This
image illustrates the partnership between Perplexity AI and Nvidia,
highlighting Perplexity's use of Nvidia's new "Vera" CPU to power its AI
agent workloads. The visual indicates that the Vera chip delivers 1.5
times faster performance for continuous AI tasks compared to traditional
processors from competitors like Intel and AMD. Generated by AI
Perplexity Taps Nvidia's "Vera" CPU for AI Agent Workloads
 |
This image illustrates the partnership between Perplexity AI and Nvidia, highlighting Perplexity's use of Nvidia's new "Vera" CPU to power its AI agent workloads. The visual indicates that the Vera chip delivers 1.5 times faster performance for continuous AI tasks compared to traditional processors from competitors like Intel and AMD. Generated by AI |
AI startup Perplexity confirmed on Tuesday that it plans to utilize Nvidia’s new central processing units (CPUs). This major move comes as the chip giant actively works to broaden its market reach and take on entrenched industry players like Intel and Advanced Micro Devices (AMD).
Nvidia is projecting massive financial returns from this hardware pivot, expecting to generate $20 billion in sales from its "Vera" CPU by the end of this fiscal year. Unlike Nvidia's highly specialized AI accelerators, the Vera chip is a more generic computing processor. These chips represent a key part of Nvidia’s strategy to diversify its sales, especially as major artificial intelligence companies—such as OpenAI and DeepSeek—begin developing their own proprietary AI chips.
A Crowded Market Meets a New Kind of AI Demand
By pushing the Vera CPU, Nvidia is entering a crowded server processor market that has long been dominated by Intel and AMD. Historically, these two companies have supplied CPUs for everything from standard consumer laptops to massive web servers.
However, many of those traditional chips were designed well before the rise of AI "agents". These agents are advanced AI systems that can carry out complex, multi-step tasks autonomously after receiving initial instructions from human users.
The workload demands of an AI agent are drastically different from typical computing:
Unlike human users who inevitably take breaks between tasks, AI agents operate continuously without stopping.
Traditional CPUs can struggle to handle the sustained, rapid-fire processing loops required for agentic workflows.
Why Perplexity is Making the Switch
For Perplexity, the decision to adopt Nvidia's hardware comes down to pure performance. Nate Kupp, Perplexity’s Vice President for Computer Enterprise and Infrastructure, revealed that Nvidia’s CPU executed AI agent coding tasks approximately 1.5 times faster than the traditional CPUs they had been using.
“Vera really stood out to us as just like a dead-on fit for a lot of the core workloads that we have,” Kupp said in an interview.
While Perplexity declined to disclose exactly how many Nvidia CPUs it plans to purchase, they are joining a rapidly growing list of high-profile adopters. Nvidia has previously disclosed that major tech players, including OpenAI, Anthropic, and Oracle, also plan to use its new CPUs.