Nvidia expands its influence into the central processor market after winning cloud company Modal to run AI agent workloads on Vera chips, which doubled performance in initial testing.
Modal adopts Vera central processors to run AI agents
The integrated role of central and graphics processors in smart agent architecture
The mixed infrastructure environment and Modal’s financial growth context
Nvidia’s Vera sales projections and the limits of preliminary results
Frequently asked questions
Modal adopts Vera central processors to run AI agents
Modal, a provider of cloud computing services for artificial intelligence, announced on October 19, 2026, that it plans to begin using Nvidia’s “Vera” central processors to run workloads associated with AI agents. Until now, Modal had relied on central processors supplied by Intel and AMD to power this vital part of its infrastructure.
This announcement, reported by Reuters based on interviews with Modal and Nvidia officials, represents a significant development for Nvidia as it seeks to expand its market position beyond the graphics processors that currently dominate most advanced AI tasks. Reuters reported that early testing conducted by Modal indicated that the new Vera central processor can double the amount of work completed by AI agents compared to the processors the company was previously using.
The integrated role of central and graphics processors in smart agent architecture
The central processing unit handles general computing operations, while the graphics processing unit excels at executing thousands of mathematical calculations in parallel. Modern AI infrastructure largely relies on both types together; massive model calculations run on graphics processors, while central processors coordinate software tasks, data processing, network operations, and supporting activities.
AI agents can generate massive workloads specifically for central processors because they repeatedly execute software tools, manage instructions, interact with applications, and continuously process intermediate results. Therefore, the commercial opportunity available to Nvidia is much broader than simply selling accelerators used directly in model calculations.
The mixed infrastructure environment and Modal’s financial growth context
Modal explained that its testing of the Vera processor showed tangible benefits in agent hosting tasks, but its CEO, Erik Bernhardsson, clearly emphasized that the company is not replacing all of its Intel and AMD processors. Instead, Modal plans to run a mixed infrastructure environment and distribute workloads to the most suitable hardware based on observed actual performance, especially since the comparison included processors currently available to the company while Intel and AMD develop newer generations that may yield different results.
Modal’s technical decision follows a separate funding event for the company the previous week. Reuters reported that Modal raised $750 million at a valuation of approximately $15.75 billion, and expects to generate nearly $1 billion in sales over the coming year—figures that highlight the scale of the new customer, even though the October 19 event relates exclusively to the adoption of Vera processors.
Nvidia’s Vera sales projections and the limits of preliminary results
Nvidia previously informed investors that it expects to generate at least $20 billion in sales from Vera processors during its current fiscal year, a figure that represents a forward-looking projection for the company rather than realized revenue from completed sales. Modal’s addition to the list of companies adopting the Vera processor indicates that Nvidia’s competition with Intel and AMD is strongly extending into the general processor infrastructure surrounding AI agents, pending actual production measurements to confirm the initial test results.
Modal’s cloud decision to adopt Nvidia’s Vera central processors to run AI agent workloads is a clear indicator of changing computing load dynamics in the era of autonomous agents. While the language model training phase relied almost entirely on graphics processors, running smart agents requires constant interaction with software tools, databases, and networks, placing a heavy burden on central processing units.
By offering the Vera processor, which is capable of doubling agent productivity according to preliminary tests, Nvidia is opening a direct competitive front in the heart of Intel and AMD’s traditional market. It aims to achieve its target of $20 billion in Vera sales during its current fiscal year, while cloud companies like Modal maintain hybrid environments that balance various processors based on performance efficiency.
Frequently asked questions
Question: What did cloud company Modal announce regarding its infrastructure?
Answer: Modal announced its intention to adopt Nvidia’s Vera central processors to run AI agent workloads after previously relying exclusively on Intel and AMD.
Question: What did Modal’s early testing of the Vera processor show?
Answer: The company’s preliminary testing indicated that the Vera central processor can double the volume of work completed by AI agents compared to its existing processors.
Question: Will Modal completely phase out Intel and AMD processors?
Answer: No, CEO Erik Bernhardsson emphasized that Modal will run a mixed infrastructure environment and distribute tasks across different processors based on actual performance.
Question: What are Nvidia’s revenue expectations for Vera processors during its current fiscal year?
Answer: Nvidia informed investors that it expects to achieve at least $20 billion in Vera processor sales during the current fiscal year.
