NVIDIA funds customers building rivals

Nvidia is financing customers that are also trying to reduce their dependence on Nvidia. Axios reported on August 5 that the company is weighing more than $750 billion in AI investments, financing deals and partnerships tied to customers building data centers, custom chips and other infrastructure around artificial intelligence. The company’s bet, Axios said, is that growth in the broader compute market will still flow back to Nvidia’s core chip and networking businesses. That approach comes as Nvidia remains the central supplier to much of the AI buildout, even while large customers search for ways to lower costs and diversify suppliers. ### Why would Nvidia fund companies trying to replace parts of Nvidia’s stack? Axios reported that Nvidia is “bankrolling the same customers” racing to build alternatives to its chips. The article said Nvidia is putting its balance sheet behind customers building alternatives to some of its most profitable products, including efforts around custom silicon, servers and infrastructure that could compete with Nvidia offerings in specific parts of the stack. (axios.com) The logic, as described by Axios, is that Nvidia expects overall AI demand to expand fast enough that even partial substitutes can still increase demand for Nvidia components, software and interconnects. That makes the company less a pure component vendor than a financier and infrastructure partner to customers building large AI systems. ### Which customers and partners show how this model works? OpenAI and CoreWeave are two of the clearest examples in Nvidia’s public disclosures and related reporting. (axios.com) Nvidia and OpenAI said in September 2025 that they planned a strategic partnership to deploy at least 10 gigawatts of Nvidia systems for OpenAI’s next-generation AI infrastructure, with Nvidia intending to invest up to $100 billion progressively as each gigawatt is deployed. The first gigawatt was scheduled for deployment in the second half of 2026 on the Vera Rubin platform, according to Nvidia’s investor release. CoreWeave said in a Nvidia Newsroom release that Nvidia invested $2 billion in CoreWeave Class A common stock at $87.20 per share as part of an expanded relationship to build more than 5 gigawatts of AI factories by 2030. Axios separately reported in May that Groq had previously signed a $20 billion licensing deal with Nvidia. Together, those deals show Nvidia using capital, long-term supply commitments and commercial agreements to anchor customers inside its ecosystem. (investor.nvidia.com) ### If Nvidia is so dominant, why do customers keep building alternatives? Large AI buyers are still pursuing custom chips and other alternatives because the scale of spending is enormous. Axios reported in May that Nvidia posted quarterly revenue of $81.6 billion, ahead of analyst estimates, underscoring how much of the current AI buildout still runs through Nvidia hardware. At the same time, another Axios report in July said the bottleneck is not only chip design but also manufacturing capacity, high-bandwidth memory and advanced packaging. (nvidianews.nvidia.com) Those constraints help explain why customers want more than one path. Even when companies design their own silicon, they still need foundry capacity, memory, networking, software integration and power-heavy data center buildouts. Nvidia’s financing can help keep those projects moving while preserving demand for Nvidia systems in adjacent layers. That is an inference from the company’s disclosed partnerships and Axios’s reporting on supply-chain bottlenecks. (axios.com) ### What does the market reaction say about Nvidia’s position? TS2 reported on August 6 that Nvidia’s market value rose by about $177.6 billion on Wednesday after SpaceX opted to rely solely on Nvidia chips. The same coverage said export controls and chip-supply restrictions could still delay revenue, even as investor enthusiasm remained strong. July market data cited by TS2 also showed how exposed Nvidia remains to China-related policy decisions. (axios.com) TS2 reported on July 9 that Nvidia shares closed at $204.12 on July 8, giving the company a market capitalization of about $4.94 trillion, while a U.S. interagency review was weighing possible H200 chip sales to China under a licensing framework. ### Where could this strategy run into trouble? (ts2.tech) U.S. export controls remain one clear pressure point. Reuters-cited reporting carried by Investing.com said on August 5 that the Federal Communications Commission was working on rules to bar imports of new Chinese optical transceivers used inside data centers, with officials aiming to publish the measure this year. That reporting suggested the policy fight is widening beyond flagship AI chips to the components that connect AI systems. (ts2.tech) The next test will come through Nvidia’s customer deployments and regulatory milestones. Nvidia has said the first gigawatt of systems for OpenAI is due in the second half of 2026, while investors are also watching future earnings reports, export-license decisions and supply-chain capacity for memory, packaging and power infrastructure. (investor.nvidia.com) (ts2.tech)