AMD Helios AI rack system takes direct aim at Nvidia with Microsoft, OpenAI backing

AMD takes on Nvidia with its Helios AI rack scale system

Chipmaker AMD is taking aim at competitor Nvidia with its latest hardware release: a rack-scale system designed to power computing needs of the world's largest AI labs. At the company's sold-out Advancing AI conference in San Francisco on Thursday, AMD Chair and CEO Dr. Lisa Su promoted the new AI rack system known as Helios -- along with its growing list of customers, including Microsoft -- as the company prepares to ship it later this year. Su also pitched the company's newest chips that are designed to feed the compute-hungry dragon that is the AI industry. Rack systems combine many processors into a single high-powered unit. They are built for data centers, where they train and run AI models and other compute-intensive workloads. Su called Helios the tech industry's "highest performance AI rack," adding that it was "built to train and run the most demanding frontier models in the world at massive scale." The system will be deployed by leading AI companies at gigawatt-scale, the company said. Nvidia has historically dominated this market with its Vera Rubin and Grace Blackwell rack-scale systems. AMD is clearly looking to get in on the action. And Helios' performance metrics appear to give it a real chance, beating out Vera Rubin by a number of metrics, the Register reported. Helios, which was first revealed in 2025 and shown onstage in January at CES 2026, already has several well-known customers, including OpenAI, Meta, Oracle, Anthropic, and Microsoft, all of which have plans to deploy the system. Microsoft CEO Satya Nadella said Monday that the company would expand its Azure infrastructure with Helios. Meanwhile, Anthropic and AMD announced a strategic partnership Wednesday for deploy up to two gigawatts of GPUs via the new rack system. AMD also introduced Thursday its Venice-X CPU, which is designed for data centers and to handle high-computing workloads. The Venice-X is expected to launch in 2027. During her remarks, Su commented on the trajectory of the chip industry, claiming that, by the year 2030, chips that power AI will become a massive part of the overall computing market. This's because the industry is "seeing a step change in compute demand" driven largely by the rise of agentic AI, she said. "When you ask the agent to do something, it actually has dozens of steps, and it has to reason, and it has to call tools, and it has to access data, and it has to keep doing it over and over until it solves the problem, and so you need lots of GPUs to do all that," the executive said. "We're now expecting that by 2030, the AI accelerator market is going to reach about $1.4 trillion," Su said. "What that means is, by the end of the decade, the AI accelerator market is going to approach the size of the entire semiconductor market today." "We do expect that GPUs are going to make up the vast majority of that market because the algorithms are still very much in their infancy, and we're still continuing to see the workloads change, and that favors programmability in the overall silicon ecosystem," she added.

AMD takes the wraps off its Instinct MI455X AI accelerator -- CDNA 5 and Helios rack-scale architecture combine to take the fight to Nvidia in the data center

Massive performance gains, higher memory capacity, and a large scale-up domain make for AMD's strongest AI chip yet At AMD's Advancing AI event this week, the company revealed more details of its upcoming MI455X GPU and the Helios rack-scale architecture that will join 72 of those GPUs into a coherent accelerator -- the largest such system that AMD has built so far and its first to truly compete with Nvidia's NVL72 rack-scale design, as used in the Blackwell and Rubin generations. AMD calls the MI455X "by leaps and bounds the most advanced AI accelerator we've ever built," and from what we've seen, it's the most competitive product at both the chip level and at rack scale that AMD has ever put up against Nvidia's thorough dominance of the AI compute race. The full MI455X GPU is a massive chip encompassing 320 billion transistors, and it's built up using advanced packaging technologies. Four Accelerator Complex Dies (XCDs) are stacked on top of each Fabric and Cache Die (FCD) using hybrid bonding. In turn, the two FCDs are each joined to six stacks of HBM4, the two I/O dies, and to one another using TSMC's CoWoS-L technology. This chiplet design lets AMD use the most advanced TSMC 2N gate-all-around (GAA) process technology on the XCDs, where it's most beneficial for power and performance, while the FCDs and I/O dies, which contain elements that don't benefit from the densest process technologies, are fabricated on TSMC N3P. CDNA 5 represents a large shift in the shape of the CDNA architecture. AMD now calls the fundamental building block of the CDNA 5 Accelerator Complex Die a "Work Group Processor" instead of a "Compute Unit," but in practice, the basic layout of the rest of the Accelerated Complex Die (XCD) is largely similar. The number of WGPs on the MI455X remains the same as on the MI355X at 256. Because there hasn't been a change to the number of fundamental compute resources on the chip, the per-WGP throughput on the CDNA 5 MI455X has to be much higher than on the MI355X to deliver its large performance boost. Among the many other changes for this generation, CDNA 5 marks a major shift for the programming model of an Instinct GPU. The width of a wavefront, or group of work items or threads that each workgroup processor addresses, is now 32 instead of 64, a choice AMD says improves instruction latency, branch divergence penalties, and register pressure. It further explains that a 32-wide approach increases the flexibility of the architecture for interacting with different tensor tile sizes and mapping compute kernels to the hardware. RDNA GPUs have used a native wavefront size of 32 since their introduction. CDNA 5 greatly enhances compute performance over CDNA 4, theoretically doubling and in some cases quadrupling the peak FLOPS possible from the chip. The MI455X especially benefits lower-precision floating-point formats now common for use in inference. OCP MXFP8 and MXFP4 formats are theoretically up to 4X faster than on the CDNA 4 MI355X. Here's an overview of the MI455X's theoretical peak FLOPS compared to both the previous-generation MI355X and Nvidia's Rubin GPU. At least on paper, the MI455X offers peak performance that bests what Nvidia has shown for Rubin so far, and in combination with the Helios rack-scale architecture that finally gives AMD the same 72-GPU coherent domain as Nvidia's NVL72 rack-scale design, AMD is more competitive in the AI data center capacity race than ever before. And that's translating into real customer wins, as AMD has announced pivotal deals with Microsoft and Anthropic this week. But we'd be careful about drawing too many conclusions from these head-to-head peak FLOPS numbers, as the realized performance of these GPUs in the real world is likely to be much lower in practice, as AMD itself admitted in one of our sessions. The ongoing challenge will be to optimize software and applications to achieve as much of that theoretical performance as is possible. Beyond its general compute improvements, AMD is also touting improved transcendental math performance from the CDNA 5 Transcendental Unit, which has wide-reaching implications for performance on essential AI functions like softmax, neural network activations, and attention. AMD says that the CDNA 5 Transcendental Unit doubles throughput versus CDNA 4. The Transcendental Unit also adds support for an explicit tanh instruction, which is useful for certain operations on hidden layers within neural networks. AMD also deeply revised the cache and memory hierarchy on the MI455X. The large Infinity Cache on CDNA 4 has been ditched in favor of a smaller but higher-bandwidth shared L2 cache on each Fabric Compute Die. Each FCD has a 96MB L2 slice for 192MB in total, compared to just 32MB backed by the 256MB Infinity Cache on the MI355X. AMD says this cache offers 1.5X higher bandwidth per FCD compared to the CDNA 4 Infinity Cache, so in aggregate, the MI455X has 3X the L2 bandwidth compared to the MI355X. Within the WGP, larger, faster, and more flexible caches are now available. The local data store (LDS) or scratchpad memory has doubled in size versus CDNA 4, to 320 KB, and in total, there is 96 MB of LDS across the chip, twice that of the largest CDNA 4 implementation on the MI355X. AMD says the LDS SRAM offers higher read and write bandwidth per clock than in the previous generation, but it didn't provide specifics. Critically, the main memory architecture of MI455X has moved to HBM4 for this generation. AMD uses six stacks of HBM4 per FCD for a total of 12 on the chip, offering a memory capacity of 432GB and memory bandwidth of 23.3 TB/s per GPU. High memory capacity and bandwidth are both crucial for keeping ever-expanding AI model weights and KV caches close to the GPU for the best performance. This HBM implementation is one of the MI455X's strongest advantages over Nvidia's Rubin in isolation. It offers both much higher capacity and slightly higher bandwidth than the first Rubin GPU that Nvidia detailed this week. In its initial configuration, Rubin only offers 288GB of HBM4 with up to 22 TB/s of bandwidth. At rack scale, the MI455X's higher HBM capacity adds up to 31.1 TB of HBM across 72 GPUs, or a whopping 50% more than the 20.7 TB aggregate capacity of Vera Rubin. And the Helios rack-scale architecture is AMD's first to join all of that GPU memory into a single coherent domain. Since AI inference performance is dominated by both data locality and memory bandwidth, AMD's advantages in this regard are sure to attract plenty of attention. Efficient data movement across the chip is a key consideration for reducing power consumption and improving performance on the MI455X. The CDNA 5 Tensor Data Mover is an improved version of the data movement engine in past CDNA generations. Like Nvidia's Tensor Memory Accelerator, the TDM accelerates the production of certain tensor-specific memory addresses and facilitates moving the related data from memory without involving the shader engines. In the CDNA 5 generation, the TDM gains the ability to move data directly from DRAM to the WGP LDS cache without staging data at intermediate cache levels. AMD has also added multicast support for memory reads to the WGPs to take advantage of the fact that AI workloads often need to work on the same weights and activations across multiple WGPs at once. By using multicast, a single memory read can be amplified across many WGPs, increasing effective memory bandwidth, reducing duplicated traffic on the bus, and lowering power consumption. Nvidia has had a similar capability in its GPUs since Hopper, but given AMD's intense focus on efficient data movement this generation, it's perhaps unsurprising to see a version of it added here. All told, the MI455X is AMD's most compelling Instinct product yet. It delivers a massive generational performance improvement, it puts up peak FLOPS on paper that are competitive with or even superior to Nvidia's upcoming Rubin GPU, and it offers a massive 432GB pool of HBM4 memory that's 1.5x larger and even slightly faster than Rubin's 288GB complement. Combined with the Helios rack-scale architecture and its 72-accelerator scale-up domain -- AMD's first true answer to Nvidia's NVL72 rack-scale design -- we should expect the race for AI compute superiority to really heat up in the second half of this year as both Nvidia and AMD begin delivering their next-generation products to customers. Follow Tom's Hardware on Google News, or add us as a preferred source, to get our latest news, analysis, & reviews in your feeds.

AMD attacks the rack with Helios systems that rival Nvidia's

Could Nvidia's days of datacenter dominance be threatened? With the launch of Helios, AMD's first true rack-scale AI platform, the House of Zen is done playing catch up. The company claims the rack system is powered by the fastest AI accelerators on the market. And, at least on paper, the 72-GPU system is not only bigger and faster by nearly every metric than Nvidia's existing Blackwell-based rack systems but Vera Rubin as well, and that includes the rack. Measuring in at 1.2 meters wide and 44OUs high, the OCP Open Rack Wide form factor is nearly twice the size of Nvidia's NVL72, and AMD has clearly put the extra space to good use. Compared to Vera Rubin, Helios boasts 50 percent more HBM4 and scale out bandwidth and between 15 and 25 percent higher performance for AI training. Vera Rubin's adaptive compression tech will supposedly give it a 25 percent lead over Helios at FP4, but, as we understand it, that's only for inference workloads. For applications that can't, Helios offers 15 percent higher peak FP4 FLOPS. To be clear, it's not the first time we've seen AMD pull ahead on memory or FLOPS. The difference is those products usually launched a year after Nvidia's equivalent part. Helios launches right in time with Nvidia's Vera Rubin platform. AMD estimates Helios' higher peak performance will give it a 30 percent performance per dollar lead over the competition. The belly of the beast Helios' performance gains are rooted in an all-new GPU built on AMD's 5th-gen CDNA compute architecture. Much like the rack it powers, the Instinct MI455X is massive, though the chip is underselling it a bit. Just like the MI300 series, AMD's latest datacenter GPU is a silicon sandwich that stitches together I/O, compute, and memory in a single package. Including memory, MI455X features 24 chiplets using a combination of 2.5D and 3D packaging technologies. The chip's eight compute dies are fabbed on TSMC's bleeding edge 2nm process tech, which are stacked atop a pair of 3nm fabric and cache dies (FCDs). The FCDs are an interesting twist on the formula. They function as a cache heavy interposer with 96 MB of L2 cache each and the memory controllers for the chip's 12, 36 GB HBM4 stacks. Unlike past Instinct accelerators in which the I/O die was located under the compute, the MI455X breaks these out into two new dies -- also fabbed on TSMC's 3 nm -- which are responsible for chip-to-chip communication. One benefit to this architecture is that the chip can be made to function as one big GPU or two smaller ones depending on which NUMA configuration you opt for. The chip also supports spatial partitioning into up to eight virtual GPUs. Under the hood, AMD's CDNA architecture brings some notable improvements over the last generation. Compared to last year's MI355X, the MI455X promises as much as 4x higher floating point performance for AI workloads. Most notably, the MI455X forgoes FP64 entirely in order to dedicate as much die area to AI-centric datatypes like MXFP4 and MXFP8 as possible. This generation also adds support for 16 and 32 block scale data types. For those needing FP64 compute, that functionality will be served by a different HPC centric SKU. This isn't the only physical change. As we mentioned earlier, for this generation, AMD has opted for a larger shared L2 cache and ditched its last level "Infinity" cache altogether. The benefit, AMD fellow Alan Smith says, is higher bandwidth and a simplified data path compared to last gen. "The bandwidth delivered for one of these L2 caches in MI455X is 1.5x the aggregate bandwidth of the Infinity cache on MI355X." Along with prioritizing low precision compute, AMD configured the chip's execution engines to boost IPC and implemented a new direct memory access (DMA) engine to minimize data movement. Racking it up While bigger than Nvidia's NVL72, AMD's Helios rack architecture is remarkably similar. Both systems pack 72 GPUs spread across 18 liquid-cooled compute blades. Each blade is equipped with four MI455Xs, but rather than two CPUs per blade like you'd find on NVL72, AMD has opted for a single 96-core Venice Epyc clocking up to 5 GHz, though OEMs and hyperscalers could theoretically spec each blade with up to 256 cores, if they wanted to. The scale-up fabric used to make the six dozen GPUs behave as one great big one is, it shouldn't surprise anyone, also quite similar to Nvidia's NVL72. But rather than NVLink, AMD is tunneling Ultra Accelerator Link over Ethernet (UALoE). This means AMD doesn't need bespoke switches and system builders can get away with merchant silicon from the likes of Broadcom -- and that's exactly what they've done. The Helios reference design uses 12 of Broadcom's 102.4 Tbps Tomahawk 6 switch ASICs spread across six switch trays. Each provides 512 lanes of 200 Gbps connectivity, just enough to feed each MI455X with 3.6 TB/s of bidirectional bandwidth. Helios' scale out networking is also quite a bit beefier than Nvidia's Vera Rubin. The Nvidia platform pairs each GPU with a single 1.6 Tbps ConnectX-9 superNIC. By comparison, each MI455X gets three of AMD's 800 Gbps Pensando Vulcano network cards for a total of 2.4 Tbps of scale out bandwidth per accelerator. These scale out NICs are designed for large scale inference and training workloads that require multiple racks' worth of GPUs. For front-end communications, which include management, API traffic, and storage, each of Helios' 18 compute blades is equipped with a 400 Gbps Pensando Salina data processing unit (DPU). All of this is powered by a 50 volt liquid-cooled DC bus bar located at the back of the system that under load can deliver between 225 and 245 kW of power. Nvidia hasn't shared system-level power consumption for its Vera Rubin racks just yet, but our best estimates put peak power consumption at somewhere between 240 and 250 kW. If Helios can deliver the performance claimed and our power estimates are accurate, that would not only make the platform faster than Vera Rubin, but more power efficient as well. There are FLOPS and then there are achievable FLOPS Unfortunately for AMD, Helios realistically won't touch 40 petaFLOPS -- not in the real world, but then again Nvidia probably won't hit its claimed FLOPS figures for the same reasons. Peak theoretical FLOPS are a mathematical approximation derived from OPS per clock cycle per unit of compute at a given frequency. The problem is modern accelerators don't run at a fixed clock. They follow boost curves tied to available power and thermals, and even with liquid cooling, these things get hot. And that's before we take into consideration the workloads and the software used to run them. AI is dominated by matrix multiplication, and if the matrix isn't the right shape, the max achievable matmul FLOPS (MAMF) won't get anywhere close to peak. Poorly optimized GPU kernels or compute libraries only compound this issue, and no software is perfect. What this means in practice is that the FLOPS you see advertised on these systems' spec sheets and marketing are really a ceiling, one that in the real world is functionally impossible to reach. AMD knows this and has quietly begun tweaking its marketing to focus more on measured performance than the theoretical. We saw this with the recently unveiled MI355X, where the company listed both peak and achieved FLOPS and with Helios. Speaking with the press ahead of Thursday's keynote, Anush Elangovan, VP of AI software at AMD, boasted that in real world testing, MI455X achieved 20 petaFLOPS of FP4 performance -- half of peak. Whose roofline is higher is still a useful metric, but the more meaningful, albeit difficult to pin down, is who gets closer to it. So while 50 percent of peak performance may not sound very good, Elangovan remains confident that the MI455X still delivers higher performance than any accelerator on the market. "It's delivered bandwidth and flops, and it is still the best in the industry today. There is no other accelerator that I have seen that can hit that yet," he said. AMD's candidness around real world performance hasn't dissuaded the major model builders and hyperscalers. The House of Zen now counts Meta, Microsoft, and Oracle as customers. Meanwhile, OpenAI has committed to deploying gigawatts of MI455X GPUs, in exchange for a roughly 10 percent stake in the chipmaker. Similarly Anthropic plans to deploy two gigawatts of the chips in exchange for an up to $5 billion investment in the model dev. Much more to come Helios may be AMD's flagship GPU platform, but the company recognizes that not everyone is ready for a 225-plus kW rack. Alongside the MI455X, AMD is working on several other CDNA 5-based GPUs including one optimized for enterprise deployments. First teased at CES last winter, the MI440X is a cut down version of the MI455X that will be offered in a more traditional eight GPU form factor. Meanwhile for those less concerned with AI and more interested in high-performance computing, AMD is also working on a specialized version of the chip, called the MI430X, that swaps the machine learning-optimized compute dies for ones designed to maximize single (FP32) and double (FP64) precision performance. These data types are essential for a wide range of high-performance computing (HPC) and scientific computing taking place in research centers and national labs around the globe. Each chip is expected to deliver north of 200 teraFLOPS, making it the most potent supercomputing accelerator AMD has ever built. And, notably, AMD claims that unlike Nvidia, the chip won't need FP64 emulation to be competitive. While details are thin, the MI430X has already won AMD two high profile contracts including EuroHPC's Alice Recoque supercomputer and Oak Ridge National Laboratory's next-gen Discovery system, which are expected to come online in 2027 and 2028 respectively. A long and winding road to Venice Then, of course, there are AMD's sixth-gen Venice Epycs. At Advancing AI this week, we caught a glimpse of AMD's full datacenter CPU lineup, which will be rolling out over the next year and a half. For this generation, AMD has opted for two separate sockets. SP7, which is what AMD is using in Helios and the only Venice CPUs shipping before the end of the year, will support up to 256 cores and 512 threads in a 600 watt socket. Like AMD's earlier Genoa generation, AMD will also support a version of the chip aimed at HPC applications that uses its 3D V-Cache tech to boost L3 cache to more than a gigabyte, but these won't arrive until late next year. For lower-power enterprise applications, AMD's SP8-based Venice CPUs, due out in the first half of next year, will offer more modest core counts ranging from eight to 128 in a lower power package. Meanwhile for AI head nodes, AMD is working on a custom version of Venice called Verano that promises 5 GHz boost clocks along with LPDDR5x memory when it arrives late next year. ®

Microsoft will deploy AMD's Helios rack-scale AI accelerator 'at scale' on Azure - Radeon Instinct MI455X and Epyc Venice power will be available through Redmond's cloud infrastructure

But it's not clear just how much AMD AI compute Microsoft is buying The demand for AI compute is already insatiable, and it seems only poised to grow in the wake of the introduction of frontier-class open models like Kimi K3 that anybody can potentially fine-tune and serve. Against this backdrop, Microsoft and AMD are teaming up to get Redmond more AI FLOPS for both internal and external use. The two companies announced this morning that Microsoft will commit to adding AMD's Helios rack-scale AI accelerator in volume to run frontier-model workloads in its own data centers, as well as for Azure AI infrastructure customers and services. The partnership makes next-gen AMD AI compute available to Azure customers like AI labs for AI training and inference serving workloads, and it'll also underpin managed compute for enterprise customers looking to deploy AI workloads through Microsoft Foundry. The two companies didn't indicate the exact size of Microsoft's Helios deployment in either watts or dollars, but the commitment would seem to be another major win for AMD as it seeks to grab data center GPU share from Nvidia. AMD has struck massive partnerships with OpenAI and Meta in the past year with gigawatts of compute installations and hundreds of billions of dollars potentially hanging in the balance. For a quick refresher, the Helios rack-scale accelerator will take the fight to Nvidia's Vera Rubin NVL72 system when it arrives later this year. Helios joins together 72 next-generation Instinct MI455X GPUs with an aggregate of 31.1TB of HBM4 memory capacity across the system. Those GPUs offer as much as 1.4 exaFLOPS of FP8 compute and 2.9 exaFLOPS of FP4 for AI models using those OCP AI data types. AMD is targeting 260 TB/s of scale-up bandwidth within the rack, on par with Nvidia's Vera Rubin NVL72 rack-scale system, and 43 TB/s of scale-out bandwidth using UALink over Ethernet, or about twice that of Vera Rubin, although the performance of UALink over Ethernet in practice remains to be seen. Microsoft and AMD also announced that Azure will add two new VM series built on AMD's upcoming sixth-gen Epyc Venice CPUs: the HDv2 series for "agentic AI and data pipelines," and the HXv2 for semiconductor design workflows. Microsoft will also leverage its existing deployment of AMD Pensando DPUs to integrate that hardware into its Azure Boost offerings to accelerate networking and storage processing operations. Tom's Hardware will be on the ground at AMD's Advancing AI event this week, where we expect to learn more about AMD's AI ambitions for the second half of this year and beyond. Stay tuned for our coverage from that event. Follow Tom's Hardware on Google News, or add us as a preferred source, to get our latest news, analysis, & reviews in your feeds.

AMD launches Helios, its first rack AI system to rival Nvidia, adding Microsoft as newest buyer

AMD rivals Nvidia with first AI system, signing Microsoft and Meta After a decade-long comeback, chip giant Advanced Micro Devices is preparing to ship its first rack-scale system for artificial intelligence, called Helios, to a growing list of customers that now includes Microsoft. It's the first rival to Nvidia's wildly popular Grace Blackwell and Vera Rubin systems, and is aiming to give the world's most valuable chipmaker its first real competition in years. Microsoft announced Monday it will use the Helios system in its data centers, joining Meta, OpenAI, Oracle and others in a race to grab as much compute as possible. AMD will begin shipping to customers, including Microsoft, later this year. Details about financial terms or the amount of compute capacity weren't disclosed. "We are expanding the Azure infrastructure portfolio with AMD Helios to give customers the performance, scale and choice they need to build and run the next generation of AI applications," Microsoft CEO Satya Nadella wrote in a press release. The new Helios system will power frontier model inference for Microsoft, its AI customers and support Azure AI services. Microsoft will also add two new computing instances run on AMD's latest "Venice" central processing units, or CPUs, one for agentic AI and data pipelines, and another for semiconductor design. It's the continuation of a longtime partnership, with AMD chips powering Microsoft's Surface PCs and Xbox gaming consoles for many years. In 2023, Microsoft was also the first to adopt AMD's MI300X graphics processing unit, or GPU, that rivaled Nvidia's AI chips. Microsoft also deploys its own Maia chips in its data centers. Like its peers, Microsoft needs as much compute as possible, especially as it ramps up its own model development and allocates more computing capacity to research and development. In June, it announced seven models built in-house. Microsoft's AI efforts thus far have seen mixed results, from its 365 Copilot AI assistant to its GitHub Copilot coding agent. It's the worst-performing "Magnificent Seven" stock so far this year. Microsoft is part of a growing number of big companies turning to AMD for AI acceleration. AMD says eight of the top 10 AI companies run workloads on its Instinct GPUs, including OpenAI, Cohere and Elon Musk's SpaceXAI, which is part of SpaceX.

AMD's Helios rack is finally shipping, and it's coming for Nvidia's AI dominance

Serving tech enthusiasts for over 25 years. TechSpot means tech analysis and advice you can trust. In context: When a market is utterly dominated by a single player as AI infrastructure has been by Nvidia, it's tough for any rival to make a dent. No company knows this better than AMD. They've been battling in many segments against Nvidia, as well as other semiconductor giants like Intel, for decades now. That challenge hasn't deterred AMD. If anything, it seems to have sharpened the company's focus. At its recent Advancing AI event in San Francisco, AMD rolled out an unusually dense wave of announcements spanning AI infrastructure for the cloud and enterprise data centers, AI-powered client devices, and physical AI for robotics. The big news was the official launch of the Helios rack, AMD's first full AI infrastructure rack solution powered by 6th-gen Epyc CPUs, Instinct MI455X datacenter GPUs, and Pensando networking chips. AMD first previewed Helios at last year's event, but actually shipping it is an important step, both philosophically and practically. Nvidia has trained the market to expect rack- and datacenter-scale solutions as table stakes for AI infrastructure, and without a shipping rack product of its own, AMD was at a real disadvantage. With Helios shipments now set to begin later this quarter, AMD can credibly call itself a serious contender in AI infrastructure, and organizations shopping for the technology now have a legitimate alternative to weigh. That new confidence was reflected in the customers and partners AMD paraded on stage. From Anthropic and OpenAI on the frontier-model side to Meta and enterprise customers like AT&T, there was a genuine sense of energy and partnership running through the event. CEO Lisa Su went out of her way to highlight how closely AMD had co-designed Helios with its earliest customers, a reminder of how much it pays to actually listen to what buyers want before finalizing a product. AMD didn't stop at building a credible alternative; it also made some aggressive claims about outperforming the competition outright. How well benchmark claims translate to real-world performance is always hard to judge in advance, but AMD touted 20% greater per-core performance for its latest Epyc CPUs over Nvidia's new Vera CPU, among other figures. The company also highlighted Helios's larger HBM4 memory capacity and faster memory bandwidth compared with similar Nvidia rack systems. However those numbers hold up once independent, third-party benchmarks arrive, it's fair to say AMD is now offering a hardware alternative to Nvidia. For customers chasing low-latency inference, AMD also announced a partnership with Cerebras that will let a Cerebras rack sit side by side with a Helios rack for demanding workloads. Of course, AI infrastructure is about more than hardware, and one of the more intriguing announcements from Advancing AI was a new version of AMD's software stack, called ROCm.ai. It's designed to use AI-assisted programming to make building software optimized for AMD hardware significantly easier - including, for example, porting code originally written in CUDA over to ROCm. Given Nvidia's roughly two-decade head start with CUDA, a wholesale industry shift won't happen overnight, but it does make the so-called "CUDA moat" look less daunting than it once did. A ROCm.ai feature called Hyperloom, purpose-built to optimize AI applications for ROCm, looks particularly compelling: at the event, Anthropic described how an engineer let Claude run autonomously over a single weekend to adapt and tune performance on an AMD Instinct MI355X chip - and came back Monday to a working, steadily improving performance curve. Beyond the data center, AMD also had something to say about PCs and robotics. On the PC side, the company detailed a new AI-focused deskside computing device, codenamed Gorgon Halo, built on an enhanced version of its Ryzen AI Max APU that pairs a Zen 5 CPU with an RDNA 3.5 GPU. The device will offer up to 192GB of unified memory, at a price still to be determined but likely steep given current DRAM costs. Even so, it's an interesting entry in the growing class of developer-focused AI systems hitting the market. More surprising was a new partnership with Cisco to bring AI agent management, observability, and tokenomics tools to this class of device. Leveraging Cisco's Cloud Control, AI Defense, and Splunk technologies, the joint solution is aimed at organizations planning to deploy AI-focused PCs at scale. Details are still thin, but it's a notable partnership that shows how AMD and Cisco are working to bring more comprehensive AI tooling to the enterprise. Finally, AMD made several announcements in physical AI and robotics. Few people realize AMD has had a substantial presence in industrial robotics for years, largely thanks to the technology it picked up when it acquired Xilinx in 2022. The kind of reprogrammable FPGAs Xilinx has built for years are essential to many robotics systems, both for adapting to new AI algorithms and for sensor fusion - tying together the camera and sensor feeds most robots rely on. Combined with AMD's traditional CPUs and GPUs, that gives the company a broad portfolio to draw from. The result is the new Kria AI SOM, a system-on-module, and the Kria AI Robotics Developer Platform that pairs it with a carrier board for robotics designs. Both are built around AMD's Ryzen AI Embedded X100 series SoCs, alongside new open-source Kria AI Robotics Platform software. Together, it's a comprehensive toolkit for robotics developers and further cements AMD's role as a significant player in the field. Nvidia will almost certainly remain the AI infrastructure leader for the foreseeable future, but it's now clear AMD has come to play, and that's good news for everyone. As in any market, competition drives faster, better advancements, and it will be worth watching how the AI semiconductor landscape continues to evolve from here. Bob O'Donnell is the founder and chief analyst of TECHnalysis Research, LLC a technology consulting firm that provides strategic consulting and market research services to the technology industry and professional financial community. You can follow him on Twitter @bobodtech

AMD unveils Helios rack and MI455X accelerator in bid to close the gap with Nvidia

AMD launched MI455X chips, Helios racks, and Venice Epyc CPUs at Advancing AI, with OpenAI and Cerebras pledging to deploy the hardware at scale AMD unveiled a suite of new data centre products at its Advancing AI event in San Francisco on Thursday, claiming they will outperform Nvidia's competing hardware across AI training and inference. The announcements included the MI455X AI accelerator, the Helios server rack that packs 72 of those chips into a single system, and the Venice generation of Epyc server processors built on TSMC's two-nanometre process. AMD predicted the total market it is pursuing will reach $2 trillion by 2030, with AI accelerators alone accounting for more than a trillion dollars of that figure. Investors were unimpressed, sending AMD shares down about four percent during the presentation despite the stock having more than doubled in value this year. The scepticism reflects the gap between AMD's ambitions and Nvidia's dominance: Nvidia's Vera Rubin platform is already in full production and shipping to customers including OpenAI, CoreWeave, and Microsoft. AMD CEO Lisa Su has delivered a remarkable growth run, but Wall Street's expectations were already baked into a stock that had risen roughly 145 percent year to date heading into the event. Su brought executives from major AI companies on stage to endorse AMD's hardware. OpenAI's Sachin Katti said his company expects to deploy Helios "at massive scale" as it scrambles to add infrastructure capacity. The event came a day after AMD announced it would invest up to $5 billion in Anthropic and deploy two gigawatts of MI450 series GPUs in Helios racks to run Claude, with the first gigawatt shipping in the first half of 2027. AMD also announced a partnership with Cerebras, whose wafer-scale chips specialise in ultra-fast AI inference, to combine Helios racks with Cerebras servers for customers who need the lowest possible latency. AMD hardware will handle the work of deciphering queries while Cerebras chips provide rapid answers. Cerebras CEO Andrew Feldman said the combined product will ship from Cerebras-owned data centres starting in the fourth quarter and will beat a similar offering Nvidia is assembling from its recent Groq acquisition. The Venice Epyc processor, meanwhile, is AMD's answer to Nvidia's push into server CPUs with its Vera chip, which Nvidia has claimed gives it a performance edge. AMD said Venice will keep it well ahead of Vera, setting up a direct contest that independent benchmarks have yet to settle. The rivalry now spans accelerators, server processors, and complete rack-scale systems, with both companies pitching end-to-end data centre solutions rather than individual components.

AMD's Helios puts 72 GPUs and 31 terabytes of HBM4 in one rack. It is AMD's answer to Nvidia's NVL72.

AMD Helios packs 72 MI455X GPUs, 31TB HBM4, and 2.9 exaflops of inference into one rack. Built on open standards. Engineering samples H2 2026, mass production Q2 2027. AMD's Helios is a single rack containing 72 Instinct MI455X GPUs, 31 terabytes of HBM4 memory, and 2.9 exaflops of FP4 inference compute. It is AMD's first rack-scale AI system and its direct answer to Nvidia's Vera Rubin NVL72. The system uses 18 compute trays, each holding four MI455X accelerators on the new CDNA 5 architecture and one sixth-generation EPYC "Venice" CPU. Engineering samples ship in the second half of 2026. Mass production begins Q2 2027. The architecture bet is open standards. Helios uses UALink for scale-up interconnect between GPUs within the rack, Ultra Ethernet Consortium specifications for scale-out networking between racks, and the OCP Open Rack Wide form factor. Nvidia's competing NVL72 uses proprietary NVLink. AMD is betting that data centre operators who do not want to be locked into a single vendor's interconnect will pay for the flexibility. AMD Pensando AI NICs handle the networking with programmable hardware and UEC-ready RDMA. The numbers are designed to compete on memory, not just compute. Each MI455X GPU carries HBM4 with 19.6 TB/s of bandwidth. The full rack delivers 260 TB/s of scale-up bandwidth and 43 TB/s of scale-out bandwidth. That memory capacity matters for frontier model training and long-context inference, where the bottleneck has shifted from raw compute to how much data the system can hold and move. The AI-driven memory crisis has pushed HBM prices up sharply, and 31TB of HBM4 in a single rack represents an enormous materials cost that only hyperscaler and sovereign compute budgets can absorb. Supermicro showed Helios hardware at Computex in June. AMD committed billions to UK AI infrastructure at London Tech Week, and Helios is the hardware those commitments will run on. The ROCm software stack supports PyTorch, TensorFlow, and JAX, which means developers do not need to rewrite code to move from Nvidia's CUDA ecosystem, at least in theory. Whether AMD can close the software gap that has kept it behind Nvidia in AI compute is the question Helios is designed to force. The hardware specs are competitive. The ecosystem is the test.

AMD launches Helios rack system and MI450 GPUs to rival Nvidia

The company is formally introducing its Helios rack-scale system, which combines Instinct MI455X GPUs, EPYC "Venice" CPUs, Pensando networking chips, and ROCm software in an integrated platform for AI training and inference, AMD said. The company is also expected to formally launch its Venice CPU for data centers, according to Reuters. The launch follows two major customer commitments announced this week. On Wednesday, AMD announced a partnership with Anthropic to deploy up to 2 gigawatts of Instinct MI450 Series GPUs in Helios rack-scale systems, with the first gigawatt of deployment beginning in the first half of 2027. AMD also committed to make a strategic equity investment of up to $5 billion in Anthropic, the company said.

Five thoughts from Lisa Su's keynote at AMD Advancing AI

Advanced Micro Devices Inc.'s Advancing AI 2026 keynote this week was Chief Executive Lisa Su's bid to redefine the company from a "graphics processing unit alternative" to a full-stack artificial intelligence infrastructure vendor and to make this the year the central processing unit officially rebounds as a first-class AI platform. The message from Su (pictured) was ambitious and often compelling, but it also sharpened the competitive contrast with Nvidia Corp. and Intel Corp., raising as many questions as it answered. Though the obvious theme of the event was that AI is moving from pilots to production, that has been the theme of every event I have attended this year. Beyond that, there were several other sub-themes. Here are the five most notable: 1. Helios and MI450: Finally a credible rack-scale alternative -- with a catch Su opened by turning Instinct MI450 and the Helios rack into a single, rack-scale product story aimed squarely at Nvidia's system-level dominance. In AMD's benchmarks, Helios delivers "an average of 10% to 15% more performance than the competition" at fixed rack power on "the highest throughput workloads" and "leading inference modes," and she translated that into "up to 30% more tokens per dollar than the competition." Though Nvidia has set the standard for the systems approach, if AMD can deliver on the savings it claims at comparable performance, it can use that to position itself as a credible alternative. The partner lineup was the strongest evidence that these claims are real. Su said demand for Helios is "extremely strong... from the largest AI labs to hyperscalers," and she highlighted OpenAI as "one of our deepest and earliest partners deploying Helios," noting that joint engineering teams are already running GPT-class workloads. OpenAI's infrastructure lead described AMD and OpenAI engineers "working side by side to optimize the software stack" and said they expect to deploy Helios "at massive scale, starting towards the end of this year, and then accelerating toward 2027." The catch is that "average of 10% to 15% more performance" and "30% more tokens per dollar" remain AMD-run numbers against unnamed "competition" and unspecified model mixes. Until cloud instance specs, public benchmarks and customer case studies show comparable gains in the wild, Helios is only a strong narrative and a promising design, not yet a proven market-share shift. Nvidia still owns software mindshare and the incumbent installed base; AMD must convert a handful of flagship design wins into a durable ecosystem. 2. CPU rebounds: Venice turns agentic AI into a three-tier compute story One of the more interesting and underappreciated parts of the keynote was AMD's aggressive effort to reset the CPU narrative in AI. For the past few years, CPUs have been cast as glorified I/O controllers in GPU boxes. Su pushed back hard on that, arguing that AI infrastructure is splitting into three CPU roles: * GPU servers where "the CPU's job is basically to drive the GPUs." * Dense "agent servers or what we call agent sandboxes" where "the priority is actually density and the highest-performing cores per watt to run thousands of agents at once." * Traditional genera-purpose servers where it's "all about efficiency" for databases, data services and enterprise apps. Venice, AMD's new Epyc family on Zen 6 and TSMC 's two-nanometer process, is the company's attempt to own all three tiers. Su called it "one of the largest generational gains in the history of Epyc," claiming "up to 1.8 times more performance than Turin" and up to 512 threads per socket. She then broke Venice into a family: Venice HF for GPU host nodes at up to 5 GHz; a 256-core Venice with "the highest compute density in the industry" for agent sandboxes; and a 128-core part tuned for enterprise performance per dollar, plus variants like Venice X and Verano for HPC and AI host interconnect. Her punchiest line of the keynote, "Epyc is the only CPU portfolio that leads across all use cases," was clearly aimed at both Intel's latest Xeons and the growing crop of Arm server CPUs. She claimed that Venice delivers more than twice the agents per watt for agent sandboxes versus x86 and up to 3.3 times more performance per watt at the rack than unnamed Arm competitors in a 100-kilowatt rack. She also pointed out that x86 software compatibility still matters when you're "adding thousands of employees to your enterprise" in the form of agents. The critical angle: AMD is right that agentic AI gives CPUs a second life, but this space is now intensely contested. Intel is not standing still on core count, memory bandwidth, or AI offload, and Arm vendors are pushing hard on efficiency and custom silicon for cloud providers. Venice's generational gains are impressive on paper, but AMD still has to win OEM designs, cloud footprints and independent software vendor certifications at scale, all while customers are also considering Arm and custom accelerators tailored to their agent workloads. 3. Jeetu Patel and Santosh Janardhan: CPUs and GPUs become 'conjoined things' If Su provided the product narrative, Jeetu Patel from Cisco Systems Inc. and Santosh Janardhan from Meta Platforms Inc. provided the architectural reality check, and they largely backed AMD's thesis that CPUs are back in the spotlight. Patel argued that "it's not just a GPU game anymore. It's CPUs and GPUs. If anything, I think CPUs are becoming at least as important, if not more." His view is grounded in enterprise knowledge, which Cisco has more of than all the other companies on stage combined. Long-running agent workflows, tools, databases and networks all must be orchestrated around frontier models. He framed compute as a heterogeneous fabric where "you must think about CPUs and GPUs as conjoined things. You hand off workloads depending on other workloads. You employ the right hardware." That is exactly the kind of messaging AMD needs enterprise CIOs to hear to become a more strategic vendor. Meta's Janardhan pushed the same idea at the data center scale. For him, the AI problem is no longer about squeezing every percentage point out of a single chip; it's about "the whole data center as one integrated system -- servers, hardware, networking, cooling, power." He noted that data centers and silicon "take years to build," and argued that the industry needs to be "sitting down in a room, co-designing today for what we need to deploy in 2027 and 2028." In other words, CPUs, GPUs, memory, networking and power are now co-equal design levers. It was great to see these companies on stage with Su, because they aren't second-tier logos; they're two of the largest AI and networking players on the planet, validating AMD as a co-design partner, not a backup supplier. For the established players, that's a warning that their traditional lock-in at the CPU and GPU levels may not survive an era in which hyperscalers want multi-vendor, co-designed systems to manage risk, cost and power. 4. ROCm.AI and Hyperloom: AMD tries to leapfrog on AI-native tooling On software, AMD went straight at its perceived weakness, the software gap versus Nvidia's stack, with a different approach: Let AI write and optimize more of the GPU code. Senior Vice President Vamsi Boppana introduced ROCm.AI as "an agentic AI platform that brings the capabilities of AI-assisted GPU programming to developers." Built on AMD's open software stack, ROCm.AI adds an AI optimization layer called Hyperloom that can "analyze the workload, tune configurations, select and tune kernels, adjust parallelism strategies and iterate towards performance goals." Internally, AMD has already pushed "a suite of 14,000 models through Hyperloom," generating optimizations that would have been "impossible even with a large team of engineers before." In a live example, an AI agent targeting MiniMax M3 on MI355s with VLLM identified an opportunity to write a more optimized Mixture-of-Experts GEMM kernel, delivering a 38 percent improvement in tokens per second. Engineers onstage admitted that some AI-generated kernels are "shockingly good, sometimes better than the most manually tuned versions." AMD's competitive bet is that an open stack plus AI-assisted optimization can close the software-ecosystem gap with Nvidia faster than traditional hand-tuning ever could. The risk is that this is still early-stage technology: automatic code generation must be safe, reproducible and debuggable at scale, and customers will want to see real-world workloads, not staged demos. Nvidia will not sit still here either; it has every incentive to build its own AI-native tooling on top of CUDA and its closed ecosystem. 5. Co-designed, open systems: AMD plays the long game against lock-in The final through line in Su's keynote was an embrace of co-design and openness as AMD's strategic wedge against incumbents. OpenAI's infrastructure chief, Sachin Katti, described a future where AI becomes "a problem at a data center scale," not just a "rack-level problem," and stressed the need to co-design "from CPUs to GPUs to memory, networking, storage, power distribution and the cooling systems that go with it." Cerebras CEO Andrew Feldman, announcing a joint solution that marries AMD's Helios racks with Cerebras' wafer-scale engine, argued that customers who used to choose between "high throughput" and "extraordinary speed" can now get both -- "five times the throughput while continuing to deliver this extraordinary speed" for ultra-low-latency inference. Su tied these threads together by leaning hard into open software. Because AMD's compiler stack and drivers are largely open, partners and even AI agents can see all the way down to the ISA. OpenAI's Philippe Tillet credited that openness with enabling "very, very significant performance gains" and faster portability of GPT-class models to AMD hardware. In a world where "recursion," that is AI helping design the next generation of AI systems, is becoming a reality, AMD is betting that openness will let it harness that flywheel more effectively than a closed stack. The competitive implication is that while Nvidia still has the deepest, most entrenched software ecosystem, AMD aims to be more open to attract a broader base of partners and developers. Intel has the x86 incumbency but has struggled to convert it into AI mindshare. AMD is positioning itself as the third pole: not just cheaper GPUs, but an open, co-designed CPU-plus-GPU platform tuned for the agentic, data-center-as-a-system future. Whether that bet pays off will depend on execution in silicon delivery, software quality, and ecosystem traction, not just on keynotes. But if Advancing AI 2026 is any indication, AMD has stopped talking like a fast follower and started acting like a company that expects to set the terms of the AI infrastructure debate. Final thoughts AMD used Advancing AI 2026 to make a statement: It is no longer content to be "the alternative" to Nvidia, and it now has credible hardware at the rack level, a resurgent CPU portfolio tuned for agentic AI, and a software story that leans into AI-assisted optimization and open ecosystems. Helios plus MI450 gives hyperscalers and frontier labs a rack-scale option that can be argued on performance per watt and tokens per dollar, while the Venice Epyc family targets the emerging three-tier AI compute stack, comprised of GPU hosts, dense agent sandboxes, and general-purpose enterprise, at a moment when CPUs are quietly becoming the control plane and workhorse for agentic workflows. On software, AMD has stopped pretending it can out-CUDA Nvidia head-on; instead, it is betting that an open stack plus AI-native tools like ROCm.AI and Hyperloom can compress the time it takes to get real workloads performant on its silicon. The uncomfortable reality for AMD is that none of this makes Nvidia "trail" in the broader AI race. Nvidia still owns the dominant software ecosystem, the bulk of the deployed AI accelerator footprint, and a deeply integrated toolchain. In most large AI shops, AMD remains the second platform to be qualified rather than the assumed default. Openness is AMD's best strategic tool against that incumbency, lowering switching costs and inviting partners and, eventually, AI agents to co-design and tune all the way down to the instruction set, which differs from Nvidia's model. The question is whether that openness, combined with competitive performance and TCO, is enough to move AMD from "necessary diversification" to "first choice" for a meaningful share of new deployments. Right now, openness looks necessary but not sufficient, and AMD will need multiple cycles of flawless execution on silicon, software and the ecosystem to turn this impressive keynote positioning into durable market power. Zeus Kerravala is a principal analyst at ZK Research, a division of Kerravala Consulting. He wrote this article for SiliconANGLE.

Microsoft to deploy AMD Helios AI system on Azure

"We are expanding the Azure infrastructure portfolio with AMD Helios to give customers the performance, scale and choice they need to build and run the next generation of AI applications," Microsoft Chairman and CEO Satya Nadella said in a statement. Beyond Helios, the expanded partnership covers additional AMD products across Azure. Two new VM series -- Azure HDv2, aimed at agentic AI and data pipelines, and Azure HXv2, designed for semiconductor work -- will run on 6th Gen AMD EPYC "Venice" processors, the company said. AMD's Pensando data processing units will also see broader deployment in Azure networking infrastructure.

AMD Instinct MI400 GPUs Target Frontier AI and FP64 HPC

AMD has announced its Instinct MI400 Series of data-center GPUs during Advancing AI 2026. The portfolio currently includes two products with distinct roles: the Instinct MI455X for frontier AI and rack-scale deployments, and the MI430X for sovereign AI, scientific research and high-performance computing. The MI455X is based on AMD's CDNA 5 architecture and integrates 432 GB of HBM4 memory. AMD specifies memory bandwidth of up to 19.6 TB/s per GPU. The combination is intended to accommodate larger AI models and working data sets without requiring data to be divided across as many accelerators. AMD is using the MI455X in its Helios rack-scale platform. A complete Helios system contains 72 MI455X GPUs, approximately 31 TB of combined HBM4 memory, sixth-generation EPYC Venice processors and AMD Pensando networking hardware. AMD rates a full Helios rack for up to 1.4 exaFLOPS of FP8 throughput and 2.9 exaFLOPS at FP4. These lower-precision formats are used to increase AI training and inference throughput. The figures represent theoretical compute performance and do not account for model structure, memory access, interconnect overhead or software efficiency. The Instinct MI430X is designed around a different set of requirements. AMD positions this accelerator for scientific computing, sovereign AI and installations that combine conventional HPC with AI processing. It delivers up to 288 TFLOPS of hardware-based FP64 performance. FP64 remains necessary for many technical workloads that require considerable numerical accuracy. Potential applications include climate research, engineering simulation, fluid dynamics, molecular modelling, financial analysis and other calculations where moving to a lower-precision format could influence the result. Sovereign AI installations are also part of the target market. These systems are generally operated within national or organizational boundaries and can have stricter requirements concerning data residency, infrastructure ownership, software control and security. AMD lists secure boot, encrypted GPU-to-GPU links and hardware-based protection among the MI400 Series security features. These mechanisms are intended to protect workloads during system startup and while information moves between accelerators. A complete technical description of the security implementation has not yet been released. Both MI400 products use the AMD ROCm software environment. ROCm provides programming interfaces, compilers, optimized libraries, runtimes and deployment tools for AI and HPC applications. AMD is continuing to focus on open standards and software portability, although the difficulty of moving existing workloads will depend on their frameworks and reliance on vendor-specific code. The MI455X is intended primarily for dense rack-scale platforms such as Helios, where many GPUs operate together during model training and large inference jobs. The MI430X can be deployed in more conventional mesh-based HPC clusters, allowing it to address research and technical computing environments with different networking and system-layout requirements. AMD has not disclosed complete MI430X specifications, including HBM4 capacity, memory bandwidth and board power. Individual accelerator pricing and detailed availability information also remain unannounced. Independent benchmarks will be needed to determine real-world scaling, power efficiency and performance against competing accelerator platforms. Specification AMD Instinct MI455X AMD Instinct MI430X Primary market Frontier AI and AI factories Sovereign AI and HPC Primary workloads Large-scale inference, model training and fine-tuning Scientific computing and converged AI-HPC workloads Architecture AMD CDNA 5 Not fully detailed Memory 432 GB HBM4 HBM4, capacity not specified Memory bandwidth Up to 19.6 TB/s Not specified FP64 performance Not specified Up to 288 TFLOPS System architecture AMD Helios and other rack-scale AI systems Traditional mesh-based HPC deployments Software platform AMD ROCm AMD ROCm Security Secure boot, encrypted GPU links and hardware-based protection Secure boot, encrypted GPU links and hardware-based protection

AMD unveils Helios AI rack to challenge Nvidia

AMD is targeting competitor Nvidia with its latest hardware release: a rack-scale system designed to meet the computing needs of the world's largest AI labs. At the Advancing AI conference in San Francisco, AMD Chair and CEO Dr. Lisa Su presented the new AI rack system, named Helios, along with its expanding customer base, featuring Microsoft, as the company gears up for shipments later this year. Helios combines multiple processors into a single high-powered unit optimized for data centers to train and operate AI models alongside other compute-intensive workloads. Su described Helios as the tech industry's "highest-performance AI rack," specifically built to handle the most demanding frontier models at scale. The system will be deployed by leading AI companies at gigawatt-scale. Nvidia has historically dominated the AI rack market with its Vera Rubin and Grace Blackwell systems. However, early performance metrics indicate that Helios outperforms the Vera Rubin across several key areas, presenting AMD with a potential competitive edge, according to The Register. Helios was initially revealed in 2025 and demonstrated at CES 2026. It has already garnered significant interest from prominent customers, including OpenAI, Meta, Oracle, Anthropic, and Microsoft. Microsoft CEO Satya Nadella announced plans to enhance the Azure infrastructure with Helios. Additionally, AMD and Anthropic formed a strategic partnership to deploy up to two gigawatts of GPUs using the Helios system. In conjunction with the Helios launch, AMD unveiled its Venice-X CPU, designed for high-computing workloads in data centers, which is expected to debut in 2027. During her address, Su discussed the future of the chip industry, predicting that chips powering AI will significantly influence the overall computing market by 2030. This shift is driven by what she described as a "step change in compute demand" resulting from the rise of agentic AI. Su elaborated on the growing complexity of AI tasks, stating, "When you ask the agent to do something, it actually has dozens of steps, and it has to reason, and it has to call tools, and it has to access data." She emphasized the need for numerous GPUs to effectively tackle these processes. Su forecasted that the AI accelerator market would reach approximately $1.4 trillion by 2030, potentially nearing the size of the current semiconductor market. "We do expect that GPUs are going to make up the vast majority of that market," Su added. She noted that as algorithms evolve and workloads change, the demand for programmable solutions within the silicon ecosystem will continue to expand.

AMD Helios rackscale solution for AI offers 15% more AI Compute than NVIDIA Vera Rubin NVL72

AMD has officially launched its AMD Helios rackscale solution for AI infrastructure and large-scale foundational-model training and inference. As a supercomputer, AMD Helios combines the new AMD Instinct MI455X GPU, 72 of them, with the latest 6th Gen AMD EPYC 'Venice' server CPUs, AMD ROCm software, and AMD Pensando networking. It's a lot of cutting-edge tech coming together, delivering 2.9 ExaFlops of FP4 Compute. According to AMD, it's enough to deliver 15% more compute (peak FP4 performance) than NVIDIA's latest Vera Rubin NVL72 rack-scale solution. And when it comes to raw specs, there is a notable difference in favor of Helios. For example, the 31TB of HBM4 memory capacity is 50% more than Vera Rubin. And with 43 TB/s scale-out bandwidth and 1.7 PB/s of memory bandwidth, AMD notes that Helios can offer AI companies 30% more tokens per dollar. "As demand grows for AI factories and frontier AI, customers need infrastructure that scales efficiently from a single rack to large gigawatt-scale AI clusters while maximizing performance, performance per watt and cost per token," AMD explains. "AMD Helios delivers an integrated rack-scale building block that simplifies deployment while providing the compute, networking and open software foundation needed for the next generation of AI infrastructure." Of course, AMD Helios isn't confined to a single rack, as it's designed for gigawatt-scale AI clusters thanks to the new AMD Pensando Vulcano 800 AI NICs providing high-bandwidth, low-latency connectivity for growth. And with that, AMD Helios is already being deployed to a wide range of leading AI companies like Anthropic, Meta, Microsoft, Oracle, OpenAI, as well as supporting the infrastructure of companies like Dell Technologies, Lenovo, IBM, and others.

AMD debuts next-generation AI infrastructure for frontier models, agentic workloads and autonomous robots

Advanced Micro Devices Inc. is pushing harder than ever to grab even more market share from Nvidia Corp. in the artificial intelligence chip industry. At its Advancing AI 2026 event today in San Francisco, the chipmaker announced a slate of updated hardware, including its next-generation AMD Instinct MI400 Series graphics processing units and the sixth-generation AMD EPYC central processing units for the most demanding agentic AI applications. The chipmaker didn't stop at just silicon. It also revealed a major expansion into the realm of physical AI and autonomous robots. Its strategy here centers on the new Kria AI Solutions, the Ryzen AI Embedded X100 series processors and the new AMD Robotics Partner Network, taking on Nvidia in one of the industry's most competitive and crucial industries. At the event, AMD Senior Vice President of AI Vamsi Boppana said it's becoming necessary to use more specialized infrastructure to maximize the performance of the most advanced frontier models today. "The next generation of AI will span frontier AI, sovereign AI and scientific computing, and each requires infrastructure optimized for its unique demands," he said. "The AMD Instinct MI400 Series extends our AI portfolio with purpose-built solutions optimized for the full spectrum of AI and HPC deployments, built on an open software foundation that gives customers the flexibility to innovate at every scale." Specialized silicon for every workload AMD built the Instinct MI400 Series GPUs (pictured above) to address this challenge specifically, Boppana said. There's a number of different chips in the series, including the Instinct MI455X GPU that's engineered for high-volume inference workloads, AI training and fine-tuning within sprawling "AI factories." Meanwhile, the Instinct MI430X GPU is geared toward sovereign AI and high-performance computing environments, delivering 288 teraflops of performance for advanced scientific applications. Both chips integrate the industry's top-tier HBM4 high-bandwidth memory chips to support the deployment of larger models, and there are also encrypted GPU-to-GPU network connections to enhance the security of these workloads. Alongside these AI factories and scientific applications, AMD also wants to position itself as a key enabler of agentic AI in enterprise computing environments and power automated business operations. Agentic AI systems don't just generate text - they can reason over data, make plans, execute code and use external tools to perform various kinds of work. But these sophisticated workloads require massive computational resources that cannot be provided by GPUs alone. Rather, what they really need are good, old-fashioned CPUs to perform tasks such as running agent sandboxes, coordinating background operations, managing memory states and ensuring data is constantly flowing into the GPUs. According to Dan McNamara, senior vice president and general manager of Compute and Enterprise AI at AMD, agentic workloads are made much more efficient when they're powered by a combination of GPUs and CPUs. That's why AMD is also unveiling its newest EPYC 9006 series server CPUs (below), made up of four specialized chips. The EPYC 9006 SP7 is designed for high-density agent sandbox execution, packing 256 cores and 512 threads to support greater numbers of agents per watt of energy. There's also the EPYC 9006 SP8 CPUs, which are meant to provide right-sized efficiency for power-constrained devices with choices ranging from eight to 128 cores. The EPYC 9006X SP7, meanwhile, is aimed at data-intensive HPC and modeling workloads, and can hit speeds of up to 5.15 gigahertz with three times as much L3 cache available per core, compared to the standard SP7. Finally, AMD introduced the EPYC 9006 LP "Verano" CPU, which is a specialized low-power chip that's purpose-built to act as an AI host node. For public cloud infrastructure providers and enterprises that need to deploy AMD's chips at large scale, the company is offering the new Helios rack-scale data center system (below), which makes it simpler to get massive clusters of them up and running in the fastest time possible. Traditionally, data center operators faced a heavy burden in building up clusters of chips that can work in concert with one another, having to fudge together various fragmented networking components. This would often lead to integration headaches that would delay deployment timelines. To get around this, the Helios rack-scale system provides data center customers with preconfigured building blocks that can integrate AMD's latest Instinct GPUs with the 6th Gen EPYC chips, plus its ROCm software and Pensando networking. Essentially, everything that's required to get clusters up and running comes in a convenient package. According to AMD, a single Helios rack has 18 open-rack trays that can house a total of 72 GPUs, delivering up to 2.9 exaflops of peak FP4 performance, 1.4 exaflops of peak FP8 compute, and 31 terabytes of HBM4 memory. Focus on physical AI AMD doesn't want to be confined to the cloud. It's also a big believer in AI's potential to impact the physical world by powering the next generation of autonomous robots. It wants to be at the forefront of that push, and it's making a bid to do that with the launch of its new AMD Kria AI solutions (below), which are centered on the Kria AI Robotics Developer Platform. This is an integrated development platform that combines GPUs, CPUs, neural processing units and field-programmable gate arrays to power everything from a robot's physical body to its intelligent brain. These resources can help autonomous robots to make more than 8,000 decisions-per-second while simultaneously enabling sub-millisecond vision-language-action reasoning, the company said. The AMD Kria platform is powered by AMD's new Ryzen AI Embedded X100 Series processors, which combine up to 16 "Zen 5" CPU cores with an RDNA 3.5-integrated GPU and an energy-efficient NPU on a single system-on-chip. They're designed to provide industrial-grade reliability in any environment, with AMD stressing confidence they'll be able to operate in temperatures ranging from 105°C to -40°C without any problems for up to 10 years. To further its robotics push, AMD also launched the AMD Robotics Partner Network, which aims to bring together original design manufacturers, sensor providers and software vendors into an open ecosystem. The network will standardize on AMD's ROCm software stack, which preserves around 75% of CUDA code during application migrations, allowing robotics makers to transition from Nvidia's silicon with minimal trouble. With today's announcements, AMD has shown that it can match the frenetic pace of silicon development set by its arch-rival Nvidia and provide enterprises with a real alternative to that company's chips, which still dominate most AI data centers. In fact, AMD claims that its Helios rack-scale solution is a superior option for AI factories specifically, offering 15% greater peak FP4 performance and 50% greater HBM capacity while reducing token costs by up to 30% compared with Nvidia's NVL72 rack systems. AMD didn't say anything about when the new Instinct MI400 GPUs and EPYC 9006 series CPUs will be made available, but customers can expect the first samples to ship later in the year.

Microsoft to Deploy Next-Gen AMD Instinct and AMD EPYC Processors

AMD today announced an expanded strategic partnership spanning AMD GPUs, CPUs, networking and software on Microsoft Azure. At the center of this expansion, Microsoft will deploy the AMD Helios Rackscale Solution, to power frontier model AI inference for Microsoft, its AI customers and support Azure AI services. Azure will also add two new AMD EPYC CPU-powered VM series and broaden its deployment of Pensando DPUs to support Azure networking services. AMD will begin shipping Helios to customers, including Microsoft, in the second half of 2026. AMD Helios combines AMD Instinct MI455X GPUs, AMD EPYC "Venice" CPUs, Pensando networking and ROCm software in an open, integrated rackscale platform built for large-scale AI training and inference. The Azure deployment will use Helios for inference workloads spanning frontier models, Azure AI services and customer applications. "AMD and Microsoft have spent years building high-performance infrastructure together, and today we're extending that partnership across the full stack of AMD AI solutions on Azure," said Dr. Lisa Su, Chair and CEO, AMD. "Microsoft's new AMD deployments mark an important milestone as we deliver leadership compute solutions to Azure customers and scale the next generation of AI infrastructure together." "Customers are looking for AI infrastructure that is optimized for a wide range of workloads, from training and inference to data preparation, search, and reinforcement learning," said Satya Nadella, Chairman and CEO, Microsoft. "Through our collaboration with AMD, we are expanding the Azure infrastructure portfolio with AMD Helios to give customers the performance, scale and choice they need to build and run the next generation of AI applications." The collaboration expands access to AMD AI infrastructure across Azure. Frontier model builders can now leverage AMD-powered infrastructure to train and serve large-scale AI models, while enterprise customers can deploy and scale production AI workloads through Azure Foundry Managed Compute. Azure's new VM series, Azure HDv2 for agentic AI and data pipelines, and Azure HXv2 for semiconductor design, will be powered by 6th Gen AMD EPYC "Venice" processors. Together, the new VM series broaden Azure's AMD EPYC portfolio across AI, data and engineering workloads. The collaboration also extends into the networking layer that connects and scales Azure infrastructure. Building on Microsoft's broad deployment of AMD Pensando DPUs, the companies are integrating Azure Boost with AMD technologies to improve networking performance, efficiency and connection processing at cloud scale. As AI demand accelerates, AMD and Microsoft will continue to deliver open, high-performance infrastructure that gives customers flexibility, efficiency and scale to build what's next.

AMD Instinct MI455X GPU launches with 432GB of HBM4 memory and next-gen CDNA 5 architecture

Coinciding with AMD's Advancing AI 2026 event, the company has launched its latest AMD Instinct data center GPUs purpose-built for the AI era, with the Instinct MI400 Series led by the flagship AMD Instinct MI455X GPU. Built on a 2nm process, it sports 320 billion transistors, a maximum clock speed of 2.4 GHz, and 432 GB of high-speed HBM4 memory with a peak memory bandwidth of 23.3 TB/sec. According to the company, the AMD Instinct MI455X GPU is its first GPU to be designed and built for rack-scale AI deployments, which is why it serves as a key part of the new AMD Helios rack, which features 72 Instinct MI455X GPUs for 31TB of HBM4 memory with an impressive 1.7 PB/s memory bandwidth and up to 2.9 exaflops of compute performance. And with the AMD Instinct MI455X being built for Helios, this means the GPU has been designed with scale-up and scale-out capability with shared-memory capacity, communication bandwidth, and more connectivity compared to previous AMD Instinct GPU generations. Powering the AMD Instinct MI455X GPU is the company's latest CDNA 5 architecture, which introduces several improvements to accelerate and expand AI performance. Like its predecessor, the AMD Instinct MI355X, the new MI455X features eight Accelerator Complex Dies, or XCD, chiplets. However, where previously you had 32 Active Compute Units per XCD, the AMD Instinct MI455X GPU has the same number of Work Group Processors (WGPs), with each XCD divided into two Shader Engines (Ses) with 16 WGPs each. The new WGP design comes from CDNA 5, which optimizes performance and latency. Each WGP itself is comprised of four 32-thread SIMD (single-instruction, multiple-data) execution Units, and four scalar execution units. With parallel processing, peak throughput in math operations per clock (MXFP8 and MXFP4) has increased by up to 4X compared to the AMD Instinct MI355X, and up to 2X when looking at tensor and vector data. And when it comes to token throughput measured in tokens per second, the AMD Instinct MI455X GPU is reportedly an impressive 34 times faster than the Instinct MI355X. This in turn, leads to a significant reduction in token cost. In addition to the flagship AMD Instinct MI455X GPU, AMD has also launched the Instinct MI430X GPUs for sovereign AI and HPC, which can deliver up to 288 TFLOPS of hardware-based FP64 performance.

AMD says its newest AI server is in full production, will ship in months

The Helios servers will contain AMD's latest MI455X AI accelerator with its new Venice central processor, both of which are made by Taiwan Semiconductor Manufacturing Co. During Su's keynote address at an event in San Francisco, OpenAI's top computing executive joined Su to say that the ChatGPT creator plans to start using the Helios racks later this year. Advanced Micro Devices' second generation of AI servers, which aim to compete against Nvidia's flagship products, are in full production and will start shipping in the coming months, CEO Lisa Su said on Thursday. The Helios servers will contain AMD's latest MI455X AI accelerator with its new Venice central processor, both of which are made by Taiwan Semiconductor Manufacturing Co. During Su's keynote address at an event in San Francisco, OpenAI's top computing executive joined Su to say that the ChatGPT creator plans to start using the Helios racks later this year. "Helios is in full production," Su said, adding that shipments are slated to start at the end of the third quarter. "Customer demand ⁠for Helios ⁠is extremely strong." AMD is attempting to capture market share from Nvidia in the fast-growing data center chip sector, especially for inference computing, which is the data crunching that occurs when a user queries a chatbot such as OpenAI's ChatGPT. When asked at a press conference whether AMD was content to remain in second place, Su said, "We're actually doing another major leap forward. We actually believe we will have leadership in the scale-up compute domain." Su said AMD thinks the total computing market will hit $2 trillion by 2030, with $1.4 trillion of that from chips that speed up AI and $220 billion of it from central processing units (CPUs), an area in which it has long competed against Intel and that Nvidia has entered. The total computing market was worth an estimated $365 billion in 2025, according to ⁠AMD's estimates. "There's no one company that can solve it all," Su said. Cerebras Systems CEO Andrew Feldman joined Su on stage to say AMD and Cerebras would partner to offer a combination of their AI servers, first in Cerebras data centers. Cerebras was up about 5% and AMD down about ⁠2% as of the market close. AMD's share drop likely reflected overall weakness in the U.S. stock market and a selloff of semiconductor stocks, said Jacob Bourne, an analyst at Emarketer. Data centre hardware unveiling This week, Nvidia released a spate of technical details about its Vera CPU, which aimed to show that the chip, when combined with Nvidia's Rubin graphics processing unit (GPU), will do the best job at maximizing how much work AI agents can do with a given amount of electricity. In San Francisco, hundreds of executives and engineers gathered to take in technical presentations and mingle on a showroom floor. During Su's keynote, Sachin Katti, the vice president of compute strategy at OpenAI, said the company was embracing AMD's new hardware. "We expect that we'll be deploying Helios at massive scale starting towards the end of this year, and then accelerating throughout 2027," Katti said. Katti added that OpenAI plans to use AMD's next generation of MI500 AI chips. In October, AMD announced a multiyear deal with OpenAI that would also bring in tens of billions of dollars in annual revenue while giving the ChatGPT creator the option to buy up to roughly 10% of the chipmaker. On ⁠Wednesday, AMD announced plans to sell up to two gigawatts of its Instinct MI450 chips to AI lab Anthropic beginning in the first half of 2027. The deal also includes an investment of as much as $5 billion in the Claude maker. Tom Brown, an Anthropic co-founder, joined Su during her keynote on Thursday to say that Claude had been able to set up AMD's AI servers on its own, over a weekend. "Anyone, human or AI, can now build real models on your platform," Brown said. Su said in the press conference that she expects Anthropic to consume AMD chips through cloud computing partners and potentially through some of Anthropic's own data centers. Their partnership also focuses on helping developers use Claude to build software on AMD, she said. AMD displayed the Helios data center rack amid booths from cloud computing providers such as Vultr and TensorWave. Both cloud providers operate data centers with AMD hardware.

Microsoft will use AMD's AI-optimized Helios racks in Azure

Microsoft Corp. will use Advanced Micro Devices Inc.'s upcoming Helios rack design to power some Azure services. The cloud and operating system giant announced the partnership today alongside three new instance families. Helios is a reference design, a blueprint that AMD's manufacturing partners can use to make data center racks. Each system contains 72 of the chipmaker's upcoming Instinct MI455 graphics processing units. AMD has to date shared only a few details about the GPU. It will feature 432 gigabytes of HBM4 memory, 19.6 terabits per second of bandwidth and a new core architecture called CDNA 5. Helios' GPUs are supported by Pensando data processing units and Epyc central processing units. Pensando chips are optimized for infrastructure management tasks such as coordinating storage equipment and encrypting network traffic. According to AMD, they can run such workloads more efficiently than CPUs. That lowers costs while making more CPU capacity available for customer applications. The CPUs in Helios are from AMD's upcoming Venice data center processor series. In May, the company started ramping up production of the chips using Taiwan Semiconductor Manufacturing Co.'s two-nanometer process. The Venice series also uses a second TSMC technology called SoIC that makes it possible to stack chiplets atop one another. Helios organizes its CPUs, GPUs and DPUs in modules called trays. The trays are wider than a standard rack server to accommodate more hardware. They use liquid cooling to dissipate heat from their chips and exchange data using an open-source network protocol called UALoE. "AMD and Microsoft have spent years building high-performance infrastructure together, and today we're extending that partnership across the full stack of AMD AI solutions on Azure," said AMD Chief Executive Officer Lisa Su. AMD will start shipping Helios racks to Microsoft and other customers later this year. The tech giant will use the systems to power a new family of Azure instances called the ND MI455X v7 series. According to Microsoft, the virtual machines are optimized for inference workloads such as artificial intelligence agents and search tools. The company debuted the ND MI455X v7 series alongside two other instance families that will also run on AMD silicon. The HDv2 series is optimized for tasks that AI applications carry out using CPUs rather than CPUs. That includes the process of preparing datasets for analysis by AI agents. Each instance includes up to 500 Epyc Vulcan cores, four terabytes of memory and 32 terabytes of flash storage. The third addition to Azure's virtual machine portfolio is an instance series called HXv2. It's an improved version of an existing Azure instance series optimized for EDA, or electronic design automation, applications. Those are programs that engineers use to design chip. Microsoft says that HXv2 supports a broader range of workloads including scientific simulations. Each HXv2 virtual machine features 176 Epyc Vulcan cores with a clock speed exceeding 5GHz. According to Microsoft, each core will feature 50% more cache than previous-generation hardware. Customers can configure their virtual machines with up to four gigabytes of memory. "The significantly increased per VM and per core performance, and the inclusion of 800 Gb InfiniBand, enable large-scale MPI-based simulations and make HXv2 an ideal fit for a wide variety of HPC customers," Scott Guthrie, Microsoft's executive vice president of cloud and AI, wrote in a blog post. The company's new collaboration with AMD also extends to a technology called Azure Boost. It offloads the computations involved in running virtualization software from CPUs to more efficient, specialized chips. Microsoft will work with AMD to optimize Azure Boost for the latter company's products.

AMD Advancing AI 2026: Top News On AI Chips, CPUs, Robotics

AMD revealed new offers around Helios rackscale, MI400 GPUs, new Epyc CPUs, a robotics partner network and more during its annual conference. Advanced Micro Devices' new Robotics Partner Network, an updated roadmap for future semiconductor releases and the Helios rackscale artificial intelligence infrastructure platform were among the biggest reveals during AMD's annual Advancing AI conference. Although the Santa Clara, Calif.-based semiconductor maker's focus for the conference was on product innovation and reaching the growing total addressable markets (TAMs) in data center AI accelerators and central processing units (CPUs), AMD CEO Lisa Su told a CRN reporter after her keynote that services partners and system integrators remain an important section of her ecosystem looking ahead. "We want to make sure that enterprises have a way to easily deploy," Su said. "System integrators are a big piece of that." AAI runs through Thursday in San Francisco. [RELATED: AMD Expands Microsoft Foothold With New Azure AI Deal] AMD Advancing AI 2026 AMD has more than 100 partners in North America, according to CRN's 2026 Partner Program Guide. Chris Bogan, vice president of sales at Houston-based IBM solution provider Mark III Systems, told CRN in an interview that Mark III's CPU business, which includes AMD, has seen customers looking to the solution provider for help with optimization choosing the right components based on the workload. Although GPUs remain a big business and powerful force in bringing AI use cases into reality, CPU usage has been growing as customers invest in data centers and other hardware projects. "The demand had been there -- our chip business across the board has been growing -- it is just people are paying attention now versus I think before everything that was making headlines was only GPUs," Bogan said. Here are some of the biggest product, innovation and partnership reveals coming out of AMD's Advancing AI 2026 conference.

AMD Unleashes Instinct MI455X GPU, A 320 Billion Transistor Behemoth That Is Designed to Tackle NVIDIA's Rubin With 50% More HBM4 Memory & Up to 40 PFLOPs of AI Compute

AMD's Instinct MI455X GPUs extend the company's AI roadmap, bringing leading HBM4 capacities and over 40 PFLOPs of compute for Agentic AI, rivaling NVIDIA's Rubin chip. AMD Has An Answer To NVIDIA's Rubin, It's Called Instinct MI455X & It's An Engineering Marvel For Agentic AI With More HBM4 Than Any Other AI Chip On The Planet The AMD Instinct MI455X is the GPU that will power the Helios AI rack. This GPU is based on the latest CDNA 5 architecture, and packs 320 billion transistors, just 16 billion transistors shy of the NVIDIA Rubin chip. MI455X is designed to offer: * Purpose-Built AI Infrastructure: The AMD Instinct MI400 Series portfolio includes the AMD Helios rack-scale solution powered by AMD Instinct MI455X GPUs for frontier AI and AI factory deployments alongside AMD Instinct MI430X GPUs for sovereign AI and HPC. Together, the portfolio provides purpose-built solutions optimized for hyperscale AI factories, national infrastructure, research institutions and leadership-class HPC environments. * Leadership Performance Across AI and HPC: AMD Instinct MI455X GPUs deliver the compute, memory and networking performance required for high-volume inference, frontier-model training and fine-tuning. AMD Instinct MI430X GPUs deliver uncompromised accuracy and throughput across converged AI and HPC workflows with up to 288 TFLOPS of hardware-based FP64 performance for scientific computing. Across the portfolio, industry-leading HBM4 memory and high memory bandwidth help customers support larger models, reduce infrastructure complexity and improve efficiency at scale. * Open Software Foundation: Powered by AMD ROCm software, the MI400 Series provides an open software foundation with programming models, compilers, libraries, runtimes and deployment tools for AI and HPC. This commitment to open standards supports interoperability and long-term portability across software, system and networking choices, helping customers innovate without vendor lock-in. * Secure and Scalable by Design: AMD Instinct MI400 Series GPUs incorporate advanced security capabilities, secure boot, encrypted GPU-to-GPU links and hardware-based protections to help safeguard sensitive AI and HPC workloads. Combined with scalable system architectures - from the AMD Helios rackscale solution for frontier AI to traditional mesh-based HPC deployments - AMD Instinct MI400 series GPUs enable customers to deploy AI infrastructure with flexibility, confidence and long-term operational consistency. For the Instinct MI400 series, AMD will have three products; the first two are the Instinct MI455X & the MI450X, which are aimed at scale AI Training & Inference workloads. The MI455X is powering the Helios rack. There's also a cost-optimized 6-HBM variant. The third chip is the MI430X, which is aimed at HPC & Sovereign AI workloads, featuring the "highest performance" FP64 capabilities, hybrid compute (CPU+GPU), and the same HBM4 memory as the MI455X. The AMD Instinct MI455X is a 40 PFLOPs of FP4 & 20 PFLOPs FP8 compute, which is double the compute capability of the MI350 series, making it a disruptive offering for AI. For comparison, an NVIDIA Rubin GPU offers 50 PFLOPs of FP4 and 17.5 PFLOPs of FP8 compute. In addition to the compute capability, AMD is also going to leverage HBM4 memory for its Instinct MI400 series. The new chip will offer a 50% memory capacity uplift from 288GB HBM3e to 432GB HBM4. The HBM4 standard will offer a massive 19.6 TB/s bandwidth, more than double that of the 8 TB/s for the MI350 series. For comparison, the Rubin GPU comes with 288 GB of HBM4 at 22 TB/s. AMD has positioned its Instinct MI400 GPUs against NVIDIA's Vera Rubin, and the high-level comparison looks something like the following: * 1.5x Memory Capacity vs Competition * Same Memory Bandwidth vs Competition * Same FP4 / FP8 FLOPs vs Competition * Same Scale-Up Bandwidth vs Competition * 1.5x Scale-Out Bandwidth vs Competition For the MI400 series, there will be two products; the first one is the Instinct MI455X, which is aimed at scale AI Training & Inference workloads. The other one is MI430X, which is aimed at HPC & Sovereign AI workloads, featuring hardware-based FP64 capabilities, hybrid compute (CPU+GPU), and the same HBM4 memory as the MI455X. In 2027, AMD will be introducing its next-gen Instinct MI500 series AI accelerators. Since AMD is shifting to an annual cadence, we are going to see updates on the datacenter and AI front at a very rapid pace, similar to what NVIDIA is doing now with a standard and an "Ultra" offering. These will be used to power the next-gen AI racks and will offer a disruptive uplift in overall performance. According to AMD, the Instinct MI500 series will offer next-gen compute, memory, and interconnect capabilities. AMD Instinct AI Accelerators: Follow Wccftech on Google to get more of our news coverage in your feeds.

AMD Unveils Helios AI Platform to Power Agentic AI at Scale

As AI expands from training to inference and agentic workloads, compute demand is accelerating rapidly. AMD delivers an open, full-stack AI platform that gives customers the flexibility to deploy the right compute for every workload. AMD launched its next-generation AI infrastructure and physical AI portfolio at Advancing AI 2026, led by AMD Helios rackscale solutions, now in production to be deployed by leading AI companies at gigawatt scale. As AI expands from training to inference and agentic workloads, compute demand is accelerating rapidly. AMD delivers an open, full-stack AI platform that gives customers the flexibility to deploy the right compute for every workload. "The next phase of AI will span frontier models, agents and physical AI, creating new opportunities to bring intelligence everywhere," said Dr. Lisa Su, chair and CEO, AMD. "Realizing that potential will take the entire industry working together. AMD is partnering across the ecosystem to deliver leadership compute and open platforms that give customers the performance, flexibility and choice to scale AI from the data center to the edge." AMD Helios: The Highest Performance Rack-Scale AI Solution Delivering frontier AI requires a fully integrated rack architecture, with every part of the stack pushing the boundaries of performance. AMD Helios rackscale solutions are built for this, with co-optimized silicon spanning 72 high-performance AMD Instinct™ MI455X GPUs and 18 powerful 6th Gen AMD EPYC™ "Venice" CPUs, connected by AMD Pensando™ front-end, scale-up and scale-out networking, and accelerated by AMD ROCm™ open software. AMD Helios combines leadership compute performance, memory capacity and networking bandwidth to deliver up to 30% more tokens per dollar than the leading competitive solution1. Leading AI labs and cloud providers are choosing AMD Helios for its open, full-stack performance. They include OpenAI, Anthropic, Meta, Microsoft, Oracle, HUMAIN, Tensorwave, Vultr, Cirrascale and others. Systems will be available from leading OEMs, including Bull, HPE, Lenovo and Supermicro, as well as infrastructure partners Sanmina and Wiwynn. At Advancing AI, AMD partners detailed how they deploy AMD AI infrastructure at scale for frontier training and inference: * OpenAI and AMD are partnering to optimize the full AI stack, from silicon to software. Leveraging OpenAI's Triton framework with AMD ROCm software, the companies are optimizing GPT-class workloads on AMD Instinct MI455X GPUs and AMD Helios racks. OpenAI expects to bring Helios online beginning in the fourth quarter of 2026, with deployments accelerating throughout 2027. * Meta and AMD are co-designing for gigawatt-scale deployments, optimizing AMD's full AI compute stack for Meta workloads. Meta is now validating 6th Gen EPYC CPU platforms in its labs and has begun testing and validating workloads on AMD Helios racks as they prepare to deploy at scale. * Cerebras and AMD are collaborating to deliver a combined solution of Cerebras ultra low-latency AI compute and AMD Helios high-throughput rack-scale infrastructure to help improve inference efficiency, scalability and economics for ultra-low-latency inference serving. Delivering the Highest Performance Data Center CPUs and GPUs 6th Gen EPYC processors deliver the broadest server CPU portfolio for agentic AI,2 spanning cloud, enterprise, general-purpose and high-performance computing (HPC) workloads. With leadership per-core performance and the highest thread density3, they enable the most agents per watt, per dollar and per rack.4,5,6 For AI host nodes, 6th Gen EPYC CPUs deliver the speed and memory bandwidth to keep accelerators fully fed. And for general-purpose servers, they bring leadership performance and energy efficiency to run business critical applications and AI support tasks. With AMD Instinct™ MI400 Series GPUs, AMD delivers powerful performance for cloud, enterprise and HPC workloads. AMD Instinct MI455X GPUs deliver 34x higher token throughput compared to MI355X GPUs7. For high-precision workloads, the AMD Instinct™ MI430X accelerator is the most advanced for HPC and sovereign AI with up to 288 TFLOPS of hardware-based FP64 performance for scientific computing. Instinct MI430X accelerators are powering the next wave of exascale-class supercomputers across the U.S. and Europe. AMD also launched the Instinct MI350P GPU, bringing seamless AI acceleration to existing infrastructure with leadership token economics. MI350P GPUs deliver up to 4.2x more tokens per second per dollar than the competition8. Advancing the Open Software Ecosystem For developers, AMD ROCm is the open software platform with the performance, flexibility and ecosystem support needed to build and deploy AI on AMD hardware. Building on that foundation, AMD is introducing ROCm.ai, an AI-driven development platform that helps developers build, optimize and deploy GPU software faster across AMD platforms. ROCm.ai brings AI-assisted GPU programming to developers by enabling popular coding agents such as Claude, Codex and Cursor to understand AMD platforms and ROCm natively. ROCm.ai is accelerating software enablement for AMD Instinct MI455X GPUs while optimizing performance. Leading open-source frameworks including PyTorch, Hugging Face, vLLM and SGLang are already enabled on MI455X and seeing great results. Accelerating Next-Generation AI Infrastructure AMD is extending its annual cadence of CPU, GPU, networking and rack-scale innovation through 2030. The company shared new details on its roadmaps, including: At Advancing AI, AT&T illustrated how it is deploying flexible enterprise AI using AMD technology across cloud, on-premises and air-gapped environments. AT&T is also using AMD Instinct GPUs and ROCm software to power its OTel 2.0 model, an open-source model trained specifically for telecoms. With the AMD Ryzen™ AI Halo developer platform, AMD delivers performance, efficiency and simplicity that makes local AI development accessible. More AMD Ryzen AI Halo platforms, powered by Ryzen™ AI Max PRO 400 Series processors, will be available later this year from AMD and OEM partners. Cisco and AMD are collaborating to combine AMD high-performance inference engines, including AMD Ryzen AI Halo systems, with Cisco networking, observability and security capabilities, so enterprises can deploy, govern and manage hybrid and local agentic AI at scale. Advancing the Next Frontier of Physical AI As AI expands across cloud, enterprise and local systems, the next frontier is bringing intelligence into machines that perceive, reason and act in the physical world. Building on a long legacy in robotics with AMD FPGAs and adaptive SoCs, AMD introduced AMD Kria™ AI solutions, extending the company's robotics capabilities from the robot body to the robot brain. AMD uniquely brings AI perception, reasoning and agentic decision-making and control together on a single platform to deliver the performance required for demanding real-world robotic systems. The portfolio includes new AMD Kria AI system-on-modules (SOMs), powered by the new AMD Ryzen AI Embedded X100 Series processors, and the AMD Kria AI Robotics Developer Platform, the first open, turnkey integrated platform for autonomous robotics combining CPU, GPU, NPU and FPGA compute. Together with an expanded open software ecosystem, AMD Kria AI solutions remove vendor lock-in and help developers and customers accelerate the path from prototype to production for next-generation physical AI systems.

AMD Expands Microsoft Foothold With New Azure AI Deal

Dominic Daninger, vice president of engineering at Burnsville, Minn.-based systems integrator Nor-Tech, tells CRN that it's good to see AMD attacking the data center opportunity with its latest collaboration with Microsoft. AMD said Monday that it expanded its collaboration with Microsoft in a deal that would see the chipmaker deploy its latest Helios AI rack-scale architecture and next-generation EPYC data center processors to Azure. As part of the deal, Redmond, Wash.-based Microsoft will deploy Helios to power "frontier model AI inference for Microsoft, its AI customers and support Azure AI services," the companies said. AMD and Microsoft both said that Azure will also add two new AMD EPYC CPU-powered VM series and broaden its deployment of Pensando data processing units (DPUs) to support Azure networking services. Santa Clara, Calif.-based AMD will begin shipping Helios to customers, including Microsoft, in the second half of 2026. Dominic Daninger, vice president of engineering at Burnsville, Minn.-based systems integrator Nor-Tech, told CRN that with more deployments of chips, "sometimes parts might be a little harder to get" given the current global chip shortage. But it's good to see AMD attacking the data center opportunity, said Daninger, who praised AMD CEO Lisa Su's leadership. "Competition always breeds innovation." Helios rack-scale architecture integrates AMD Instinct GPUs with EPYC Venice CPUs to deliver high-performance AI infrastructure. In AMD's quarterly earnings call in May, Su said that the company is "seeing significant customer interest" around Helios and AMD's Instinct MI450 data center AI accelerator chip. Su cited AMD's partnerships with OpenAI and Meta, saying those collaborations are "going really well," according to a transcript of the call. "We appreciate the deep co-engineering that has gone on there," said Su at the time. "Based on our current visibility, how those forecasts are coming in with all of our customers, we're actually seeing it above our initial plans that we had planned for 2027. I think the encouraging thing is we're seeing a breadth of customers who are now very interested in deploying at significant scale, MI450 series." The Microsoft deal would significantly expand the availability of the Helios product to another technology powerhouse. "Our approach to AI infrastructure is designed to support the breadth of how AI systems are built and run," said Scott Guthrie, Microsoft's executive vice president of cloud and AI, in a blog post. "We closely work with industry innovators like AMD as well as our own purpose-built silicon and systems to provide customers with a comprehensive, open and heterogenous platform to achieve the best performance, cost and energy efficiency outcomes." An AMD spokesman declined to disclose additional details on the scale and the financial terms of the collaboration. A Microsoft spokesperson pointed CRN to Microsoft's blog post and AMD press release. Shares of AMD stock rose nearly 2 percent on Monday to $503.57. In the last quarter, AMD grew its data center revenue to a record $5.8 billion -- once again higher than what its larger rival, Intel, made for the segment. Even with the growth in sales, AMD is still behind leader Nvidia in sales, which reported record data center revenue of $75.2 billion in its last quarter, up 92 percent from a year ago.

AMD Unveils Helios, Its Next-Gen AI Powerhouse With MI455X & 6th Gen EPYC, Challenging NVIDIA's Rack-Scale Dominance

AMD has unveiled its next-generation Helios AI Rack, which aims to become the leading solution for frontier AI and sovereign computing, powered by its MI455X GPUs and 6th Gen EPYC CPUs. AMD's Helios AI Racks Are A Sight To Behold: Combining Next-Generation Instinct GPUs, EPYC CPUs, & Pensando Networking Chips For Large-Scale Inferencing It was last year at Advancing AI 2025 when AMD gave us the first glimpse at its Helios AI rack, a solution designed to meet the growing demands of AI workloads in the Agentic era. After giving us the official unveiling, we got to see the first Helios platforms in the flesh at OCP 2025. And this year, we got to see Helios Rack on full display at the Computex show floor. The plan is simple: to roll out the platform that is the best at AI and disrupts the current market hierarchy dominated by NVIDIA's existing Oberon and upcoming Kyber racks. This will be AMD's first full-stack rack-level solution for AI and brings AMD's open standards to the industry. But the Helios AI rack isn't just a standard rack with some hardware components; full stack means full stack, and AMD has deployed its latest and greatest to ensure that firms using Helios get the full taste of its capabilities. Although AMD's Helios AI Rack will officially launch at the Advancing AI 2026 event, the company has officially disclosed the full platform details, so we decided to give you all a detailed rundown. There are three key components of the Helios AI rack: * AMD Instinct MI455X GPUs * 6th Gen AMD EPYC CPUs * AMD Pensando AI NICs CNBC got a real nice close-up of these chips, which you can see below: There are several other technologies that power Helios, which include: * AMD Pensando DPU * AMD Infinity Fabric * AMD ROCm Software Stack So let us start by looking at the key components that power the AMD Helios AI platform. AMD Instinct MI455X - The Inference GPU Powerhouse For Helios The AMD Instinct MI400 series are the GPUs that will power the Helios AI rack. These GPUs are based on the latest CDNA 5 architecture, which brings: * Increased HBM4 Capacity & Bandwidth * Expanded AI Formats with Higher Throughput * Standard-Based Rack-Scale Networking (UALoE, UAL, UEC) For the Instinct MI400 series, AMD will have three products; the first two are the Instinct MI455X & the MI450X, which are aimed at scale AI Training & Inference workloads. The MI455X is powering the Helios rack. The third chip is the MI430X, which is aimed at HPC & Sovereign AI workloads, featuring the "highest performance" FP64 capabilities, hybrid compute (CPU+GPU), and the same HBM4 memory as the MI455X. The AMD Instinct MI455X is a 40 PFLOPs of FP4 & 20 PFLOPs FP8 compute, which is double the compute capability of the MI350 series, making it a disruptive offering for AI. For comparison, an NVIDIA Rubin GPU offers 50 PFLOPs of FP4 and 17.5 PFLOPs of FP8 compute. In addition to the compute capability, AMD is also going to leverage HBM4 memory for its Instinct MI400 series. The new chip will offer a 50% memory capacity uplift from 288GB HBM3e to 432GB HBM4. The HBM4 standard will offer a massive 19.6 TB/s bandwidth, more than double that of the 8 TB/s for the MI350 series. For comparison, the Rubin GPU comes with 288 GB of HBM4 at 22 TB/s. AMD Instinct AI Accelerators: AMD 6th Gen EPYC Venice - The Host CPU For Helios & The First Zen 6 On Market The other key component, and the one that is really shaping up as the king of the Agentic AI era, is the CPU. For Helios, AMD is leveraging its brand new Zen 6 core architecture that will be used on its 6th Gen EPYC CPUs, codenamed Venice. AMD's EPYC Venice chips are the first HPC product to enter volume production on TSMC's 2nm process technology, and their readiness for Helios AI racks proves that AMD will always deliver the best of the best for its high-performance customers. The TSMC 2nm process technology transitions from FinFET to Nanosheet transistors (GAA), and offers 10-15% higher performance at the same power, 25-30% lower power consumption at the same performance, and up to 15% higher transistor density. AMD's 6th Gen EPYC Venice CPUs were teased earlier this year at CES, offering up to 256 cores and 512 threads with eight massive compute dies and two even larger I/O dies. For this chip, AMD has promised over 70% performance & efficiency improvement with its EPYC Venice CPUs, with a >30% increase in thread density. Once again, the requirement for high-performance CPUs has risen tenfold with Agentic AI workloads. And competitors are firing across all cylinders with CPUs that are optimized around agents. Just like GPUs, NVIDIA is AMD's primary competitor in this space, which is leveraging Vera CPUs based on its custom Arm IP to power the Vera Rubin NVL72 racks. AMD's Helios AI racks powered by Zen 6 EPYC chips not only offer more cores, but faster performance and improved single-core capabilities, which is important for sustained throughput. In early benchmarks, AMD has shown that not only do its 5th Gen Turin "Zen 5" CPUs lead against Vera, but the 6th Gen Venice "Zen 6" chips lead massively, driving more performance & offering better TCOs at a similar power scale. AMD EPYC CPU Families: AMD Pensando "Vulcano" AI NICs & Pensando "Salina" DPU - Enabling Heightened Scale-Up & Scale-Out Networking/Fabric Solutions Networking and Fabric play a core role in shaping data center and enterprise solutions. AMD's Pensando stack delivers these technologies, and Helios is fitted with the latest Vulcano 800 AI NIC & Salina DPU, which rival NVIDIA's ConnectX-8 and Bluefield 3/4 DPUs. Starting with the AMD Pensando "Vulcano" 800 AI NIC, this is an 800 Gbps high-performance switch with 800 Gbps Ethernet network throughput. It is currently the only NIC to offer up to 2.4 Tbps of scale-out bandwidth per GPU, backed by full hardware and software programmability. Each GPU gets access to up to 8x the scale-out bandwidth, and with a UAL / PCIe Gen6 host interface, Helios drives super-low-latency communication between the GPUs. Vulcano is also UEC-ready with an RDMA Ethernet that is optimized for large-scale AI clusters. The scale-up fabric is built on an open networking approach, leveraging UALink over Ethernet (UALoE) to deliver high-bandwidth, low-latency GPU connectivity at rack scale. The fabric provides seamless interconnect for up to 72 GPUs, which is the full GPU count that each Helios AI rack supports. The AMD Pensando DPU bridges the AI servers to enterprise networks, accelerating network, security, and storage offloads to boost AI server efficiency. The Salina DPU features 16 Arm N1 cores that are used for offloading networking, security, and storage for front-end server-to-client connectivity. Each Salina DPU offers a 40% speed-up versus CPU-only processing and is twice as capable as AMD's prior generation DPUs while offering 40% performance uplifts versus NVIDIA's BlueField-3 DPUs. Helios - The Ultimate AI Rack From The High-Performance Leaders With all three core components outlined, we now want to shed some light on how Helios AI racks bring them together. The AMD Helios AI Rack makes use of Meta's Open Rack Wide standard submitted to the OCP (Open Compute Project). The Helios rack is a fully liquid-cooled design with 18 compute trays with six switches, and each tray features four Instinct MI455X GPUs and a single EPYC Venice "Zen 6" CPU. The system leverages AMD's Pensando "Salina" 400 DPU & the Pensando "Vulcano" 800 AI NIC for networking and interconnection. Each AMD EPYC Venice "Zen 6" CPU comes with up to 256 cores based on the Zen 6C architecture, and each Instinct MI455X GPU packs several thousand compute units. There are a total of 72 GPUs onboard the Helios AI Rack. Each GPU is placed under a liquid cold plate made out of copper. All put together, the Helios AI rack weighs roughly 5000 pounds and costs around $5-$5.5 million, as per CNBC. Each rack will consume around 225-245kW of power. The Helios rack scales up to 2.9 Exaflops of FP4 compute, 1.4 Exaflops of FP8 compute, 31 TB of HBM4 memory, 1.4 PB/s of aggregate bandwidth, 43 TB/s of scale-out bandwidth, 260 TB/s of scale-up interconnect bandwidth, and up to 4600 CPU + 18,000 GPU cores. All of these enable Helios to deliver a generational leap for trillion-parameter model training and large-scale AI inference. Comparison Table Software Stack & An Impressive Line of First Customers ROCm is AMD's industry standard for AI and HPC. Being the direct competitor to NVIDIA's CUDA, AMD has come a long way with its ROCm stack, now in v7.14. ROCm brings open hardware and software standards that reduce integration complexity and accelerate AI deployment across all AMD platforms. For Helios, AMD is bringing native support for leading frameworks such as PyTorch, TensorFlow, JAX, Hugging Face, vLLM, SGL, Deepspeed, ONNX, llm-d, OpenXLA, MLID, Llama Stack, and many more with Day-0 support. And now, coming to why this all matters. The explosive demand for Agentic AI has led to increased demand for more compute, and Helios brings with it some massive capabilities. Today, AMD and Microsoft announced that Helios Rackscale solutions will be deployed for its customers and also to support Azure AI services. The following are some of the Azure instances that will be offered by Microsoft: * Azure HDv2 (CPU-Focused) - Nearly 500 6th Gen EPYC CPU Cores, 4 TB RAM, 32 TB NVMe Storage, 400 Gb Azure Boost Networking * Azure HXv2 (Agentic AI Driver) - 176 6th Gen EPYC CPU Cores per VM (3D V-Cache, 5 GHz+ Clocks, 50% More Cache), 2-4 TB RAM, 800 Gb Infiniband * ND MI455X v7 (Helios Rack) - 72 MI455X GPUs, 6th Gen AMD EPYC CPUs Besides Microsoft, AMD has also announced various customers for its Helios AI Rackscale solutions, 6th Gen EPYC CPUs, and MI455X GPUs, which include OpenAI, META, Oracle, HPE, TCS, Celestica, Nutanix, US DOE. AMD's Helios AI Rack represents a bold and comprehensive leap forward in the Agentic AI era, delivering a fully integrated, open-standards rack-scale platform that combines cutting-edge Instinct MI455X GPUs, powerful 6th-Gen EPYC Venice CPUs, and advanced Pensando networking technologies into a liquid-cooled powerhouse. With superior HBM4 memory capacity, exceptional scale-up and scale-out bandwidth, and a mature ROCm software ecosystem offering broad framework support, Helios is positioned to challenge the established order by providing higher memory density, greater flexibility, and compelling performance-per-dollar advantages over competing solutions. Backed by an impressive roster of early adopters including Microsoft, OpenAI, Meta, and Oracle, this innovative system promises to accelerate trillion-parameter model training and large-scale inference while fostering a more open and interoperable AI infrastructure. As it prepares for official launch at Advancing AI 2026 in just a few days, Helios underscores AMD's commitment to high-performance leadership and stands as a compelling testament to the rapid evolution of AI hardware. Follow Wccftech on Google to get more of our news coverage in your feeds.

AMD just landed its biggest AI deal yet

AMD has broken a chip monopoly before. In the 2000s, it took server market share away from Intel's dominant x86 processors, forcing years of price and performance competition that reshaped enterprise computing. Two decades later, AMD is attempting a similar disruption, this time against Nvidia's grip on the chips that power artificial intelligence. That earlier fight took years to move the needle. This one appears to be moving faster. On Monday, July 20, AMD landed its most significant customer yet for Helios, its first rack-scale AI system, and the buyer has one of the deepest existing relationships with Nvidia in the industry. Microsoft said it will deploy Helios racks across its Azure data centers to power AI inference for itself and its cloud customers, joining Meta, OpenAI and Oracle as early adopters of the platform. Microsoft framed the announcement around choice rather than replacement. Azure customers want "performance, scale and choice," Microsoft CEO Satya Nadella said, according to Seeking Alpha. AMD shares rose as much as 5% Monday. Rosenblatt and UBS separately raised their price targets on the stock the same day, to $655 and $700, respectively, according to Yahoo Finance. Microsoft shares were little changed, a sign the deal matters far more to AMD's growth story than to Microsoft's infrastructure budget. Microsoft's three-vendor strategy Microsoft already runs Nvidia GPUs and its own Maia chips inside Azure, a dual-track approach common among hyperscalers hedging against reliance on a single vendor. Helios adds a third supplier to that mix rather than replacing either of the other two. Wall Street had already priced some of this in. Jefferies analysts told clients they believe Microsoft was already using AMD's MI450 chips based on industry checks, before Monday's announcement made the relationship official, according to TipRanks. The commitment goes beyond chips. Helios bundles AMD's Instinct MI455X GPUs, EPYC Venice CPUs, Pensando networking hardware and ROCm software into one integrated rack. AMD's Venice CPU began its production ramp in May, with more customers validating it than any earlier EPYC generation. That is a full stack architecture decision, not a one time purchase. AMD is not winning on price, and Nvidia is not standing still The common assumption is that AMD competes with Nvidia by undercutting it. The numbers say otherwise. Futurum Group estimates Helios racks will cost between $5 million and $5.5 million, compared with $3.5 million to $4 million for Nvidia's next-generation Vera Rubin system, according to CNBC. AMD's pitch instead rests on cost per completed AI task rather than sticker price, a framing its executives have repeated at nearly every public appearance this year. Wolfe Research analyst Chris Caso said Microsoft's decision to return to AMD GPUs "attests to Helios' competitiveness," according to CNBC. Nvidia has already signaled it will not cede ground. The company has unveiled its next generation Kyber rack, which will double the number of GPUs per rack to 144 from 72. The stakes explain the urgency on both sides. AMD's data center revenue climbed from $6.5 billion in 2023 to $16.6 billion in 2025, and analysts expect it to reach roughly $31.2 billion this year, according to Yahoo Finance. Bloomberg / Getty Images Wall Street is re-rating AMD faster than the news itself Monday's price target increases did not happen in isolation. Goldman Sachs raised its target to $640 from $450 earlier this month, citing surging demand for high performance CPUs tied to agentic AI workloads. AMD now carries an 82.4% buy rating among covering analysts, according to Bloomberg data cited by Yahoo Finance. Jefferies has a $615 target and expects AMD to announce Anthropic as a customer at its Advancing AI event this week, on top of the Microsoft, Meta and OpenAI relationships it already has, according to TipRanks. Analysts covering the stock are split 28 buy ratings, eight holds and zero sells, with an average price target implying further upside, according to TipRanks. That conviction exists despite AMD holding a small piece of the market it is chasing. Nvidia still controls more than 95% of data center GPUs, while AMD holds roughly 4.5%. Wall Street is betting that gap narrows faster than the headline share numbers suggest. The end of single supplier reliance? Microsoft's three-vendor approach fits a pattern of building since the pandemic-era chip shortage exposed how fragile single-supplier dependency can be. Cloud providers that once bought almost exclusively from Nvidia are increasingly spreading orders across multiple chipmakers, and in some cases building custom silicon of their own. That shift changes the competitive question facing Nvidia. It is no longer only whether Nvidia's chips are the fastest available. It is whether hyperscalers are willing to concentrate hundreds of billions of dollars in annual capital spending with a single supplier when credible alternatives now exist. AMD still has to prove Helios performs at scale once it leaves the lab, and its next test comes fast. The company's Advancing AI event this week, where Jefferies expects an Anthropic announcement, will show whether Monday's momentum was a one time win or the start of a broader customer list that makes Nvidia's dominance look less permanent than it did a year ago.

AMD Puts Nvidia in Its Sights with Latest Rack-scale System Launch

Chipmaker AMD has finally showcased its networking stack Helios and in the process put Nvidia's NVL72 into its crosshairs. This rack-scale system is designed to power the computing needs of the world's largest AI labs via a programmable front-end, a single hop scale-up fabric, high bandwidth, and flexi-protocol AI network interface cards. The company made the revelations at its Advancing AI Conference held in San Francisco on Thursday where Chairperson and CEO Dr. Lisa Su promoted Helios and announced that it would be shipping it to a growing list of customers including Microsoft later this year. Su also pitched AMD's latest chip series designed to power the AI industry. The company also reported that the rack-scale system's performance metrics while also noting that Anthropic joined a growing list of AI labs, neo-clouds and hyper-scalers who have since signed up to deploy Helios. These rack systems combine several processors into a single high-powered unit and are built for datacentres where AI models get trained and perform compute-intensive workloads. Su noted that Helios was the tech industry's "highest-performance AI rack" that is "built to train and run the most demanding frontier models in the world at massive scale." Over the past years, Nvidia has dominated this market with its Vera Rubin and Grace Blackwell rack-scale systems, which now AMD is looking to break into and the performance metrics that the company shared does appear to give it a chance to compete strongly. In fact, a report in The Register noted that Helios actually beat Vera Rubin on several metrics. Incidentally, AMD had revealed Helios in 2025 and had promoted it during the CES 2026, which resulted in several known customers including the likes of OpenAI, Meta and Oracle, besides Microsoft and Anthropic. Microsoft CEO Satya Nadella had stated last week that his company would be expanding its Azure infrastructure with Helios. At the same time Anthropic also announced a strategic partnership with AMD to deploy up to two gigawatts of GPUs via the new rack system. Another reveal from AMD during the Thursday event was its Venice-X CPU designed for datacentres that handles high-computing workloads. The Venice-X is expected to launch in 2027 with Su noting that by 2030 chips that power AI will become a massive part of the overall computing market to adjust to the "step change in compute demand" driven by the rise of agentic AI options. "When you ask the agent to do something, it actually has dozens of steps, and it has to reason, and it has to call tools, and it has to access data, and it has to keep doing it over and over until it solves the problem, and so you need lots of GPUs to do all that," Su told the gathering. "We're now expecting that by 2030, the AI accelerator market is going to reach about $1.4 trillion. What that means is, by the end of the decade, the AI accelerator market is going to approach the size of the entire semiconductor market today." "We do expect that GPUs are going to make up the vast majority of that market because the algorithms are still very much in their infancy, and we're still continuing to see the workloads change, and that favours programmability in the overall silicon ecosystem," she revealed. On the front-end, Helios leverages AMD's Salina 400 Gb/s (400G) data processing unit where the accelerator chip is programmable and can be upgraded live without traffic disruption. It adds a security layer for the front-end allowing users to encrypt all traffic while also offloading infrastructure tasks from the central processing unit to free up valuable cores. The scale-up happens within a Helios pod that works on a UALink over Ethernet and pools 72 M1455X GPUs to generate a bandwidth of 260 TB/second per pod. The design also includes a single-tier one-hop Ethernet fabric that supports 31 TBs of high-bandwidth memory hour. In addition AMD also offers fixed latency, congestion avoidance and resilience. However, it is in the rack-to-rack network where Helios really stands up to Nvidia as it is built on the second-generation Pensando AI NIC dubbed the "Vulcano" and packs three NICs per GPU to provide a highly dense configuration capable of delivering 43 TB/s, which is actually 50% more bandwidth than Nvidia's Vera Rubin NVL72, AMD claims. In a media briefing, Soni Jiandani, SVP and GM of AMD's networking technology and solutions group, said the shift from LLMs training to inferencing and agentic workflows means that the data movement and latency on the network are core determinants of throughput, GPU utilisation, and costs per token. "While agentic AI is driving even more interactions between the compute, memory, and storage, the common denominator in all of this is data movement at scale, where the network is fundamental to deliver the performance in a deterministic manner and have the ability to do it at a system level, she said.

Microsoft to Deploy AMD Helios and 6th Gen EPYC Chips to Power Azure AI

At the center of this expansion, Microsoft will deploy the AMD Helios Rackscale Solution, to power frontier model AI inference for Microsoft, its AI customers and support Azure AI services. AMD announced an expanded strategic partnership spanning AMD GPUs, CPUs, networking and software on Microsoft Azure. At the center of this expansion, Microsoft will deploy the AMD Helios Rackscale Solution, to power frontier model AI inference for Microsoft, its AI customers and support Azure AI services. Azure will also add two new AMD EPYC CPU-powered VM series and broaden its deployment of Pensando DPUs to support Azure networking services. AMD will begin shipping Helios to customers, including Microsoft, in the second half of 2026. AMD Helios combines AMD Instinct™ MI455X GPUs, AMD EPYC™ "Venice" CPUs, Pensando™ networking and ROCm™ software in an open, integrated rackscale platform built for large-scale AI training and inference. The Azure deployment will use Helios for inference workloads spanning frontier models, Azure AI services and customer applications. "AMD and Microsoft have spent years building high-performance infrastructure together, and today we're extending that partnership across the full stack of AMD AI solutions on Azure," said Dr. Lisa Su, Chair and CEO, AMD. "Microsoft's new AMD deployments mark an important milestone as we deliver leadership compute solutions to Azure customers and scale the next generation of AI infrastructure together." "Customers are looking for AI infrastructure that is optimized for a wide range of workloads, from training and inference to data preparation, search, and reinforcement learning," said Satya Nadella, Chairman and CEO, Microsoft. "Through our collaboration with AMD, we are expanding the Azure infrastructure portfolio with AMD Helios to give customers the performance, scale and choice they need to build and run the next generation of AI applications." The collaboration expands access to AMD AI infrastructure across Azure. Frontier model builders can now leverage AMD-powered infrastructure to train and serve large-scale AI models, while enterprise customers can deploy and scale production AI workloads through Azure Foundry Managed Compute. Azure's new VM series, Azure HDv2 for agentic AI and data pipelines, and Azure HXv2 for semiconductor design, will be powered by 6th Gen AMD EPYC "Venice" processors. Together, the new VM series broaden Azure's AMD EPYC portfolio across AI, data and engineering workloads. The collaboration also extends into the networking layer that connects and scales Azure infrastructure. Building on Microsoft's broad deployment of AMD Pensando DPUs, the companies are integrating Azure Boost with AMD technologies to improve networking performance, efficiency and connection processing at cloud scale. As AI demand accelerates, AMD and Microsoft will continue to deliver open, high-performance infrastructure that gives customers flexibility, efficiency and scale to build what's next.

AMD Unveils Next-Generation AI Infrastructure in Direct NVIDIA Challenge

The company plans to present its Helios server racks, Instinct MI450 chips, and Venice processors. AMD aims to gain ground against NVIDIA in the fast-growing AI infrastructure market. AMD is expected to formally present Helios, its first full server rack for artificial intelligence workloads. The system combines processors, graphics chips, networking tools, and software. Helios will use AMD Instinct MI455X graphics processors from the MI450 series. It will also include EPYC Venice processors and Pensando networking technology. The rack-scale design targets both AI training and inference. Inference covers the computing work that occurs when users send requests to AI services. AMD will market Helios against NVIDIA's integrated server systems. NVIDIA is preparing its second-generation platform using Rubin graphics chips and Vera processors. NVIDIA recently released more technical details about Vera. The company says the processor will improve work completed by while controlling power use. Meanwhile, AMD displayed Helios at the Moscone West convention center. Cloud providers Vultr and TensorWave also presented systems using AMD hardware.

AAI 2026: AMD Delivers Full-Stack Compute for the Agentic AI Era

AMD launches broad high-performance computing and physical AI portfolios, including its first rack-scale AI solution and the world's most powerful AI rack, AMD Helios. AMD today launched its next-generation AI infrastructure and physical AI portfolio at Advancing AI 2026, led by AMD Helios rackscale solutions, now in production to be deployed by leading AI companies at gigawatt scale. As AI expands from training to inference and agentic workloads, compute demand is accelerating rapidly. AMD delivers an open, full-stack AI platform that gives customers the flexibility to deploy the right compute for every workload. "The next phase of AI will span frontier models, agents and physical AI, creating new opportunities to bring intelligence everywhere," said Dr. Lisa Su, chair and CEO, AMD. "Realizing that potential will take the entire industry working together. AMD is partnering across the ecosystem to deliver leadership compute and open platforms that give customers the performance, flexibility and choice to scale AI from the data center to the edge." AMD Helios: The Highest Performance Rack-Scale AI Solution Delivering frontier AI requires a fully integrated rack architecture, with every part of the stack pushing the boundaries of performance. AMD Helios rackscale solutions are built for this, with co-optimized silicon spanning 72 high-performance AMD Instinct™ MI455X GPUs and 18 powerful 6th Gen AMD EPYC™ "Venice" CPUs, connected by AMD Pensando™ front-end, scale-up and scale-out networking, and accelerated by AMD ROCm™ open software. AMD Helios combines leadership compute performance, memory capacity and networking bandwidth to deliver up to 30% more tokens per dollar than the leading competitive solution1. Leading AI labs and cloud providers are choosing AMD Helios for its open, full-stack performance. They include OpenAI, Anthropic, Meta, Microsoft, Oracle, HUMAIN, Tensorwave, Vultr, Cirrascale and others. Systems will be available from leading OEMs, including Bull, HPE, Lenovo and Supermicro, as well as infrastructure partners Sanmina and Wiwynn. At Advancing AI, AMD partners detailed how they deploy AMD AI infrastructure at scale for frontier training and inference: * Anthropic and AMD further outlined Wednesday's strategic partnership announcement to deploy up to 2 gigawatts of AMD Instinct MI455X GPUs in AMD Helios rackscale solutions. The companies are launching a multiyear engineering collaboration to use Claude to accelerate AMD software Specifically, the teams will use Claude to optimize workloads for AMD Instinct GPUs and accelerate ROCm software development. AMD will also broadly adopt Claude across its engineering and product development teams. * OpenAI and AMD are partnering to optimize the full AI stack, from silicon to software. Leveraging OpenAI's Triton framework with AMD ROCm software, the companies are optimizing GPT-class workloads on AMD Instinct MI455X GPUs and AMD Helios OpenAI expects to bring Helios online beginning in the fourth quarter of 2026, with deployments accelerating throughout 2027. * Meta and AMD are co-designing for gigawatt-scale deployments, optimizing AMD's full AI compute stack for Meta workloads. Meta is now validating 6th Gen EPYC CPU platforms in its labs and has begun testing and validating workloads on AMD Helios racks as they prepare to deploy at scale. * Cerebras and AMD are collaborating to deliver a combined solution of Cerebras ultra-low-latency AI compute and AMD Helios high-throughput rack-scale infrastructure to help improve inference efficiency, scalability and economics for ultra-low-latency inference serving. Delivering the Highest Performance Data Center CPUs and GPUs 6th Gen EPYC processors deliver the broadest server CPU portfolio for agentic AI,2 spanning cloud, enterprise, general-purpose and high-performance computing (HPC) workloads. With leadership per-core performance and the highest thread density3, they enable the most agents per watt, per dollar and per rack.4,5,6 For AI host nodes, 6th Gen EPYC CPUs deliver the speed and memory bandwidth to keep accelerators fully fed. And for general-purpose servers, they bring leadership performance and energy efficiency to run business critical applications and AI support tasks. With AMD Instinct™ MI400 Series GPUs, AMD delivers powerful performance for cloud, enterprise and HPC workloads. AMD Instinct MI455X GPUs deliver 34x higher token throughput compared to MI355X GPUs7. For high-precision workloads, the AMD Instinct™ MI430X accelerator is the most advanced for HPC and sovereign AI with up to 288 TFLOPS of hardware-based FP64 performance for scientific computing. Instinct MI430X accelerators are powering the next wave of exascale-class supercomputers across the U.S. and Europe. AMD also launched the Instinct MI350P GPU, bringing seamless AI acceleration to existing infrastructure with leadership token economics. MI350P GPUs deliver up to 4.2x more tokens per second per dollar than the competition8. Advancing the Open Software Ecosystem For developers, AMD ROCm is the open software platform with the performance, flexibility and ecosystem support needed to build and deploy AI on AMD hardware. Building on that foundation, AMD is introducing ROCm.ai, an AI-driven development platform that helps developers build, optimize and deploy GPU software faster across AMD platforms. ROCm.ai brings AI-assisted GPU programming to developers by enabling popular coding agents such as Claude, Codex and Cursor to understand AMD platforms and ROCm natively. ROCm.ai is accelerating software enablement for AMD Instinct MI455X GPUs while optimizing performance. Leading open-source frameworks including PyTorch, Hugging Face, vLLM and SGLang are already enabled on MI455X and seeing great results. Accelerating Next-Generation AI Infrastructure AMD is extending its annual cadence of CPU, GPU, networking and rack-scale innovation through 2030. The company shared new details on its roadmaps, including: * Next-generation EPYC server CPUs based on the "Zen 7" architecture are coming in 2028. The "Florence," "Ferrara" and "Fidenza" CPUs are expected to extend AMD's leadership in density, performance, performance-per-system dollar and performance-per-watt. * "Ravenna" CPUs based on the "Zen 8" architecture are coming in 2030, designed to continue AMD server CPU leadership. * Next-generation AMD Instinct MI500 Series GPUs are coming in 2027, with next-generation compute, memory and interconnect technologies for leadership * AMD Instinct MI600 Series GPUs are coming in * The next-generation AMD Helios 500 rackscale solution will be powered by AMD Instinct MI500 Series GPUs and AMD EPYC "Verano" CPUs, with next-gen Pensando "Como" and "Monza" The AMD Helios 600 rackscale solution will follow, powered by AMD Instinct MI600 Series GPUs, EPYC "Ferrara" CPUs and Pensando "Palma" and "Levanzo" networking. Scaling AI Across Enterprise Leading enterprises run on AMD infrastructure, from cloud, hybrid and on-prem data centers to AI-enabled PC fleets. AMD technologies are helping customers scale quickly and accelerate enterprise transformation. At Advancing AI, AT&T illustrated how it is deploying flexible enterprise AI using AMD technology across cloud, on-premises and air-gapped environments. AT&T is also using AMD Instinct GPUs and ROCm software to power its OTel 2.0 model, an open-source model trained specifically for telecoms. With the AMD Ryzen™ AI Halo developer platform, AMD delivers performance, efficiency and simplicity that makes local AI development accessible. More AMD Ryzen AI Halo platforms, powered by Ryzen™ AI Max PRO 400 Series processors, will be available later this year from AMD and OEM partners. Cisco and AMD are collaborating to combine AMD high-performance inference engines, including AMD Ryzen AI Halo systems, with Cisco networking, observability and security capabilities, so enterprises can deploy, govern and manage hybrid and local agentic AI at scale. Advancing the Next Frontier of Physical AI As AI expands across cloud, enterprise and local systems, the next frontier is bringing intelligence into machines that perceive, reason and act in the physical world. Building on a long legacy in robotics with AMD FPGAs and adaptive SoCs, AMD introduced AMD Kria™ AI solutions, extending the company's robotics capabilities from the robot body to the robot brain. AMD uniquely brings AI perception, reasoning and agentic decision-making and control together on a single platform to deliver the performance required for demanding real-world robotic systems. The portfolio includes new AMD Kria AI system-on-modules (SOMs), powered by the new AMD Ryzen AI Embedded X100 Series processors, and the AMD Kria AI Robotics Developer Platform, the first open, turnkey integrated platform for autonomous robotics combining CPU, GPU, NPU and FPGA compute. Together with an expanded open software ecosystem, AMD Kria AI solutions remove vendor lock-in and help developers and customers accelerate the path from prototype to production for next-generation physical AI systems.

AMD CEO says AI servers in full production, will ship this By Investing.com

Investing.com -- Advanced Micro Devices announced Thursday that its second generation of AI servers designed to compete with Nvidia's products are in full production and will begin shipping at the end of the third quarter. CEO Lisa Su revealed that the Helios servers, which contain AMD's latest MI455X AI accelerator and new Venice central processor, are ready for deployment. Both chips are manufactured by Taiwan Semiconductor Manufacturing Co. "Helios is in full production," Su said during her keynote address in San Francisco. "Customer demand for Helios is extremely strong." Get instant alerts on market-moving headlines on InvestingPro -- now 60% off. OpenAI's top computing executive joined Su on stage to announce that the ChatGPT creator plans to start using the Helios racks later this year. AMD aims to capture market share from Nvidia in the data center chip sector, particularly for inference computing, which processes user queries to chatbots like ChatGPT. Su said AMD estimates the total computing market will reach $2 trillion by 2030, with $1.4 trillion coming from chips that accelerate AI and $220 billion from central processing units. The company valued the total computing market at $365 billion in 2025. "Frankly, the only way you're going to be able to service a market like that is for us to work together as an ecosystem," Su said. "There's no one company that can solve it all." AMD also announced that its new EPYC processor delivers 20% more performance than Nvidia's CPU. Su said Venice is in full production with strong customer demand. The company is collaborating with chip designer Cerebras to provide fast inference through Cerebras Cloud Service. AMD shares traded roughly 3% lower Thursday following the announcements, paring some of earlier losses.

Microsoft to Deploy Next-Gen AMD Instinct and AMD EPYC Processors as the Companies Expand Their Long-Term Strategic Partnership

AMD today announced an expanded strategic partnership spanning AMD GPUs, CPUs, networking and software on Microsoft Azure. At the center of this expansion, Microsoft will deploy the AMD Helios Rackscale Solution, to power frontier model AI inference for Microsoft, its AI customers and support Azure AI services. Azure will also add two new AMD EPYC CPU-powered VM series and broaden its deployment of Pensando DPUs to support Azure networking services. AMD will begin shipping Helios to customers, including Microsoft, in the second half of 2026. AMD Helios combines AMD Instinct™ MI455X GPUs, AMD EPYC™ "Venice" CPUs, Pensando™ networking and ROCm™ software in an open, integrated rackscale platform built for large-scale AI training and inference. The Azure deployment will use Helios for inference workloads spanning frontier models, Azure AI services and customer applications. "AMD and Microsoft have spent years building high-performance infrastructure together, and today we're extending that partnership across the full stack of AMD AI solutions on Azure," said Dr. Lisa Su, Chair and CEO, AMD. "Microsoft's new AMD deployments mark an important milestone as we deliver leadership compute solutions to Azure customers and scale the next generation of AI infrastructure together." "Customers are looking for AI infrastructure that is optimized for a wide range of workloads, from training and inference to data preparation, search, and reinforcement learning," said Satya Nadella, Chairman and CEO, Microsoft. "Through our collaboration with AMD, we are expanding the Azure infrastructure portfolio with AMD Helios to give customers the performance, scale and choice they need to build and run the next generation of AI applications." The collaboration expands access to AMD AI infrastructure across Azure. Frontier model builders can now leverage AMD-powered infrastructure to train and serve large-scale AI models, while enterprise customers can deploy and scale production AI workloads through Azure Foundry Managed Compute. Azure's new VM series, Azure HDv2 for agentic AI and data pipelines, and Azure HXv2 for semiconductor design, will be powered by 6th Gen AMD EPYC "Venice" processors. Together, the new VM series broaden Azure's AMD EPYC portfolio across AI, data and engineering workloads. The collaboration also extends into the networking layer that connects and scales Azure infrastructure. Building on Microsoft's broad deployment of AMD Pensando DPUs, the companies are integrating Azure Boost with AMD technologies to improve networking performance, efficiency and connection processing at cloud scale. As AI demand accelerates, AMD and Microsoft will continue to deliver open, high-performance infrastructure that gives customers flexibility, efficiency and scale to build what's next

AMD Starts Production of Its Helios AI Server and Steps Up the Pace Against Nvidia

Helios integrates the MI455X AI accelerator and the Venice processor, both manufactured by TSMC. According to CEO Lisa Su, demand is "extremely strong," and shipments will begin at the end of the third quarter. OpenAI confirmed it will start deploying the platform before year-end and expects to ramp up in 2027, while also signaling its intention to use the next generation of MI500 chips. AMD estimates the global computing market will reach $2,000bn by 2030, including $1,400bn tied to AI accelerators. The company is also expanding partnerships to speed its push into AI. After a multi-year agreement struck with OpenAI in October, AMD announced a new partnership with Anthropic covering the supply of up to two gigawatts of Instinct MI450 chips starting in 2027, alongside an investment that could reach $5bn. At the San Francisco event, Anthropic said its Claude model was able to configure AMD's servers on its own, highlighting advances by AI agents in complex technical tasks. AMD shares were down nearly 3.5% during trading yesterday.

Microsoft to Deploy AMD's Next-Generation AI Chips on Azure Under Expanded Partnership

Advanced Micro Devices is expanding its strategic partnership with Microsoft, under which the software company is set to deploy AMD's next-generation artificial-intelligence chips and processors across its Azure cloud platform. Microsoft will deploy AMD's Helios rack-scale AI platform to power frontier-model inference workloads for its own AI services and Azure customers, the companies said on Monday. AMD said it expects to begin shipping the Helios platform to customers, including Microsoft, in the second half of 2026. Helios combines AMD's Instinct MI455X graphics processors, sixth-generation EPYC Venice central processing units, Pensando networking technology and ROCm software in an integrated AI computing platform, the companies said. Microsoft said it will also introduce two new Azure virtual-machine series powered by AMD's sixth-generation EPYC processors. The Azure HDv2 series is designed for agentic AI and data pipeline workloads, while the Azure HXv2 series targets semiconductor-design applications. The companies also said they are expanding deployment of AMD Pensando data-processing units across Azure's networking infrastructure and integrating AMD silicon with Microsoft's Azure Boost technology to improve cloud-networking performance. "Microsoft's new AMD deployments mark an important milestone as we deliver leadership compute solutions to Azure customers and scale the next generation of AI infrastructure together," AMD Chief Executive Lisa Su said in a statement. Microsoft Chief Executive Satya Nadella said the expanded partnership will provide customers with additional infrastructure options for AI training, inference and other workloads. Financial terms of the partnership weren't disclosed. Shares of AMD were up 3.5% Monday morning at $513.29. Shares of Microsoft were down around 0.7% at $391.24.

AMD launches Helios and steps up its offensive against Nvidia with Microsoft's backing

Helios is AMD's most ambitious offering yet for data centers dedicated to artificial intelligence. Microsoft will deploy the infrastructure in Azure to run inference for its AI models and will also incorporate AMD's new "Venice" processors into its services. The platform, which brings together graphics processors, central processors, networking gear and internally developed software, already has Meta, OpenAI, Oracle, Tata Consultancy Services and several other major industry players as clients. AMD says Helios lowers the total cost of ownership while improving performance for artificial intelligence workloads. The company says that 8 out of the 10 leading AI-focused firms already use its Instinct accelerators, even as Nvidia retains more than 95% of the data-center graphics processor market thanks to the strength of its CUDA software ecosystem. Several analysts say Helios could nonetheless allow AMD to gain market share quickly if early deployments deliver on its promises. The new platform fits into AMD's turnaround under CEO Lisa Su, following years of investment and strategic acquisitions, including Xilinx, Pensando and ZT Systems. Data centers are now the company's primary growth engine, and as from 2027, AMD expects to generate tens of billions of dollars in annual AI revenue, largely driven by Helios. However, analysts warn that commercial success will depend both on the quality of the technology and AMD's ability to close the software gap with Nvidia.

AMD launches Helios and steps up its offensive against Nvidia with Microsoft backing

Helios is AMD's most ambitious offering yet for data centers dedicated to artificial intelligence. Microsoft will deploy this infrastructure in Azure to run the inference phases of its AI models and will also integrate AMD's new 'Venice' processors into its services. The platform, which brings together graphics processors, central processors, networking equipment and in-house developed software, already counts Meta, OpenAI, Oracle, Tata Consultancy Services and several other major industry players among its customers. AMD says Helios reduces total cost of ownership while improving performance for artificial intelligence workloads. The group says eight of the 10 leading AI-focused companies already use its Instinct accelerators, even as Nvidia retains more than 95% of the data center graphics processor market thanks to the strength of its CUDA software ecosystem. Several analysts say Helios could nonetheless allow AMD to win market share quickly if early deployments deliver on its promises. This new platform is part of the turnaround led by Lisa Su at AMD, following several years of investment and strategic acquisitions, including Xilinx, Pensando and ZT Systems. Data centers are now the group's main growth engine, and it expects to generate, starting in 2027, tens of billions of dollars in annual AI revenue, largely thanks to Helios. Analysts nonetheless believe commercial success will depend as much on the quality of the technology as on AMD's ability to compete with Nvidia's software lead.