Nvidia has secured a notable early customer for its push into the central-processor market, with fast-growing AI cloud startup Modal planning to deploy the chipmaker’s new Vera CPU. The move is significant because Modal has until now relied on Intel and AMD processors for the workloads that sit outside the GPU-heavy part of its infrastructure. Early tests, according to Reuters, showed that Nvidia’s Vera processor roughly doubled the amount of work AI agents could complete compared with the Intel and AMD chips Modal currently uses.
Nvidia is moving beyond GPUs
Nvidia built its dominant AI position around graphics processors, but the company is increasingly trying to own more of the server stack. Vera is central to that strategy because CPUs still handle orchestration, memory management and many general-purpose tasks even in AI-heavy data centers.
Modal is a strategically useful customer
Modal rents computing power to AI developers and has grown rapidly with demand for agentic AI systems. The startup recently raised $750 million at a $15.75 billion valuation and says it is on track for about $1 billion in sales over the next year.
AI agents create different hardware needs
Traditional model training is overwhelmingly GPU-centric. Long-running agents, by contrast, can involve repeated tool calls, memory access, scheduling and other CPU-intensive work. That makes processor performance more important than in many earlier AI workloads.
Modal is not abandoning Intel or AMD
Chief Executive Erik Bernhardsson said the company intends to run a mixed fleet. Nvidia chips will sit alongside existing Intel and AMD systems, with jobs assigned to whichever hardware performs best.
That mixed model raises competitive pressure
For Intel and AMD, the immediate risk is not losing every workload but losing the fastest-growing ones. If AI agents become a major share of cloud demand, superior performance in those workloads could shift future infrastructure spending.
Nvidia’s software advantage may matter
Nvidia’s success in GPUs has always rested on both hardware and software. A similar strategy in CPUs could make Vera attractive to customers already deeply embedded in Nvidia’s AI ecosystem.
Modal says software changes will be required
Supporting Vera is not a plug-and-play decision. Modal said the transition will require deep software work, underlining that CPU architecture changes can carry meaningful engineering costs.
The company will keep testing rival chips
Intel and AMD are both preparing newer products that could narrow the performance gap. Modal says it plans to test them as they become available, which means Nvidia’s early advantage will have to be defended.
Nvidia sees a huge CPU market
The company has told investors it expects at least $20 billion in Vera sales in the current fiscal year and has described the broader opportunity as roughly $200 billion. That forecast implies Nvidia sees CPUs as a major new growth engine rather than a side product.
Other AI companies are also interested
Nvidia has previously said OpenAI, Anthropic, Oracle and Perplexity are among customers or partners planning to use Vera-based systems. Modal adds another high-growth AI-native company to that list.
The data-center contest is becoming broader
The AI hardware market is no longer simply about who makes the fastest accelerator. Customers increasingly care about how CPUs, GPUs, networking and memory work together. Nvidia’s attempt to compete across that entire stack could reshape spending patterns throughout the data-center industry.
Sources
- Reuters, Oct. 19, 2026
- Nvidia and Modal company statements cited by Reuters
