Perplexity signs up for Nvidia Vera CPU

- Perplexity confirmed on July 7 that it plans to use Nvidia’s Vera CPU for AI agent workloads, giving Nvidia an early customer reference. - Perplexity executive Dmitry Shevelenko told Reuters Vera was a “dead-on fit,” while BigGo Finance reported agent tasks ran about 1.5 times faster. - Nvidia says Vera systems are being adopted by server makers including Dell, HPE, Lenovo and Supermicro, with broader rollout tied to Rubin platforms.

Perplexity said on July 7 that it plans to use Nvidia’s new Vera central processing unit for AI agent workloads, handing Nvidia an early public customer as it pushes beyond graphics processors and AI accelerators. Reuters reported the confirmation from Perplexity as Nvidia tries to expand in the data-center CPU market, where Intel and Advanced Micro Devices have long been entrenched. The Next Web separately reported that Perplexity would run its agent workloads on Vera, describing the company as an early adopter. Nvidia has said it expects Vera to become a significant business this fiscal year. ### Why is Perplexity using Vera instead of sticking with standard server CPUs? Perplexity executive Dmitry Shevelenko said Vera was a “dead-on fit” for the company’s agent workloads, according to Reuters. The report said Perplexity had tested agentic coding tasks on Vera and found the chip better suited to that work than conventional server processors from Intel and AMD. (money.usnews.com) BigGo Finance reported that Perplexity said its AI agents executed tasks about 1.5 times faster on Vera than on traditional CPUs. That figure has also been echoed in secondary coverage, though the underlying benchmark details were not published in the reports reviewed. ### What exactly is Nvidia selling with Vera? Nvidia describes Vera as “the CPU for agents” on its product page and says the chip is part of the broader Vera Rubin platform. (money.usnews.com) The company’s data-center materials say a Vera Rubin NVL72 system combines 72 Rubin GPUs, 36 Vera CPUs, ConnectX-9 SuperNICs and BlueField-4 DPUs in a rack-scale design. Nvidia’s newsroom said last month that Vera is “the first CPU built for AI agents.” The company said manufacturers planning standalone Vera CPU systems at scale include Dell Technologies, HPE, Lenovo and Supermicro, along with several Asian server makers. (finance.biggo.com) ### How important is one customer endorsement to Nvidia’s CPU push? Nvidia told investors, as cited by Reuters, that it expects to generate $20 billion in Vera CPU sales by the end of the current fiscal year. (nvidia.com) That target is separate from the company’s established AI accelerator business and puts unusual weight on early customer names that Nvidia can cite publicly. (nvidianews.nvidia.com) The Next Web reported that Perplexity is among the first notable AI companies to say it will build on Vera. Nvidia’s own blog this week also listed Perplexity among “AI innovators” adopting Vera and framed the product as a new class of CPU for the “agentic AI era.” ### Where does this leave Intel and AMD? Reuters said Nvidia is using Vera to challenge Intel and AMD in general-purpose data-center computing, not just in AI accelerators. (money.usnews.com) Perplexity’s decision does not amount to broad market share data, but it gives Nvidia a named software customer discussing a specific workload category where Vera may have an advantage. (thenextweb.com) The reports reviewed did not show Intel or AMD responding publicly to Perplexity’s endorsement. BigGo Finance characterized the comparison as running against “traditional CPUs,” while Reuters identified Intel and AMD as the incumbent rivals in the market Nvidia is targeting. ### What happens next as Vera reaches the market? Nvidia has tied Vera to its Rubin generation of AI infrastructure, with official materials pointing to adoption by server makers and system builders. (money.usnews.com) The company’s technical blog and product pages position Vera as part of rack-scale deployments rather than a standalone consumer launch. (finance.biggo.com) Perplexity has not publicly given a deployment date in the reports reviewed. Nvidia’s next milestones are likely to come through Rubin-platform shipments and partner system rollouts from companies including Dell, HPE, Lenovo and Supermicro, all named by Nvidia as planned adopters. (nvidianews.nvidia.com) (developer.nvidia.com)

Get your own daily briefing

Scout delivers personalized news, insights, and conversations tailored to your role and industry.

Download on the App Store

Shared from Scout - Be the smartest in the room.