13 Sources
[1]
Intel Announces Arc Pro B-Series, "Project Battlematrix" Linux Software Improvements
Intel is using Computex 2025 to showcase their new Arc Pro B-Series graphics cards that will be available in Q3 for professional use-cases as well as focusing on AI inference workstations and edge computing workloads. Plus they are noting some significant improvements coming to their Linux software stack. Linux in fact got quite a few mentions during the advanced press briefing for the Intel Arc Pro B-Series ahead of today's embargo lift at Computex. The Intel Arc Pro B50 is a Battlemage graphics card with 16GB of memory, rated for 170 pTOPS, and has a 70 Watt total board power. Meanwhile the Intel Arc Pro B60 features 24GB of vRAM, rated for 197 pTOPS, and will have a total board power between 120 and 200 Watts. With 24GB of vRAM, the Intel Arc Pro B60 is better suited for AI inference workloads with large language models. The Intel Arc Pro B50 will have a $299 suggested price and the specs are rounded out quite nicely for the price. Both Windows and Linux support is ready. With 16GB of vRAM, it's much better equipped for modern workloads than the Arc Pro A50 that had a mere 6GB of vRAM or the competing NVIDIA RTX A1000 with 8GB. Intel is talking up 1.2x to 33.1x improvements on a generational basis. Interestingly with the Arc Pro B60 is where Intel is talking up "Linux Multi-GPU" support, 24GB of memory as a nice attribute and more than what's available with the current Intel Arc B-Series graphics cards, and also will support SR-IOV. Besides the new Intel Arc Pro B-Series, the other interesting aspect of Intel's Computex announcement is "Project Battlematrix" where they are talking about combining up to eight Intel Arc Pro GPUs for up to 192GB of vRAM to handle 70B+ parameter large language models. With Intel Project Battlematrix it's Linux-focused with an "LLM Optimized Linux Software Stack". The Project Battlematrix Linux Software Stack will still leverage oneAPI, Level Zero, and the like while featuring vLLM Serving and other features. Next quarter is when Intel is hoping their Linux software stack will be ready with vLLM staging and container deployments followed by enhancing performance and improving vLLM serving. Going into Q4'2025 is where they hope to have the SRIOV support ready as well as VDI and manageability feature. Some elements of this Project Battlematrix initiative have already been underway for open-source/Linux support and seemingly now under this umbrella. In past months I've covered the PMT telemetry coming for Battlemage as well as the ongoing SRIOV preparations for the Intel Xe driver and related patch series that are aligning with the Project Battlematrix goals. It will be very interesting to see what all ends up being incorporated into these "Battlematrix" Linux driver stack improvements. This continues building off the Intel Xe kernel driver and other existing open-source driver components maintained by Intel while better equipping their modern Battlemage graphics cards to better handle AI/LLM workloads and more. The Intel Arc Pro B50 and B60 workstation graphics cards are quite interesting and hopefully in Q we'll be able to test them at Phoronix under Linux. It's not until Q4 though where Intel is talking up "full feature" enablement for the Arc Pro B50/B60 graphics cards. It will also be interesting to see what comes of the Project Battlematrix Linux software driver enhancements so stay tuned to Phoronix as always to hear more about their open-source driver enhancements.
[2]
Intel launches $299 Arc Pro B50 with 16GB of memory, 'Project Battlematrix' workstations with 24GB Arc Pro B60 GPUs
Intel has announced its Arc Pro B-series of graphics cards at Computex 2025 in Taipei, Taiwan, with a heavy focus on AI workstation inference performance boosted by segment-leading amounts of VRAM. The Intel Arc Pro B50, a compact card that's designed for graphics workstations, has 16GB of VRAM and will retail for $299, while the larger Intel Arc Pro B60 for AI inference workstations slots in with a copious 24GB of VRAM. While the B60 is designed for powerful 'Project Battlematrix' AI workstations sold as full systems ranging from $5,000 to $10,000, it will carry a roughly $500 per-unit price tag. Intel has focused on leveraging the third-party GPU ecosystem to develop its Arc Pro cards, in contrast to its competitors, who tend to release their own-branded cards for the professional segment. That includes partners like Maxsun, which has developed a dual-GPU card based on the B60 GPU. Other partners include ASRock, Sparkle, GUNNR, Senao, Lanner, and Onix. Both the B50 and B60 GPUs are now being sampled to Intel partners, as evidenced by a robust display of partner cards and full systems on display, and will arrive on the market in the third quarter of 2025. Intel will initially launch the cards with a reduced software featureset, but will add support for features like SRIOV, VDI, and manageability software in the fourth quarter of the year. The Intel Arc Pro B50 has a compact dual-slot design for slim and small-form-factor graphics workstations. It has a 70W total board power (TBP) rating and does not have external power connectors. The GPU wields 16 Xe cores and 128 XMX engines that deliver up to 170 peak TOPS, all fed by 16GB of VRAM that delivers 224 GB/s of memory bandwidth. The card also sports a PCIe 5.0 x8 interface, which Intel credits with speeding transfers from system memory, ultimately delivering 10 to 20% more performance in some scenarios. The B50's 16GB of memory outweighs its primary competitors in this segment, which typically come armed with 6 or 8GB of memory. The card also has certified drivers that Intel claims deliver up to 2.6X more performance than the baseline gaming drivers. Intel shared a slew of benchmarks against the competing Nvidia RTX A1000 8GB and the previous-gen A50 6GB, but as with all vendor-provided benchmarks, take them with a grain of salt (we included the test notes at the end of the article). In graphics workloads, Intel claims up to a 3.4X advantage over its previous-gen A50, and solid gains across the board against the RTX A1000. It sports similar advantages in a spate of AI inference benchmarks. The Intel Arc Pro B60 has 20 Xe cores and 160 XMX engines fed by 24GB of memory that delivers 456 GB/s of bandwidth. The card delivers 197 peak TOPS and fits into a 120 to 200W TBP envelope. This card also comes with a PCIe 5.0 x8 interface. Intel supports multiple B60 GPUs on a single board, as evidenced by Maxsun's GPU, with software support in Linux for splitting workloads across both GPUs (each GPU interfaces with the host on its own bifurcated PCIe 5.0 x8 connection). Intel's benchmarks again highlighted the advantages of the B60's 24GB of memory vs the competing RTX 200 Ada 16GB and RTX 5060Ti 16GB GPUs, claiming this can impart gains of up to 2.7X over the competition in various AI models. Intel also highlighted the advantages of higher memory capacity in model size, context, and concurrency scaling. The Intel Arc Pro B60 will primarily come in pre-built inference workstations ranging from $5,000 to $10,000, dubbed Project Battlematrix. The goal is to combine hardware and software to create one cohesive workstation solution. However, the per-unit cost will be in the range of $500 per GPU, depending on the specific model. Project Battlematrix workstations, powered by Xeon processors, will come with up to eight GPUs, 192GB of total VRAM, and support up to 70B+ parameter models. Intel is working to deliver a validated full-stack containerized Linux solution that includes everything needed to deploy a system, including drivers, libraries, tools, and frameworks, that's all performance optimized, allowing customers to hit the ground running with a simple install process. Intel will roll out the new containers in phases as its initiative matures. Intel also shared a roadmap of the coming major milestones. The company is currently in the enablement phase, with ISV certification and the first container deployments coming in Q3, eventually progressing to SRIOV, VDI, and manageability software deployment in Q4. Intel's partners had multiple Project Battlematrix systems up and running live workloads in the showroom, highlighting that development is already well underway. One demo included a system running the full 675B parameter Deepseek model entirely on a single eight-GPU system, with 256 experts running on the CPU and the most frequently used experts running on the GPU. Other demos included running and finding bugs in code, an open enterprise platform for building RAGs quickly, and a RAG orchestration demo, among others. As noted above, the Intel Arc Pro B50 and Intel Arc Pro B60 will arrive on the market in the third quarter of 2025.
[3]
First Look: Intel Sends Battlemage to Workstations With 24GB Arc Pro B60
For as long as I can remember, I've had love of all things tech, spurred on, in part, by a love of gaming. I began working on computers owned by immediate family members and relatives when I was around 10 years old. I've always sought to learn as much as possible about anything PC, leading to a well-rounded grasp on all things tech today. In my role at PCMag, I greatly enjoy the opportunity to share what I know. TAIPEI -- Intel flexed its professional engineering muscles at Computex and announced its first two Intel Arc Pro B-series graphics cards, the Arc Pro B60 and Arc Pro B50, targeted at the workstation and server markets. We were on hand to check out these new GPUs in person and get the rundown on the AI-centric future of workstation computing. The B60 steals the show here as the more potent of the pair, with 24GB of RAM, ready to power the next wave of AI inferencing workloads. These B60 cards will be produced by various partners, with some special models even featuring an innovative dual-GPU design with two Battlemage GPUs and an enormous 48GB pool of RAM in one card. Though less innovative, the Intel Arc B50 is also an alluring option with 16GB of RAM and a budget price of $299. Intel Arc Pro B60 Puts Battlemage to Work With Intel still relatively new to today's graphics card industry, it's always exciting to see what it will try next. That also extends to the workstation and server markets, where Intel has been a provider for a significantly longer time. It sometimes gets overlooked, but the last time Intel attempted to break into the graphics card industry, it found the product it created was better suited for heavy compute workloads. So, Intel released it as its Xeon Phi product line, which ran from 2010 through 2020. Xeon Phi was unquestionably successful for its time, adopted for several major projects. It even powered the world's fastest supercomputer in 2013, the Tianhe-2. When Intel cancelled the Xeon Phi range, it turned to using Intel Arc graphics architecture to compete in this space instead with the Intel Arc Pro line, updated now with the Intel Arc B60 and the Intel Arc B50. The standard Intel Arc Pro B60 is the workstation counterpart of the Intel Arc B580. Both feature the Intel BMG-G21 graphics chip with 20 Xe-cores based on the Battlemage architecture. The Intel Arc B580 has 12GB of GDDR6 with 456GB/s of memory bandwidth. Intel didn't say what type of RAM the Arc Pro B60 would have, but GDDR6 seems likely, and we expect its memory bandwidth to be the same at 456GB/s. However, the Arc Pro B60 comes with double the amount of RAM, with 24GB, which will help it appeal to workstations. The B60's launch signals Intel's effort to get AI inference-capable workstation hardware into server racks and professionals' hands as soon as possible. Sampling starts now, with delivery targeting edge computing later this year. This launch should provide much more capable and scalable AI proficiency for creators and AI developers, where much of the workstation focus is shifting. We haven't tested one of these cards, but Intel claims that the Intel Arc Pro B60 with 24GB of RAM could outperform an Nvidia RTX 2000 and an Nvidia GeForce RTX 5060 Ti 16GB in some tests. The larger pool of vRAM likely helps the Arc Pro B60 in these tests. Maxsun Intel Arc Pro B60 Dual 48G Turbo Doubles Down on AI Inferencing Capacity Another more interesting option in the Intel Arc Pro line is a custom model made by Maxsun. The Maxsun Intel Arc Pro B60 Dual 48G Turbo combines two Intel Arc Pro B60 graphics chips onto a single card. Each GPU in this configuration operates independently with a PCIe 5.0 x8 connection and 24GB of RAM, giving the card 48GB total. This approach effectively isn't any different from two Intel Arc B60 graphics cards, but workstations are all about cramming as much computational power as possible into as small a space as possible. This approach makes fitting two Arc B60 cards in the same space as you'd fit one a serious advantage. We're sure some of you wonder if we might see a gaming version of this card, but that is unlikely. Supporting two GPUs on one card isn't complicated when those GPUs do compute work, but doing it while gaming is drastically different. It requires a substantial amount of additional driver work to get the best performance out of a setup like this, and even then, you'll find issues, which is why AMD and Nvidia have both largely abandoned their multi-GPU gaming technologies. While not entirely impossible, Intel seems unlikely to move in that direction. The Budget B-Series Pro Card: Intel Arc Pro B50 The Intel Arc Pro B50 is the lesser of the two new workstation graphics cards that Intel announced during Computex. Though this card is likely using the BMG-G21 graphics chip too, it has four of its Xe-cores disabled, which reduces its overall performance. It also has less memory bandwidth and just 16GB of vRAM. The advantages of this card come in a few areas. Intel calls it compact, and while it has a dual-slot thermal solution, the card itself isn't much longer than the PCIe x16 slot it plugs into. Its power draw is similarly low at less than 70 watts, meaning it doesn't require additional power beyond what the PCIe slot provides to operate. Last but not least, it's also affordable at $299. That price tag is higher than you would pay for a standard Intel Arc B580, but you don't get the firmware and extra compute horsepower that comes from a workstation graphics card that way. Competing options like AMD's Radeon Pro W7600 cost double that at full MSRP, even though it has just 8GB of RAM. Intel Arc Pro Starts Sampling Now, Shipping In Fall Intel said it is now sampling both the Intel Arc Pro B60 and the Intel Arc Pro B50 to its customers, with general availability to come sometime in Q3 of this year. The Intel Arc Pro B60 will be available from several of Intel's board partners, including ASRock and Maxsun, though we don't know how much it will cost at this time. The Intel Arc Pro B50 will be an Intel-exclusive GPU product for $299.
[4]
The Arc B770 was a no-show at Computex -- here's what Intel announced instead
Summary Intel is not launching the Arc B770 at Computex; instead, it unveiled the Arc B50 and B60 GPUs for AI workstations. The Arc B50 focuses on traditional workstations, while the B60 targets AI inference workloads with multi-GPU capabilities. Project Battlematrix platform supports up to 8 B60 GPUs, and an optimized Linux software stack for LLMs with over 70 billion parameters. Despite what appeared to be confirmation that Intel would launch the Arc B770 at Computex, the card was nowhere to be found at the show. Teasing its Computex announcements, the official Intel X account responded to several commenters a couple of weeks back, telling them to "stay tuned" in regard to the Arc B770. It looks like they'll need to stay tuned a little longer. Intel confirmed to me that it won't be launching a new consumer graphics card at Computex this year, but it still has some exciting GPU news. Intel instead pulled back the curtain on two new GPUs in the Arc Pro range -- the Arc B50 and B60. Both are workstation cards with a focus on AI, and they leverage the same Battlemage architecture we've seen in action on everything from Lunar Lake laptops to the Arc B580. In fact, the B50 and B60 use the same BMG-G21 die as the Arc B580, though with a massive bump to VRAM capacity to satisfy demanding AI workloads. Related Intel Arc B580 GPU with a massive 48GB of memory could be just around the corner A dual-GPU version of the Arc B580 is in the works Posts 4 Intel wants to make "inference workstations" with its new Arc Pro GPUs Scaling up tiny GPUs Close Intel has two cards joining the Arc Pro range, and you can see the specs Intel laid out for the B50 and B60 below. However, the critical spec here is memory capacity. AI inference is extremely demanding on VRAM, and most consumer-level graphics cards simply don't have enough VRAM to do much useful for local models. It's like the problem with 8GB graphics cards in modern games, but magnified exponentially. The two Arc Pro cards have more VRAM, but Intel is pushing the idea further. You're able to scale up the number of Arc Pro B60 GPUs you include in a workstation. Intel says it designed the software stack in such a way that it's essentially able to split models across multiple GPUs, and it says you can expect linear performance scaling up to eight graphics cards. Technically, you could use more, but Intel says it only officially supports eight. Close Intel is taking this idea to the extreme with Project Battlematrix. This is a workstation platform built specifically for AI developers. It's based on an Intel Xeon platform with support for up to eight B60 GPUs. Intel is also working on a Linux-based software stack for these workstations, which it says is optimized for LLMs in excess of 70 billion parameters. This is still a work in progress, however. Intel has a road map through the end of the year with milestones for optimization. Intel is working with OEMs to bring Project Battlematrix machines to market, and that's the primary way you'll be able to find the Arc Pro B60. Intel's board partners have a lot of leeway in their designs for the B60 -- for instance, Maxon has a dual-GPU model it's releasing -- so there isn't a list price for the card. Close The lower-end Arc B50 doesn't have multi-GPU capabilities, but it targets more traditional workstations. It's a compact, dual-slot card that still uses Intel's Arc Pro drivers, and it can still handle lightweight AI workloads. It's more akin to something like the RTX A1000, however, accelerating other workstation tasks outside of AI. Unlike the B60, Intel is selling the B50 directly to customers, and it clocks in at an MSRP of $300. Intel is already sampling the B50 and B60 to partners, and it plans to have broad availability in Q3. Development will continue past release, however, with Q4 being the target for full feature enablement.
[5]
Intel's New Arc GPUs Aren't Just For Gaming
Summary Intel introduces non-gaming Arc Pro B60 & B50 GPUs with enhanced Xe2 architecture & more memory. Focus on business applications with AI cores & ray tracing units, not really optimized for gaming. "Project Battlematrix" offers a workstation platform supporting up to 8 Arc Pro B60 GPUs for AI processing. While most of us buy graphics cards for gaming, there are graphics cards for other purposes as well. NVIDIA has its Quadro range, and AMD's serious range is sold as the Radeon PRO series. Intel has been in the graphics business for two generations now, and the company is finally branching out that second gen into other non-gaming applications. Intel made a slew of announcements at Computex 2025 today, but the most important one is the addition of new GPUs that aren't just meant for gaming. The company is expanding its Arc Pro family with the new Intel Arc Pro B60 and Intel Arc Pro B50 GPUs. These are built upon the advanced Xe2 architecture that the rest of Intel's Arc B-series GPUs use. The Intel Arc Pro B60 will feature a substantial 24 gigabytes (GB) of memory, while the B50 will offer 16 GB -- still pretty decent, as both feature more VRAM than Intel's current top-of-the-line Intel Arc B580 GPU. As a fun fact, Intel seems to have completely forgotten to announce a B770, so the B580 will continue being the higher-end one for the foreseeable future. Related It's Time to Start Taking Intel GPUs Seriously Thanks to lower prices and better drivers, Intel Arc GPUs are finally worth considering. Posts The Arc Pro B60 and B50 GPUs come equipped with Intel Xe Matrix Extensions (XMX) AI cores and ray tracing units. Technically, these can play games just like regular Arc cards can, but they're meant more for business purposes than anything. Unlike the consumer-focused Arc gaming cards, the Pro series prioritizes stability, reliability, and performance in professional applications. These go into workstations, servers, and are meant for "serious" work. They can play games, but they're not as optimized for gaming as the consumer-grade Arc cards are, and they're probably a waste of money if you're not buying them for business purposes. Still, it's good to see Intel put effort into its business-grade GPUs. Intel also revealed "Project Battlematrix," a configurable workstation-class platform based on Intel Xeon processors. This system can support up to a whopping eight Intel Arc Pro B60 24GB GPUs, mostly for AI processing purposes, which seems to be something these cards are heavily optimized for. "The Intel Arc Pro B-Series showcases Intel's commitment in GPU technology and ecosystem partnerships," stated Vivian Lien, vice president and general manager of Client Graphics at Intel. She went on to further add that "with Xe2 architecture's advanced capabilities and a growing software ecosystem, the new Arc Pro GPUs deliver accessibility and scalability to small and medium-sized businesses that have been looking for targeted solutions." These cards will become available from multiple third-party manufacturers starting next month. No one should buy these for a gaming PC, and you might want to wait on the rest of Intel's B-series consumer-grade Arc GPUs, or just buy whatever's currently on store shelves. Source: Intel
[6]
Intel Announces Arc Pro B-Series and Gaudi 3 AI Accelerators
At Computex 2025, Intel introduced two new workstation-grade graphics accelerators -- Arc Pro B60 and Arc Pro B50 -- built on the Xe2 architecture. Each card integrates Intel Xe Matrix Extensions (XMX) AI cores and hardware-accelerated ray tracing units. The B60 model offers 24 GB of dedicated frame buffer memory, while the B50 model provides 16 GB. Both cards support multi-GPU configurations and deliver deterministic performance for architecture, engineering, and construction (AEC) workflows. Certified by key independent software vendors (ISVs) and compatible with both Windows and Linux ecosystems, they leverage containerized drivers on Linux to streamline AI inference and visualization tasks. A reference platform codenamed "Project Battlematrix" complements these GPUs by accommodating up to eight B60 cards, enabling systems with as much as 192 GB of video memory for medium-sized AI models (up to 150 billion parameters). Intel also expanded its Gaudi 3 line of AI accelerators with two deployment formats: PCIe add-in cards and rack-scale server modules. The PCIe option plugs into existing x86 servers, supporting models from smaller Llama 3.1-8B up to full-scale Llama 4 Scout and Maverick configurations. PCIe cards will ship in the second half of 2025. Rack-scale reference designs accommodate up to 64 Gaudi 3 devices per rack, each rack offering more than 8 TB of high-bandwidth memory. These rack-scale systems employ cabled backplanes, blind-mate 2D cabling, and liquid cooling to optimize maintenance and thermal management. This open design aligns with Open Compute Project specifications, enabling cloud service providers to avoid vendor lock-in while meeting low-latency inferencing requirements. Finally, Intel's AI Assistant Builder framework, now available in public beta on GitHub, allows developers to construct and deploy specialized AI agents on Intel-based systems. This lightweight, modular toolkit supports on-device model inference and customization workflows, reducing dependency on external services. Intel highlighted early integrations by OEM partners such as Acer and ASUS, which demonstrated end-user applications ranging from natural language query interfaces to domain-specific automation tools. The release coincides with Intel's 40th anniversary in Taiwan, underlining the company's long-term commitment to the region's technology ecosystem and reinforcing the importance of open architectures in accelerating professional and enterprise AI adoption.
[7]
Intel Arc Pro B50 16GB and B60 24GB GPUs announced, no sign of Arc B770 for gamers
As an Amazon Associate, we earn from qualifying purchases. TweakTown may also earn commissions from other affiliate partners at no extra cost to you. At Computex 2025, Intel announced its new Battlemage Arc Pro GPUs for AI workstations, with multiple models and variants from its partners, such as ASRock, Sapphire, MaxSun, and more. The Arc Pro B-series, powered by the same Battlemage architecture as the desktop Intel Arc B580 gaming GPU, includes two models - the Intel Arc Pro B50 16GB, starting from $299, and the Intel Arc Pro B60 24GB for around $500 per GPU. That's per GPU because the Arc Pro B60 is designed for systems that can house multiple B60 GPUs in scalable AI inference workstations dubbed 'Project Battlematrix.' We have to see these systems up and running at Computex 2025, where up to 8 Intel Arc Pro GPUs offer up to 192GB of VRAM to support 70B+ parameter models with full software support and optimization. According to the performance chart below, the new Intel Arc Pro B50 is set to offer an efficient mainstream option for AI workstations compared to the NVIDIA RTX A1000. The Intel Arc Pro B60 is being positioned as a better option than the NVIDIA RTX 2000 Ada 16GB and GeForce RTX 5060 Ti 16GB GPUs due to its inclusion of 24GB of memory. Here's a look at the specs for the new individual Intel Arc Pro B50 and B60 GPUs. Both are set to arrive with consumer and pro drivers, ISV Certified SRIOV, and manageability features. And for those wondering about the heavily rumored Intel Arc B770 for desktop gamers, unfortunately, the flagship Battlemage gaming GPU was not at Computex. "The Intel Arc Pro B-Series showcases Intel's commitment in GPU technology and ecosystem partnerships," said Vivian Lien, vice president and general manager of Client Graphics at Intel. "With Xe2 architecture's advanced capabilities and a growing software ecosystem, the new Arc Pro GPUs deliver accessibility and scalability to small and medium-sized businesses that have been looking for targeted solutions." The Intel Arc Pro B60 is set to arrive first, in June 2025, followed by the Intel Arc Pro B50 in July 2025.
[8]
Intel Launches Arc Pro B-Series GPUs for AI Workloads at Computex 2025
Intel also revealed Project Battlematrix, a Xeon platform for AI models Intel Arc Pro B-series graphics processing units (GPUs) were launched by the company at Computex 2025 on Tuesday. Available in various partner board form factors, the new GPUs are built for workstations and artificial intelligence (AI) inference, handling demanding workloads with larger memory configurations and expanded software support. The Arc Pro B-series comprises two GPU models called Intel Arc B50 and Arc B60, with both of them equipped with advanced ray tracing units and Intel Xe Matrix Extensions (XMX) cores. Intel Arc Pro B50 GPU will be available for purchase via Intel-authorised resellers starting in July 2025, priced at $299 (roughly Rs. 25,500). Meanwhile, the Intel Arc B60 GPU can be purchased from add-in board partners such as ASRock, Gunnir, Lanner, Maxsun, Onix, Senao and Sparkle starting June. Its pricing is yet to be revealed. Intel says while its Arc B-series GPUs targeted consumer-grade PCs, the Arc Pro B-series GPUs are built for graphics workstations, AI inference workstations, and edge computing solutions. They are based on the company's latest Xe2 architecture and leverage XMX engines for supporting AI capabilities. As per Intel, the GPUs deliver up to 3.4 times better graphical fidelity and up to 3.5 times higher inference in Stable Diffusion compared to the Nvidia RTX A1000 8GB. The company says Arc Pro B-series GPUs have been optimised for delivering stability through independent software vendor (ISV) certifications and optimised software on AEC (architecture, engineering, construction) and inference workstations. When equipped in Linux-based systems, they support a containerised software stack for simplification of AI deployments. Intel's Arc Pro B60 GPU is equipped with 20 second-generation Xe cores, promising up to 197 tera operations per second (TOPS) of computational power and up to 200W Total Board Power (TBP). According to the company, it is equipped with 24GB of GDDR6 VRAM with a memory bandwidth of 456GB/s and supports PCIe Gen5 platform. It supports a varied range of display resolutions, including but not limited to 8K UHD at 60Hz, 5K WUHD at 240Hz, and 5K UHD at 120Hz. With this GPU, Intel is targeting advanced AI inferencing workstations and servers. Meanwhile, the Intel Arc Pro B50 GPU is claimed to be tailored for use cases involving AI, generative design, 3D simulations, ray tracing, and editing. It gets 16 Xe cores, 16GB of GDDR6 memory, and a memory bandwidth of 224GB/s. Its peak computational power is rated at 170 TOPS with a 70W TBP. Alongside, Intel has also unveiled a configurable workstation-class Intel Xeon-based platform, codenamed Project Battlematrix. It is said to be designed to reduce friction points during AI development. As per the company, it supports up to eight Intel Arc Pro B60 24GB GPUs for enabling medium-sized AI models with up to 150 billion parameters, along with up to 192GB of VRAM.
[9]
Intel Unveils Arc Pro B60 24 GB & B50 16 GB "Battlemage" GPUs For Pro & AI Workloads: Better Perf/$ & Up To 3x Faster, Dual GPU Variant Teased Too
Today, Intel is officially announcing its Arc Pro B60 & B50 "Battlemage" GPUs, offering superb VRAM capacities for AI & Professional users at a great value. At Computex 2025, Intel is offering a big graphics update for its Arc lineup. The update is that the Blue Team is expanding into new horizons with the Arc Battlemage family, with two brand-new products, the Arc Pro B60 and the Arc Pro B50. Both of the products are aimed at graphics workstations, inference workloads, and Edge Computing segments. The two key products within the lineup are the Arc Pro B60 and the Arc Pro B50. So let's take a look at the specifications on offer. First up, we have the higher-end Intel Arc Pro B60, which features the full BMG-G21 GPU die with 20 2nd Gen Xe cores, 160 XMX engines, and a peak compute output of 197 TOPS (INT8). This solution scales from a TBP of 120W up to 200W and uses a Gen5 x8 PCIe link. The peak TOPS are around 15.5% lower than the Arc B580, which is a more gaming-tuned graphics card and uses a higher TBP (190W). In terms of memory, the Intel Arc Pro B60 is equipped with 24 GB of VRAM across a 192-bit bus interface, which offers 456 GB/s of total bandwidth. These are the same 19 Gbps GDDR6 dies, but the capacity has been doubled over the Arc B580. For performance, Intel is putting the Arc Pro B60 against the RTX 5060 Ti 16 GB and the RTX 2000 Ada 16 GB. The Arc Pro B60 can be seen offering up to 2.7x performance in various AI models & it gets better once larger LLMs are tested, as the 16 GB VRAM is not enough to accommodate those. But there's more: Intel isn't stopping itself from going big, and to address the memory-scaling problem with larger AI models, the company is introducing Project Battlematrix, an inference workstation platform that enables the use of up to 8 Intel Arc Pro AI GPUs at once. For this purpose, select Intel partners (such as Maxsun) will be offering an Intel Arc Pro B60 solution with dual GPUs. These are not connected via a PLX chip but act as a separate GPU, each with its own 24 GB of VRAM for a total of 48 GB of VRAM per card. This enables a total of 8 BMG-G21 GPUs, 192 GB of VRAM, and 1280 XMX engines to power model sizes over 70 B. The platform is optimized around the use of PCIe Gen5 protocols on the Xeon ecosystem. Project Battlematrix will come with an LLM Optimized Linux software stack and feature full-stack validation, making it a potent solution for AI use cases that offers a lot of value to end users. The shining diamond within the Arc Pro B-series family is the B50. This is a cost-effective solution that packs 16 Xe cores, 128 XMX Engines, and 170 Peak TOPS (INT8). The graphics card has a TBP of 70W and utilizes a PCIe Gen5 x8 interface. VRAM includes 16 GB capacity running across a 128-bit bus interface and 224.0 GB/s of total bandwidth. The Intel Arc Pro B50 is designed to be energy efficient; as such, it uses a dual-slot and compact form factor. The card is both Linux and Windows Ready and will come with consumer and Pro drivers (ISV Certified). The reason why the Arc Pro B50 is the first GPU to be highlighted by Intel is due to its incredible value and VRAM offering. The GPU will be priced at an MSRP of $299, which makes it vastly better than the rest of the competition while offering up to 3x the uplift versus the past-generation A50. Intel shares various benchmarks of the Arc Pro B50 against the A50 and the NVIDIA RTX A1000, both of which get destroyed by the Arc Pro B50 in a range of graphics and inference workloads. With that said, Intel also shared word on its software feature roadmap for the Arc Pro lineup. In Q2, the company plans on offering the baseline Windows & Linux drivers to its partners along with Workstation ISV certifications. This will be followed up with the deployment of the first inference-optimized containers along with further improvements to the GPUs, and finally, in Q4, the company will add Virtualization updates such as SRIOV, VDI, and Manageability. The Intel Arc Pro B60 and Arc Pro B50 will be available in Q3 of this year, with customer sampling starting now. The cards will be shipped within systems from leading workstation manufacturers, but we were also told that a DIY launch might happen after the software optimization work is complete around Q4. As for price, the Arc Pro B50 will have an MSRP of $299 US, while the Arc Pro B60 will target the $500 US price point with systems around the $5000 US price range.
[10]
Intel's Next-Gen AI Tech Steals The Show At Computex 2025 - Intel (NASDAQ:INTC)
Intel Corp INTC showcased a new lineup of graphics processing units (GPUs) and AI accelerators designed for professionals and developers at the Computex 2025 on Monday. Intel launched Arc Pro B60 and B50 graphics processing units for workstations and AI inference. During the May 20-23 event in Taipei, Taiwan, Intel marks 40 years of collaboration with local ecosystem partners and expands its GPU lineup, AI accelerator capabilities, and AI assistant availability. Also Read: Intel CFO Says New Chip Output For Clients Will Be Minimal, Eyes Foundry Breakeven By 2027 Intel announces Intel Arc Pro B60 and Intel Arc Pro B50 GPUs, expanding the Intel Arc Pro family with configurations tailored for AI inference and professional workstations. Intel Gaudi 3 AI accelerators are now available in PCIe and rack-scale systems, offering scalable, open solutions for enterprise and cloud AI inferencing. Now publicly available on GitHub, Intel AI Assistant Builder enables developers to create local, purpose-built AI agents optimized for Intel platforms. This week's product launches coincide with Intel's 40th anniversary in Taiwan, which marks four decades of collaboration. New Intel Arc Pro B60 and B50 GPUs based on the Xe2 architecture feature Intel Xe Matrix Extensions (XMX) AI cores and advanced ray tracing units, bringing high-performance capabilities to creators, developers, and engineers. Expanding Intel's GPU lineup for professionals, new Intel Arc Pro B60 and Intel Arc Pro B50 GPUs are designed for demanding AI inference workloads and workstation applications. The GPUs are optimized for AEC (architecture, engineering, construction) and inference workstations, offering stability and performance through various independent software vendor (ISV) certifications and optimized software. The Intel Arc Pro B-Series GPUs are compatible with consumer and pro drivers on Windows. On Linux, the GPUs support a containerized software stack to simplify AI deployments and will be progressively upgraded with features and optimizations. The Intel Arc Pro B-Series offers creators and AI developers scalable, cost-effective solutions by combining high memory capacity with key software compatibility. Intel also revealed a configurable workstation-class Intel Xeon-based platform (code-named Project Battlematrix) designed to reduce friction points for AI developers. It supports up to eight Intel Arc Pro B60 24GB GPUs to enable medium-size (up to 150 billion parameters) and accurate AI models with up to 192GB of video random-access memory. Intel chief Lip-Bu Tan told Reuters on Monday that the company has a 55% share of the data center market. Recently, Intel announced that several contract manufacturing customers planned to build test chips for its forthcoming advanced manufacturing process as it struggles to win market share from Taiwan Semiconductor Manufacturing Co TSM. Price Actions: INTC stock is down 1.45% at $21.34 at last check Monday. Read Next: AMD Launches EPYC 4005 Chips For Small Business And Cloud Photo: Tada Images/ Shutterstock.com INTCIntel Corp$21.36-1.36%Stock Score Locked: Want to See it? Benzinga Rankings give you vital metrics on any stock - anytime. Reveal Full ScoreEdge RankingsMomentum21.70Growth9.57Quality-Value73.30Price TrendShortMediumLongOverviewTSMTaiwan Semiconductor Manufacturing Co Ltd$191.75-1.27%Market News and Data brought to you by Benzinga APIs
[11]
Intel Unleashes AI Power: Arc Pro GPUs Take Center Stage at Computex 2025!
Intel AI Assistant Builder goes public in beta, allowing developers to create and run custom local AI agents on Intel-based systems Computex 2025 was an absolute sight to behold as Intel took the spotlight, armed not just with ambition but with serious silicon strength. With calculated precision, the tech giant unveiled its latest weapons for the professional battlefield: the Arc Pro B60 and B50 GPUs.These aren't consumer-grade cards chasing frame rates. They are purpose-built for professionals who live deep in CAD workflows, media production pipelines, and AI model training grounds. The Arc Pro B60 and B50 deliver ECC memory support, certification for major ISVs, and the kind of horsepower engineers, architects, and data scientists demand.
[12]
Computex 2025: Intel Unveils New GPUs for AI and Workstations By Investing.com
Company extends the Intel Arc Pro GPU lineup to prosumers and AI developers, and announces Intel Gaudi 3 AI accelerator availability via rack scale and PCIe deployments. TAIPEI, Taiwan--(BUSINESS WIRE)--Today at Computex 2025, Intel unveils a new lineup of graphics processing units (GPUs) and AI accelerators designed for professionals and developers. Intel's announcements include: This press release features multimedia. View the full release here: https://www.businesswire.com/news/home/20250519565090/en/ Celebrating 40 Years in Taiwan This week's product launches coincide with the 40th anniversary of Intel in Taiwan " marking four decades of collaboration, innovation and shared success around the x86 architecture with one of the world's most dynamic technology ecosystems. As high-performance AI applications increase the need for more compute, the x86 architecture continues to build on its key role accelerating customer and user innovation. More: Intel at Computex 2025 (Press Kit) For the past 40 years, the power of our partnership with the Taiwan ecosystem has fueled innovation that has changed our world for the better, said Lip-Bu Tan, CEO of Intel. This week, we are renewing our commitment to our partners as we work to build a new Intel for the future. Together, we will create great products that delight our customers and capitalize on the exciting opportunities ahead. New GPUs for AI and Workstation Performance New Intel Arc Pro B60 and B50 GPUs " based on the Xe2 architecture " feature Intel ® Xe Matrix Extensions (XMX) AI cores and advanced ray tracing units, bringing high-performance capabilities to creators, developers and engineers. Expanding Intel's GPU lineup for professionals, new Intel Arc Pro B60 and Intel Arc Pro B50 GPUs are designed for demanding AI inference workloads and workstation applications. With AI-ready features, 24 gigabytes (GB) and 16GB of memory, respectively, and multi-GPU scalability, the Intel Arc Pro B-Series offers a versatile solution for creators, AI developers and professionals. The GPUs are optimized for AEC (architecture, engineering, construction) and inference workstations, offering stability and performance through a wide range of independent software vendor (ISV) certifications and optimized software. The Intel Arc Pro B-Series GPUs are compatible with consumer and pro drivers on Windows. On Linux, the GPUs support a containerized software stack to simplify AI deployments, and will be progressively upgraded with features and optimizations. By combining high memory capacity with key software compatibility, the Intel Arc Pro B-Series offers creators and AI developers scalable, cost-effective solutions. Intel also revealed a configurable workstation-class Intel ® Xeon ®-based platform (code-named Project Battlematrix) that is designed to reduce AI developer friction points. It supports up to eight Intel Arc Pro B60 24GB GPUs to enable medium-size (up to 150 billion parameters) and accurate AI models with up to 192GB of video random-access memory. The Intel Arc Pro B-Series showcases Intel's commitment in GPU technology and ecosystem partnerships, said Vivian Lien, vice president and general manager of Client Graphics at Intel. With Xe2 architecture's advanced capabilities and a growing software ecosystem, the new Arc Pro GPUs deliver accessibility and scalability to small and medium-sized businesses that have been looking for targeted solutions. The Intel Arc Pro B60 GPU will be sampling from add-in board partners including but not limited to ASRock, Gunnir, Lanner, Maxsun, Onix, Senao and Sparkle starting in June 2025. The Intel Arc Pro B50 GPU will be available from Intel-authorized resellers starting in July 2025. Intel Gaudi 3 Supports PCIe and Rack Scale Deployments Expanding on its AI strategy, Intel introduced new deployment options for Intel Gaudi 3 AI accelerators: The Intel Gaudi rack scale architecture is optimized for running large AI models and excels in real-time inferencing with low-latency performance. These configurations reinforce Intel's commitment to open, flexible and secure AI infrastructure, supporting both custom and Open Compute Project (OCP) designs for cloud service providers (CSPs). Intel AI Assistant Builder: Now Available Following its debut at CES 2025, Intel AI Assistant Builder " a lightweight, open software framework for building and running custom AI agents locally on Intel-based AI PCs " is now available in public beta release on GitHub. With new innovative solutions being unveiled at Computex by Acer and ASUS as examples, Intel AI Assistant Builder allows developers and partners to quickly create and deploy AI agents for their own organizations and direct customers. Intel Booth at Computex 2025 Intel will showcase its latest processor and graphics innovations at Computex 2025 in the Taipei Nangang Exhibition Center. Intel (Nasdaq: INTC) is an industry leader, creating world-changing technology that enables global progress and enriches lives. Inspired by Moore's Law, we continuously work to advance the design and manufacturing of semiconductors to help address our customers' greatest challenges. By embedding intelligence in the cloud, network, edge and every kind of computing device, we unleash the potential of data to transform business and society for the better. To learn more about Intel's innovations, go to newsroom.intel.com and intel.com. Notices & Disclaimers Performance varies by use, configuration and other factors. Learn more at intel.com/performanceindex. AI features may require software purchase, subscription or enablement by a software or platform provider, or may have specific configuration or compatibility requirements. Data latency, cost and privacy advantages refer to non-cloud-based AI apps. Learn more at intel.com/AIPC. © Intel Corporation. Intel, the Intel logo and other Intel marks are trademarks of Intel Corporation or its subsidiaries. Other names and brands may be claimed as the property of others. View source version on businesswire.com: https://www.businesswire.com/news/home/20250519565090/en/
[13]
Intel Launches GPUs to Expand Arc Family
Intel launched a new group of graphics processing units and expanded the availability of some artificial-intelligence accelerators. The software company on Monday said it was debuting Arc Pro B60 and B50 GPUs for workstations and AI inference. The GPUs are designed for architecture, engineering, construction and inference workstations, Intel said. They add larger memory configurations and greater software support to Intel's Arc Pro family, a set of graphics software geared toward gaming. Intel also announced three new deployment options for its Gaudi 3 AI accelerator. They are now available in PCIe and rack scale systems. Intel AI Assistant Builder is now offered on GitHub to allow customers to create personalized AI agents. The announcements come ahead of Computex, a technology showcase in Taiwan. Write to Katherine Hamilton at [email protected]
Share
Copy Link
Intel announces new Arc Pro B-Series GPUs, including the B50 and B60, designed for professional workstations and AI inference. The company also introduces 'Project Battlematrix', a platform supporting multiple GPUs for large language model processing.
Intel has unveiled its latest addition to the Arc Pro family, the B-Series GPUs, at Computex 2025. The new lineup includes the Arc Pro B50 and B60, designed specifically for professional workstations and AI inference tasks 12.
The Arc Pro B60 is the more powerful of the two, featuring 24GB of VRAM and rated for 197 pTOPS (peta operations per second). It has a total board power between 120 and 200 Watts 1. The B60 is particularly well-suited for AI inference workloads with large language models, thanks to its substantial memory capacity 2.
The Arc Pro B50, on the other hand, comes with 16GB of memory, is rated for 170 pTOPS, and has a 70 Watt total board power 1. Priced at $299, the B50 is positioned as a more budget-friendly option for traditional workstations 3.
Both GPUs are built on Intel's Battlemage architecture, utilizing the BMG-G21 graphics chip. They feature Intel Xe Matrix Extensions (XMX) AI cores and ray tracing units, prioritizing stability, reliability, and performance in professional applications 5.
Alongside the new GPUs, Intel introduced "Project Battlematrix," a workstation platform designed to support up to eight Arc Pro B60 GPUs 4. This configuration aims to handle large language models with over 70 billion parameters, showcasing Intel's focus on AI processing capabilities 2.
Intel is developing an optimized Linux software stack for Project Battlematrix, which will include features like vLLM serving and container deployments. The company plans to roll out these enhancements in phases, with full feature enablement expected by Q4 2025 1.
The Arc Pro B-Series GPUs are not primarily intended for gaming, despite their capability to run games. Instead, they target business and professional applications, competing with NVIDIA's Quadro and AMD's Radeon PRO series 5.
Intel is working with various partners, including Maxsun, ASRock, Sparkle, and others, to bring these GPUs to market 2. The B50 and B60 are currently being sampled to partners, with broad availability expected in Q3 2025 4.
Intel's entry into the professional GPU market with AI-focused hardware signifies the company's commitment to competing in the rapidly growing AI sector. The high VRAM capacity of these GPUs, particularly the B60 with 24GB, addresses the increasing memory demands of AI inference tasks 23.
The introduction of Project Battlematrix and the ability to scale up to eight GPUs in a single system demonstrate Intel's strategy to provide powerful, flexible solutions for AI developers and researchers 4.
As the AI industry continues to evolve, Intel's new offerings position the company to capture a share of the professional GPU market, challenging established players like NVIDIA and AMD in the workstation and AI inference space 5.
Google launches its new Pixel 10 smartphone series, showcasing advanced AI capabilities powered by Gemini, aiming to challenge competitors in the premium handset market.
20 Sources
Technology
7 hrs ago
20 Sources
Technology
7 hrs ago
Google's Pixel 10 series introduces groundbreaking AI features, including Magic Cue, Camera Coach, and Voice Translate, powered by the new Tensor G5 chip and Gemini Nano model.
12 Sources
Technology
7 hrs ago
12 Sources
Technology
7 hrs ago
NASA and IBM have developed Surya, an open-source AI model that can predict solar flares and space weather with improved accuracy, potentially helping to protect Earth's infrastructure from solar storm damage.
6 Sources
Technology
15 hrs ago
6 Sources
Technology
15 hrs ago
Google's latest smartwatch, the Pixel Watch 4, introduces significant upgrades including a curved display, enhanced AI features, and improved health tracking capabilities.
17 Sources
Technology
7 hrs ago
17 Sources
Technology
7 hrs ago
FieldAI, a robotics startup, has raised $405 million to develop "foundational embodied AI models" for various robot types. The company's innovative approach integrates physics principles into AI, enabling safer and more adaptable robot operations across diverse environments.
7 Sources
Technology
7 hrs ago
7 Sources
Technology
7 hrs ago