AI infrastructure startup Modal says it plans to start using Nvidia's Vera CPUs for the first time for hosting agents, alongside its current Intel and AMD chips (Stephen Nellis/Reuters)

Modal announced plans to begin using Nvidia's Vera CPUs to host AI agents. This is notable because Modal has so far used Intel and AMD chips for this purpose. The move expands the processor choices available in its AI infrastructure.

AI agents need hosted computing systems to run their software and respond to tasks. Modal's plan means Vera CPUs will join the processors already supporting those workloads. The article does not provide technical details about the Vera deployment, such as its size, timing, or performance.

The change also represents a new customer win for Nvidia's CPU business. Nvidia has traditionally been associated with GPUs, but the article identifies Modal as another fast-growing AI startup choosing Nvidia for central processor chips. Modal will use Vera alongside, rather than instead of, its existing Intel and AMD chips.

An AI agent is a software system that uses artificial intelligence to handle tasks, make decisions, or respond to instructions. Unlike a single static prediction, an agent may carry out several steps toward a goal. The source identifies Modal as hosting such agents but does not define the term.

Computing infrastructure provides the processors, memory, storage, and networking that let an agent run. A CPU can manage software instructions and coordinate activity, while other hardware may accelerate model calculations. For example, when an agent receives a request, infrastructure can run its program, pass information to an AI model, and return the result.

Hosting matters because agents must operate when users or other systems call them. Modal's planned Vera deployment is therefore about the machines supporting agents, not necessarily about changing what an agent is. The article gives no figures for the number of agents, customers, or servers involved.

Nvidia's Vera CPU is a central processing unit developed by Nvidia. The source identifies it as a processor Modal plans to use for hosting AI agents, but it does not describe Vera's design, specifications, or performance. Those details should not be inferred from the announcement.

A CPU performs general-purpose computing. It reads and executes program instructions, manages operating-system tasks, coordinates data movement, and helps applications communicate with other hardware. In an AI server, it can receive requests, schedule work, and organize the software surrounding an AI model.

This role matters because hosting an AI workload involves more than model calculations. The CPU helps keep the service running and connects incoming requests with the resources that process them. Modal currently uses Intel and AMD chips, and adding Vera gives it a third CPU supplier. The article does not state whether Vera will replace any existing processor.

After adding Nvidia, Modal will have three CPU suppliers. They are Nvidia, Intel, and AMD. The article says Modal plans to use Nvidia's Vera CPUs alongside its current Intel and AMD chips, so the change broadens its hardware mix.

The practical mechanism is supplier diversification. Different processors can support the computing systems that host AI agents, while a company chooses where to run particular workloads. The source does not explain how Modal will assign workloads among the three suppliers, or whether their systems will have different roles.

The immediate reality is coexistence, not replacement. Modal is adding Vera to an existing base of Intel and AMD processors. That gives Nvidia central processor business with another fast-growing AI startup and gives Modal access to a new CPU source. The article provides no details about purchase volumes, server counts, deployment dates, or spending.

Adding Vera changes Modal's infrastructure by expanding its CPU lineup. Before the announcement, the startup relied on Intel and AMD chips for the relevant hosting work. Nvidia's processor will now be available alongside both existing suppliers.

The key mechanism is parallel hardware support. Modal can host AI agents using systems built around more than one CPU supplier, rather than relying on only Intel and AMD. The article does not say whether this will change software, performance, pricing, capacity, or how Modal distributes workloads across the chips.

For Nvidia, the change is a business expansion into Modal's central processor needs. For Modal, it creates an additional option without the article describing any retirement of current hardware. The forward impact remains open: no deployment scale, schedule, or technical comparison is provided. What is clear is that Vera becomes part of Modal's infrastructure choices for hosting agents.

Nvidia is best known for GPUs, but AI infrastructure also depends on central processors. CPUs handle general-purpose software and coordination around computing workloads. The source reports Nvidia winning CPU business at another fast-growing AI startup, showing that its expansion includes more than graphics or accelerator chips.

The immediate example is Modal's decision to use Nvidia's Vera CPUs for hosting agents. Modal had relied entirely on Intel and AMD chips until now. Nvidia's CPU can therefore become part of the servers supporting AI-agent workloads, alongside the company's established processor suppliers.

This matters because demand for AI infrastructure is creating opportunities across the server hardware stack. Nvidia can pursue CPU sales in addition to its traditional GPU business, while startups can consider Nvidia as a processor supplier. The article does not explain Nvidia's broader strategy, market share, or technical advantages, so those implications remain limited to the reported Modal deal.

CPUs and GPUs perform different but complementary jobs in modern AI systems. CPUs handle general-purpose instructions, operating-system activity, request handling, and coordination. GPUs are designed to perform many similar calculations in parallel, which makes them useful for many AI model workloads. The source does not explain Modal's GPU setup.

For example, a server can use its CPU to receive an agent request, organize data, and schedule work. A GPU may process the model's intensive calculations, while the CPU manages the surrounding software and returns the answer. The exact division varies by system, and the article provides no technical diagram or performance data.

Both processors can therefore matter when hosting AI workloads. Modal's announcement concerns the CPU side: Vera will join Intel and AMD chips in its infrastructure. It does not say that Modal is changing its GPUs, replacing existing systems, or using Vera for model training. Those details are not stated in the source.