Modal Adopts Nvidia Vera CPUs After Internal Tests Show 2X Gains
AI cloud provider Modal is turning to Nvidia's new Vera central processors after internal evaluations revealed dramatic performance improvements for agentic AI workloads.
Radwan AbdulhadiEditor and technology writer
Nvidia Secures Major Win With Vera CPUs at Modal
Nvidia has long dominated the market selling AI accelerators, but the rapid emergence of agentic AI has elevated the role of central processing units. According to reports covered by outlets like [Tom's Hardware](https://www.tomshardware.com/pc-components/cpus/nvidia-vera-cpu-deployed-at-ai-startup-modal-alongside-existing-x86-fleet-internal-testing-showed-2x-throughput-compared-to-amd-and-intel), Nvidia has secured a major design win with its Vera CPUs at AI cloud provider Modal.
Preliminary internal evaluations conducted by the startup showed that the Vera processor could deliver twice the throughput of its current Intel and AMD processors. Despite this performance advantage, Modal does not plan to completely phase out hardware from its current CPU suppliers.
Demands of Agentic AI Workloads
Modal provides critical cloud infrastructure enabling developers to deploy AI agents capable of performing complex, autonomous operations over extended periods of time. These sophisticated workloads require substantial CPU resources to handle software execution, data processing, and auxiliary tasks that cannot be efficiently offloaded to GPUs alone.
While the firm has historically relied exclusively on AMD and Intel x86 processors for these operations, internal benchmarks indicate that Nvidia's new architecture can significantly enhance efficiency. However, because the Vera platform is based on the Arm instruction set architecture, adopting the new chips requires substantial engineering and software modifications.
Performance Details and Architectural Questions
The reported 2x throughput advantage is particularly compelling because agentic AI workflows frequently involve numerous independent tasks competing heavily for CPU cores, cache capacity, and memory bandwidth. Detailed benchmark configurations, exact processor models tested against Vera, and performance-per-watt metrics have not been publicly disclosed.
Consequently, it remains difficult to isolate whether Vera's advantage stems from its Arm v9.2-based Olympus CPU cores—which can decode and execute up to 10 instructions per cycle—its memory subsystem offering 1.2 TB/s of bandwidth, or other distinct architectural characteristics. Furthermore, the tests compared Vera against currently available AMD and Intel silicon rather than upcoming competitive generations.
Operating a Heterogeneous Server Fleet
Because Modal cannot instantly port its entire software stack to the Arm architecture overnight, the company will maintain a mixed hardware infrastructure. As noted in insights shared [According to Modal](https://modal.com/blog/nvidia-vera-sandboxes), chief executive Erik Bernhardsson confirmed that the company will continue utilizing x86 CPUs alongside the new Nvidia hardware.
Rather than committing exclusively to a single instruction set architecture for all operations, Modal intends to operate a heterogeneous server fleet. This strategy allows the firm to select the most suitable processors for each specific workload while continuing to evaluate upcoming generations of AMD and Intel silicon as they become available.
Broader Market Impact for Nvidia
For Nvidia, this agreement marks a significant milestone in establishing Vera as a standalone server processor capable of capturing a substantial share of the broader data center market. The hardware maker anticipates that Vera sales will generate at least $20 billion during its current fiscal year, while estimating the total addressable market for these CPUs to be approximately $200 billion, putting the company on track to ship millions of units.
This technological adoption follows a period of rapid financial and operational growth for Modal, which recently secured $750 million in financing led by Accel at a valuation of $15.75 billion. With service consumption surging significantly over the past year, the company continues to scale its infrastructure to meet soaring demand for autonomous AI agent solutions.
Sources
- Tom's HardwareModal taps Nvidia Vera CPUs after firm's internal tests show 2x throughput over Intel and AMD chips
Continue chronologically
Related entity coverage
Related stories
Intel's Gloomy PC Outlook Triggers $1 Trillion Chip Market Wipeout
Intel shares plunge 30% after a weak PC outlook, erasing $150 billion and sparking a wider semiconductor market bloodbath across the industry.
AMD Surpasses $1 Trillion Market Cap Milestone
AMD has officially joined the elite trillion-dollar market cap club, outpacing Intel following a significant surge in stock price fueled by data center demand.
Nvidia Vera CPUs Land in Oracle Cloud
Nvidia's standalone Vera CPUs arrive in Oracle Cloud Infrastructure Superclusters as part of a multi-gigawatt deployment.
How to Boost GPU Performance Without Buying New
Learn how to optimize your older graphics card, utilize AI supersampling, and maintain your PC hardware to boost gaming frame rates without upgrading.
