Silicon Valley’s chip heavyweight is successfully breaching territory long defended by Intel and AMD. AI search startup Perplexity confirmed it is adopting Nvidia’s new central processing unit (CPU), dubbed “Vera,” marking a crucial win for Nvidia as it works to diversify its hardware ecosystem.

The shift into traditional computing chips comes at a vital time. With major generative AI labs like OpenAI and DeepSeek increasingly building their own custom AI silicon, Nvidia is expanding into the broader CPU market—a sector historically dominated by legacy players powering web servers and laptops. Nvidia is already projecting a massive $20 billion in revenue from Vera by the end of the current fiscal year.

Perplexity’s decision to adopt the new silicon comes down to the relentless, 24/7 demands of modern artificial intelligence. While traditional CPUs were engineered for human workflows that feature frequent pauses, they struggle under the continuous, heavy operational cycles of autonomous “AI agents.” According to Nate Kupp, Perplexity’s VP of Computer Enterprise and Infrastructure, Nvidia’s Vera chip handles complex agent coding tasks roughly 1.5 times faster than standard legacy processors.

While Perplexity kept the exact size of its hardware order under wraps, they aren’t the only tech giants making the jump. Nvidia has already secured a powerhouse roster of early adopters for the Vera line, including Anthropic, Oracle, and OpenAI.