NVIDIA Vera CPU with Olympus cores delivers unprecedented ARM performance against Intel and AMD

NVIDIA Vera CPU Benchmarks: Olympus Cores Delivering The Best Performance Ever Seen On ARM Review

NVIDIA's Vera data center CPU isn't ramping up until later this year but I recently had the opportunity to try out this new ARM-based CPU designed for agentic AI workloads. NVIDIA's Vera CPU with its in-house-designed Olympus CPU cores ends up packing a heavy-hitting punch with competitiveness to Intel/AMD x86_64 CPUs that I have never seen out of any other ARM or non-x86_64 processors. Continue on with these early benchmarks of the NVIDIA Vera CPU on Linux. Vera is NVIDIA's next-gen data center CPU designed for agentic AI and similar modern data center workloads. Vera most notably will be found with NVIDIA NVL72 Vera Rubin as the host CPU for powering these powerful racks for AI while it will also be found standalone for CPU racks. Unlike NVIDIA's Grace that uses Arm Neoverse-V2 cores, Vera makes use of NVIDIA's in-house "Olympus" core design. Vera features 88 Olympus cores that claim to deliver 2x the performance of its predecessor with leading energy efficiency too. Olympus is compatible with the Armv9.2 ISA and supports FP8 precision, 176 threads in total via spatial multi-threading, and is paired with LPDDR5X memory for delivering up to 1.2TB/s of memory bandwidth. Over Grace, Vera also has double the L2 cache at 2MB per core, a larger unified L3 cache at 164MB, and supports PCIe Gen 6 as well as CXL 3.1 connectivity. The Vera CPU as tested for this initial benchmarking had a 450 Watt socket TDP. With the LPDDR5X memory it's around 50 Watts or less of power consumption. NVIDIA Vera data center CPUs remain on track for shipping in the second half of the year, but ahead of the ramp, NVIDIA invited me to their Santa Clara headquarters to run some of the first public benchmarks of this new CPU with their Olympus cores. In this article are those very initial results. But before talking performance, it's first important to note the level of Linux support. With Vera not officially out there yet I really didn't know what to expect for the upstream Linux kernel support and the like or what their plans are among the major ARM64 Linux distributions, etc. Fortunately, NVIDIA Vera is already in good shape with upstream Linux kernel support. Linux 7.1+ has the key driver support in place and Vera should work on ARM64 server Linux distributions like Ubuntu, Fedora, etc. NVIDIA will also continue to provide Base OS as their modified version of Ubuntu with all the Vera patches ready. ACPI is relied upon and no needing to deal with frustrating Device Trees or other headaches for NVIDIA Vera on Linux. With Vera being compatible with Armv9.2 and complying with Arm's Server Base System Architecture (SBSA), it ends up making use of many of the common ARM Linux drivers for support, as part of the reason it's more difficult pre-launch to track the upstream kernel support status in advance of hardware availability. Among that common ARM Linux code being used is the ongoing work around Arm Confidential Compute (CCA) for confidential computing with VMs to be supported by Vera. Last year GCC and LLVM Clang added support for the Olympus cores. That means GCC 16.1+ or LLVM Clang 21+ for building optimized binaries for Vera. It was great and ambitious of NVIDIA introducing this Olympus compiler support so early and they can be applauded for that initiative. For comparison, while NVIDIA upstreamed their Olympus support in March of 2025, it was only in December when the AMD Zen 6 (znver6) support was added to GCC and this February when it appeared in LLVM/Clang. NVIDIA's Olympus compiler upstreaming is akin to the well-in-advance support we have been used to seeing from Intel over the years. Great seeing NVIDIA take similar initiative with their Vera upstreaming for compilers and at large. Hopefully this continues with future NVIDIA CPU generations. I wasn't able to test loading different Linux distributions on NVIDIA Vera or the like, but from everything told while at NVIDIA, the upstream open-source Linux support is in good shape already for Vera. My testing was done on Ubuntu 24.04 LTS with their base OS configuration of a patched Linux 6.18 LTS kernel paired with GCC 16.1. The NVIDIA Vera with its Olympus cores were working nicely on Linux. One caveat is some power management tuning still in the process of being upstreamed. Recently I covered on NVIDIA working on ACPI CPPC v4 support for Linux and this appears to be one of the related bits. Due to that power management tuning still happening, unfortunately, NVIDIA asked that CPU power consumption monitoring not be engaged during this initial round of benchmarking. Similarly, CPU frequency monitoring was not allowed either during this first round of testing. Additionally, the NVIDIA Vera benchmarking was done from one of their pre-production, open-platform system where as it will be more relevant for looking at power and frequency in an actual production, enclosed server chassis designs later in 2026. NVIDIA also requested only specific workloads relevant to the intended workloads/domains that Vera is catering to in the data center be tested. So this first round of Vera benchmarking isn't too comprehensive across the spectrum of possible workloads but limited to the benchmarks that were permitted based on what they feel were most relevant -- plus the fact I was only spending one day at NVIDIA's offices. For these initial NVIDIA Vera benchmarks they preferred the scope of benchmarks be limited to target use-cases they feel most relevant for their modern data center customers. This isn't a sponsored article but I obliged to their requests in order to run these initial Vera CPU benchmarks. Hopefully in future rounds of Vera testing over the months ahead there will be a more widespread set of benchmarks for those curious about the Olympus CPU core performance at large. Similarly, in being able to report on the power efficiency and performance-per-Watt once their power management code is tuned.

NVIDIA names Anthropic and OpenAI among first users of its Vera chip

The company's in-house processor, built for AI agents and pitched as the successor to Grace, is now in full production. Jensen Huang spent a portion of his Computex keynote reading out a guest list. Anthropic, OpenAI, SpaceX and Oracle, the Nvidia chief executive told the audience in Taipei on Monday, are among the first big users of Vera, the company's new in-house processor. Nvidia, a firm that built its empire on graphics chips, would now like to be known for a CPU as well. Vera is the successor to Grace, Nvidia's previous data-centre processor, though the company is positioning it as a ground-up redesign rather than an iteration. It is built around 88 of Nvidia's own "Olympus" cores, a departure from the off-the-shelf Arm Neoverse cores that powered Grace, and the company says it is now in full production. The pitch is that Vera is a CPU designed for the age of AI agents, software that plans and executes tasks rather than simply answering a prompt. Nvidia claims the chip completes those agentic workloads faster than the x86 processors made by Intel and AMD, and pairs the cores with up to 1.2TB/s of memory bandwidth. Early independent benchmarks put Vera ahead of Intel's Xeon and AMD's EPYC on several measures, which is the comparison Nvidia wants drawn. The named customers matter as much as the silicon. Anthropic and OpenAI are the two labs whose compute appetite has come to define the current build-out, and Nvidia listing them as launch users is a statement about where the chip is headed. The first units were hand-delivered in May, according to Nvidia's own account, before the public announcement in Taipei. For Nvidia, the CPU is a flank manoeuvre. Its GPUs already sit at the centre of nearly every large AI system, and the data-centre processor has long been the one major component it bought rather than built. Designing its own removes a dependency and lets the company sell the CPU and GPU as a matched pair, which is the logic behind the Vera Rubin platform Huang has called the largest product launch in the island's history. Huang has spent the year describing Taiwan as the "epicentre" of that effort. Oracle Cloud Infrastructure is set to be the first hyperscaler to deploy Vera at scale, with broader availability across the other major clouds expected in the second half of 2026. That timeline is the one figure worth watching. Full production and a podium full of marquee names are easy to announce in June; volume shipments to paying customers are the harder thing, and they are still ahead. What Nvidia did not say in Taipei was pricing, or how many units any of the named labs have actually committed to. The guest list was read out. The invoices were not.

NVIDIA Vera CPU Is 'Packing a Heavy-Hitting Punch' Against Competition

In the new Phoronix benchmark, Vera delivers winning performance and memory results for agentic AI. The shift to agentic AI creates a new CPU requirement for the AI factory: fast cores, massive memory bandwidth and the ability to sustain high performance when all cores are active. Initial benchmark results published by Phoronix today show that the NVIDIA Vera CPU meets this need. For this first public look, the benchmark scope was centered on the agentic workloads Vera was designed for in the modern data center. The Vera CPU delivers the throughput AI factories need while optimizing platform power. Eighty-eight NVIDIA custom Olympus cores, 1.2TB/s of memory bandwidth and a high-speed, on-chip fabric results in a CPU platform that combines core performance and memory bandwidth in an efficient power envelope. NVIDIA Olympus Delivers Aggressive Performance At the heart of Vera are custom NVIDIA Olympus CPU cores. Fully compatible with the Armv9.2 instruction set architecture, Olympus is designed for the sequential CPU work underpinning agentic AI: branch-heavy runtimes, sandboxed code, data processing and orchestration. Vera's monolithic die, wide cores, advanced branch prediction and the second-generation NVIDIA Scalable Coherency Fabric help Vera keep data moving across all 88 cores. Phoronix's testing of a single-socket Vera CPU -- rated at 450-watt thermal design power with less than 30 watts of memory power -- showed that it delivers outstanding performance within that power profile, along with generational gains across a broad array of workloads spanning code compilation, file compression, video transcoding, Python, Java and database management. These are the same kinds of CPU-heavy tasks that agents and AI factories run every day: compiling code, executing runtimes, compressing data, querying databases and coordinating large software stacks. "Going into this, I didn't really know what to expect of NVIDIA's Vera with the new Olympus cores," wrote Michael Larabel, founder and principal author of Phoronix. "But in the end I was left realizing this is the most formidable competition to Intel and AMD x86_64 processors ever realized." 'Incredible Advantage' in Memory Performance Agentic workloads are not limited by core count alone. They need high core utilization and sustained memory bandwidth, making memory performance per watt a critical part of overall CPU efficiency. Vera incorporates a second-generation LPDDR5X memory subsystem, enabling dramatically lower energy per bit compared to DDR5. This allows Vera to offer up to a massive 1.2 TB/s of bandwidth -- up to 2x the peak memory bandwidth compared with traditional CPUs in less than 30 watts of memory power, as opposed to more than 100 watts for traditional DDR5. In Phoronix STREAM TRIAD testing, Vera sustained 90% of its peak memory bandwidth -- achieving the highest percentage of rated peak bandwidth of any CPU tested by Phoronix -- and delivered over 4x the memory bandwidth per core compared with traditional x86 CPUs. "NVIDIA Vera with its LPDDR5X memory was showing its incredible advantage in memory performance over current Intel Xeon and AMD EPYC processors," Larabel wrote. However, peak bandwidth is only part of the story. AI factory workloads run many sandboxes, tool calls and data services at the same time. In separate testing with Vera, Prime Intellect found that Vera maintained high bandwidth and low, consistent memory latency as more workloads ran in parallel -- the kind of predictable performance needed for agentic AI. A Large Generational Leap -- and Leadership in Phoronix Testing Compared with the prior-generation NVIDIA Grace CPU, Vera delivered a 1.6x geometric mean increase in Phoronix's testing -- an incredible generation-over-generation gain. "The difference from Grace to Vera was consistently exceeding my expectations for gen-on-gen performance we typically see for processors," Larabel wrote. "NVIDIA's Vera CPU with its in-house-designed Olympus CPU cores ends up packing a heavy-hitting punch with competitiveness to Intel/AMD x86_64 CPUs that I have never seen out of any other ARM or non-x86_64 processors." Vera led the tested CPU field, delivering a 1.5x overall performance advantage compared with a latest-generation 128-core x86 processor. The gains showed up in practical developer workloads. Single-socket Vera compiled a default Linux kernel in just 20 seconds, the fastest result Phoronix measured in that test. Vera delivered 2x faster Linux kernel compilation on a per-core basis compared with a 128-core processor. "On a [geometric] mean basis, the NVIDIA Vera delivered 10% better performance than the AMD EPYC 9575F 5.0 GHz high frequency processor," Larabel wrote. Vera in Customer Testing, Coming Soon From Partners At NVIDIA GTC, NVIDIA announced widespread ecosystem support for Vera, spanning AI natives, supercomputing centers, cloud service providers and infrastructure providers. NVIDIA has also delivered the first Vera CPUs to leading AI companies and cloud providers, marking an important milestone as Vera moves toward partner availability in the second half of the year. Vera will be available from partners in dual- and single-socket systems, with air-cooled and liquid-cooled options to support AI factory deployments, from standard enterprise data centers to high-density agentic AI infrastructure. Learn more about NVIDIA Vera.

NVIDIA Unveils Vera, the CPU for Agents

1.8x Faster Than x86 Processors to Drive Diverse Workloads Across Industries, Generating More Data Center Token Revenue * NVIDIA launches high-performance, energy-efficient NVIDIA Vera CPUs to drive diverse workloads across industries, including agentic AI, reinforcement learning and data processing. * Vera serves as the CPU powering standalone Vera servers, NVIDIA Vera Rubin systems and Vera BlueField-4 STX AI storage platforms. * Global AI labs planning to adopt Vera to transform their AI factories include Anthropic, OpenAI and SpaceXAI, and hyperscalers ByteDance, CoreWeave and Oracle Cloud Infrastructure. * Manufacturers building standalone Vera CPU systems at scale include Dell Technologies, HPE, Lenovo and Supermicro, along with ASUS, Compal, Foxconn, GIGABYTE, Pegatron, Quanta Cloud Technology (QCT), Wistron and Wiwynn. NVIDIA GTC Taipei -- NVIDIA today announced that the world's technology leaders are planning to adopt NVIDIA Vera, the first CPU built for AI agents. Now in full production, NVIDIA Vera is a new class of processor enabling 1.8x faster task completion compared with x86 CPUs to drive diverse workloads across industries -- including agentic AI, reinforcement learning and data processing -- generating more data center token revenue. Building on the success of NVIDIA Grace™ CPUs, which have nearly 2.5 million shipments to date, Vera takes CPU performance and energy efficiency to new levels for the most demanding AI workloads in modern data centers -- where agents move from answering basic questions to taking actions, running code, using tools and evaluating results. Customers exploring the Vera CPU include finance leader NYSE, global AI labs Anthropic, OpenAI and SpaceXAI, and hyperscalers ByteDance, CoreWeave, Lambda, Nebius, Nscale and Oracle Cloud Infrastructure (OCI). Vera is also being integrated into AI infrastructure from world-leading system manufacturers such as Dell Technologies, HPE, Lenovo and Supermicro, along with Taiwan system builders. "AI agents will be the largest users of computing," said Jensen Huang, founder and CEO of NVIDIA. "Vera is the first CPU designed for that future -- built to run agentic AI at hyperscale with extraordinary performance, efficiency and programmability." "At the NYSE, our focus is to optimize the latency, throughput and reliability of the systems underpinning our unrivaled infrastructure," said Lynn Martin, president of NYSE Group. "The NYSE processes more than 1.1 trillion messages per day, and in collaboration with Redpanda and HPE, using NVIDIA Vera CPUs, we will be scaling our capacity while further optimizing latency to power a high-performance, resilient and AI-ready market infrastructure." Anthropic, the AI innovator behind Claude, is evaluating adding Vera to scale CPU-intensive agentic workloads. "Scaling compute is an important accelerant for the growth of models," said James Bradbury, head of compute at Anthropic. "We're excited to see Vera emerge as a promising part of the ecosystem when solving for agentic workloads." OCI Supercluster powered by NVIDIA Vera represents the next frontier in hyperscale AI supercomputing. "Oracle Cloud Infrastructure is rapidly scaling AI infrastructure to meet surging demand for training, inference and agentic AI," said Mahesh Thiagarajan, executive vice president of Oracle Cloud Infrastructure. "By deploying NVIDIA Vera CPUs, OCI will support high-throughput reasoning and data processing workloads across next-generation AI environments." According to Phoronix, which offers a comprehensive, open source benchmarking suite, NVIDIA Vera delivered the fastest overall performance across agentic workloads including code compilation, Python, Java and database processing. These workloads sit on the critical path of modern AI factories, including for agent tool use and sandbox execution, where faster CPU performance delivers higher agent throughput and interactivity. A Custom CPU for the Agentic Era AI factory economics are shifting from cores per dollar to tokens per dollar, requiring CPUs that complete agentic, data-processing and orchestration work faster and more efficiently. Vera is powered by Olympus, a custom NVIDIA CPU core engineered for the CPU work behind that shift, from Python runtimes and sandboxed code execution to orchestration logic and analytics pipelines. Vera is built to process more instructions, anticipate application behavior and move data across large numbers of concurrent environments, queries and data processing tasks -- featuring 88 Olympus cores, Spatial Multithreading and a LPDDR5X memory subsystem that delivers up to 1.2TB/s of bandwidth. This helps agents spend less time waiting on CPU-bound steps and lets AI factories keep accelerators moving. The Vera CPU can also be deployed across the full AI factory -- from the standalone CPU infrastructure to tightly coupled accelerated systems. Vera helps AI factories deliver higher end-to-end throughput and faster time to solution for users, improving responsiveness and efficiency across training, inference and agentic execution. Vera serves as the host CPU for NVIDIA Vera Rubin platforms through second-generation NVIDIA NVLink™-C2C interconnect technology, which provides up to 1.8TB/s of coherent bandwidth between CPU and GPU. It extends NVIDIA Confidential Computing at rack scale, protecting agentic workloads. The NVIDIA Vera BlueField-4 STX processor integrates Vera with high-performance networking, storage acceleration and in-silicon security to create secure-by-design AI-native data platforms. Extensive Ecosystem Support Vera CPUs are available in dense, liquid-cooled racks for large-scale agentic AI and reinforcement learning environments, as well as flexible two-socket air-cooled systems for enterprise, cloud, data processing and AI factory deployments. Leading infrastructure providers offering Vera CPU-based systems include Aivres, ASRock Rack, ASUS, Compal, Dell, Foxconn, GIGABYTE, HPE, Hyve Solutions, Inventec, Lenovo, MiTAC Computing, MSI, Pegatron, Quanta Cloud Technology (QCT), Supermicro, Wistron and Wiwynn. Major original equipment manufacturers -- Dell, HPE, Lenovo and Supermicro -- will be offering Vera in standalone CPU server configurations, the first standard CPU option beyond x86. Leading cloud service providers planning to deploy Vera CPUs include Akamai, ByteDance, Cloudflare, CoreWeave, Crusoe, Lambda, Nebius, Nscale, Oracle Cloud Infrastructure, Redpanda, Starburst, Together AI and Vultr. Availability Vera systems will be available from system builders and cloud partners starting this fall. Watch Huang's keynote and learn more at NVIDIA GTC Taipei.

NVIDIA's Vera CPU With 88 Olympus "Arm" Cores Outperforms AMD EPYC & Intel Xeon In First Benchmarks

The first NVIDIA Vera CPU benchmarks have been released, showcasing a huge gain over Grace while outperforming the latest AMD EPYC & Intel Xeon chips. NVIDIA Vera CPU Brings An Impressive 63% Performance Bump Over Grace CPU, Challenges x86 Chips From AMD & Intel NVIDIA recently announced that its Vera CPUs were in full production, hand-delivering the first CPU racks to major AI firms such as OpenAI, SpaceX, Anthropic, and Oracle. Vera is a fundamental component of the Extreme Co-Design ecosystem powering the Rubin platform, but with Vera, NVIDIA is entering into a new market for the first time: Standalone CPUs. The ARM-based CPU with 88 custom Olympus cores is all set to deliver some big upgrades, with NVIDIA promising 50% better performance, twice the performance per watt (efficiency), and four times the density per rack vs traditional x86 CPUs. It's a CPU that is purpose-built for Agentic AI and Inference domains. Now, the first benchmarks of NVIDIA's Vera CPU have been published by Phoronix, and they showcase a big leap in performance. In the Geomean of all test results, the NVIDIA Vera CPU with 88 Olympus cores ended up 63% faster than its predecessor, the 72-core Grace CPU. The CPU was also 10% faster than AMD's EPYC 9575F, which has a total of 64 "Zen 5" cores clocked at 5 GHz, while also beating the Intel Xeon 6980P, a 128-core chip based on the Granite Rapids family, by 55%. These benchmarks give us a first look at the strong performance that NVIDIA's Vera CPU has to offer, and it's more important than ever to look at NVIDIA's CPU designs since they will not only land in Vera Rubin AI systems, but will also be available in standalone models, with NVIDIA aiming to become the biggest CPU supplier in 2026. The demand for Vera is huge, and since Agentic AI loves CPUs, Vera is primed to eat a big share out of Intel and AMD server segments. At the same time, some important metrics, such as performance per watt (efficiency), are missing from the tests. Phoronix states that they were not permitted to run or showcase these tests. The Vera module they got was an early pre-production hardware, so it looks like we may see some pre-release power tuning & optimizations that will further help the system achieve better performance/efficiency. Phoronix's conclusion states that this is the "most performant ARM Linux server processor" they have ever tested. Looking ahead, NVIDIA Vera CPUs do face some heated competition in the form of AMD's next-gen EPYC Venice, based on the Zen 6 core architecture, which is already in mass production and primed for a 2H 2026 release. Intel is gearing up its Diamond Rapids platform while Qualcomm and Arm are also working on their own data-center-specific CPUs for the Agentic AI race. Follow Wccftech on Google to get more of our news coverage in your feeds.

Nvidia launches Vera CPU for AI agent workloads By Investing.com

TAIPEI - Nvidia Corp. (NASDAQ:NVDA) announced Saturday the launch of its Vera CPU, designed for artificial intelligence agent workloads, with full production now underway. The $5.11 trillion semiconductor giant continues to expand its product portfolio beyond GPUs as it solidifies its position as a prominent player in the industry. The company stated the processor delivers 1.8x faster task completion compared with x86 CPUs for workloads including agentic AI, reinforcement learning and data processing, according to a press release statement. The chip features 88 Olympus cores and a LPDDR5X memory subsystem providing up to 1.2TB/s of bandwidth. Vera serves as the CPU for standalone Vera servers, Nvidia Vera Rubin systems and Vera BlueField-4 STX AI storage platforms. The processor connects to GPUs through second-generation NVLink-C2C interconnect technology, offering up to 1.8TB/s of coherent bandwidth between CPU and GPU. Organizations planning to adopt Vera include Anthropic, OpenAI, SpaceXAI, ByteDance, CoreWeave and Oracle Cloud Infrastructure. NYSE stated it will use Vera CPUs in collaboration with Redpanda and HPE to scale capacity while optimizing latency for its market infrastructure, which processes more than 1.1 trillion messages daily.The Vera launch comes as Nvidia maintains exceptional financial performance, with revenue growth of 71% over the last twelve months and a return on assets of 83%. According to InvestingPro analysis, the stock currently trades below its Fair Value, placing it among the market's most undervalued opportunities despite its strong market position. The company holds an "EXCELLENT" financial health score. "Scaling compute is an important accelerant for the growth of models," said James Bradbury, head of compute at Anthropic. "We're excited to see Vera emerge as a promising part of the ecosystem when solving for agentic workloads." System manufacturers building Vera CPU systems include Dell Technologies, HPE, Lenovo and Supermicro, along with ASUS, Compal, Foxconn, GIGABYTE, Pegatron, Quanta Cloud Technology, Wistron and Wiwynn. Nvidia reported nearly 2.5 million shipments of its Grace CPUs to date. The company said Vera systems will be available from system builders and cloud partners starting this fall.Analysts anticipate continued sales growth for the current year, according to InvestingPro Tips, which offers 18 additional exclusive insights for NVDA investors. For deeper analysis, investors can access the comprehensive Pro Research Report, available for Nvidia and 1,400+ other US equities, transforming complex data into actionable intelligence. Cloud service providers planning to deploy Vera CPUs include Akamai, Cloudflare, Lambda, Nebius, Nscale and Together AI. In other recent news, Nvidia announced that its Vera Rubin platform is entering full production, with collaboration from Taiwan's server manufacturers and global supply chain partners. This platform, designed for AI systems, consists of five purpose-built racks that function as an integrated AI supercomputer. In related developments, Lynx Equity provided a positive outlook on Nvidia ahead of the company's Computex keynote address, suggesting the event could serve as a catalyst for the company's stock. Additionally, Groq is reportedly raising up to $650 million from existing investors following a $17 billion licensing deal with Nvidia. Cadence Design Systems and Samsung Foundry have expanded their collaboration to develop memory and interface intellectual property for Samsung's 2-nanometer process technology, which is intended for AI applications. This partnership enhances Cadence's portfolio, including NVIDIA NVLink-C2C-enabled interconnects. Wolfe Research forecasts a 30% growth in the CPU market by 2028, driven by agentic and orchestration CPUs, with Nvidia's Rubin Ultra being a key player in this growth. These developments reflect ongoing advancements and strategic partnerships in the AI and semiconductor industries. This article was generated with the support of AI and reviewed by an editor. For more information see our T&C.