AI Hardware Shakeup: Perplexity Picks Nvidia Vera CPUs Over AMD
July 12, 2026 — A surprise decision by Perplexity, a rising AI coding platform, is reverberating through the chip sector. The company said it will deploy Nvidia's Vera CPU family for its multi-agent AI stack, a move that elevates speed claims and reshapes the competitive landscape. Perplexity asserts Vera delivers about 1.5x faster throughput than typical server processors in its workloads, a finding that could influence enterprise buying across cloud and on-premises data centers.
Industry watchers note the shift goes beyond a single product choice. Nvidia is marketing a bundled, end-to-end AI stack — GPUs, CPUs, networking, and software — while AMD presses on with its EPYC CPUs and MI accelerators. The outcome could widen the gap between a vertically integrated hardware software approach and a more modular, CPU-centric strategy.
Industry watchers are tagging it as nvidia amd: perplexity choosing, a shorthand for a broader shift in AI infrastructure and how buyers value a single-vendor experience versus a multi-vendor mix.
Why Perplexity’s Move Matters
The decision highlights a potential rethinking of the old server playbook. If a single vendor can supply compute, interconnect, and software tools in one package, developers may gain faster deployment times, simpler procurement, and clearer total cost of ownership for AI workloads. That has implications for cloud providers, systems integrators, and venture-backed AI shops weighing their flagship architectures.
“This isn’t just about hardware speed; it’s about the architecture and ecosystem that enable AI to scale quickly,” said a senior analyst at TechStride Partners. “Perplexity’s move draws attention to Vera’s performance in large-scale AI stacks and could influence buying patterns across enterprise buyers.”
Data Center Dynamics At a Glance
- Data Center Revenue: Nvidia’s data center business remains the largest growth engine, with revenue measured in the tens of billions of dollars in the latest quarter. AMD’s data center revenue trails, landing in the single-digit billions range.
- Non-GAAP Gross Margin: Nvidia maintains a high gross margin around the mid-70s, reflecting software and networking advantages, while AMD sits in the mid-50s to low-60s on its data center stack.
- Product Strategy: Nvidia’s stack blends Vera CPUs, GPUs, InfiniBand/Spectrum interconnects, and software tooling into a cohesive platform. AMD leans on EPYC CPUs paired with MI accelerators, relying more on partnerships for software value.
- Market Position: The Perplexity win accelerates Nvidia’s push into CPU territory, challenging traditional x86 dominance and pressuring AMD to defend its server CPU franchise while expanding accelerator leadership.
Market Pulse: Reactions And Implications
Stock and bond markets reacted with a mix of optimism and caution. Nvidia’s shares moved higher in extended trading as investors priced in stronger data center attachment and faster AI deployment cycles. AMD faced a tempered reaction, as analysts debated how quickly EPYC and MI accelerators can sustain the platform in a quickly bundling world.
Several hedge funds and tech-focused funds noted the risk of vendor lock-in should more buyers follow Perplexity’s path. At the same time, buyers say a tightly integrated stack can dramatically reduce integration headaches for large-scale AI deployments, a factor that could drive broader adoption of Vera and Nvidia-enabled workloads.
Outlook: The Road Ahead
The industry will watch closely how additional customers respond to the Vera option and whether Perplexity’s early results translate into broader purchasing momentum. If the shift accelerates, Nvidia could extend its advantage in AI training and inference workloads, not only on GPUs but across a unified platform.
AMD remains confident in its CPU and accelerator roadmap. Executives stress continued strength in EPYC performance and partnerships, while exploring incremental software enhancements to close any perceived gaps with a bundled stack. The long arc of the race remains intact: can a single-stack approach outpace a modular, best-of-breed strategy in a market defined by rapid AI experimentation?
For investors, the key question is whether nvidia amd: perplexity choosing represents a durable trend or a temporary divergence in customer preference. If Perplexity’s early results hold and more large customers migrate to a bundled Nvidia stack, the race could tilt toward Nvidia in both price-to-performance and total cost of ownership for AI workloads. If AMD answers with sharper software tie-ins and faster CPU-to-accelerator integration, the balance could shift again. In the end, the phrase nvidia amd: perplexity choosing has become a focal point for assessing the future of AI infrastructure.
Discussion