According to Reuters, NVIDIA has secured a central processing unit order from another fast-growing AI startup, a company that previously relied entirely on chips from Intel and AMD. This serves as an early indicator that NVIDIA’s strategy to enter the market long dominated by its two major rivals has begun to bear fruit.
Cloud computing startup Modal provides compute rental services for AI developers. The firm stated it plans to deploy NVIDIA’s new Vera Central Processing Units for the first time to host AI agents, AI systems capable of executing long-running tasks on behalf of humans. Previously, Modal’s workloads of this kind ran exclusively on Intel and AMD chips.
Modal completed a $750 million funding round led by Accel last week, with a post-money valuation of $15.75 billion. The company projected its revenue could reach $1 billion in the coming year. Modal reported that usage of its services has tripled, with year-over-year revenue growth exceeding 15 times.
Modal said on local-time Monday that early testing demonstrated NVIDIA’s new chip doubles task throughput for AI agents, prompting its decision to adopt NVIDIA’s chip solution for the first time. This adoption requires Modal to carry out extensive rework on its underlying software stack.
Nevertheless, CEO Erik Bernhardsson emphasized that the company will not drop its existing suppliers. Modal will adopt hybrid compute clusters, where NVIDIA chips will be deployed alongside current Intel and AMD hardware, routing different workloads to the most suitable hardware.
Bernhardsson noted that the benchmark for the tests was commercially available chips from Intel and AMD. Both rival vendors are developing next-generation products that are expected to narrow the performance gap with NVIDIA.
“We will test the new products once they launch,” he added. The hybrid hardware architecture enables the company to “deploy the highest-performing hardware available at any time.”
NVIDIA disclosed to investors that it targets at least $2 billion in Vera chip sales for the current fiscal year. This marks NVIDIA’s opening foray into this roughly $200 billion market. NVIDIA previously shared that AI players including OpenAI, Anthropic, Oracle and Perplexity also plan to adopt the chip.
(Reprinted from https://news.eccn.com/）