Why Nvidia scrapped its $100 bn OpenAI deal for a $30bn investment

Nvidia's equity bet follows the collapse of a long-term compute pact, as OpenAI secures fresh funding and expands infrastructure partnerships beyond a single chip supplier

Nvidia is close to finalising a $30 billion equity investment in OpenAI, replacing a previously announced $100 billion multiyear partnership that never progressed beyond a preliminary agreement, according to a report by the Financial Times. 

The earlier arrangement, which was announced in September last year as a letter of intent, aimed at Nvidia investing in 10 instalments of $10 billion over several years. In return, OpenAI planned to deploy up to 10 gigawatts of new computing capacity and purchase millions of Nvidia’s artificial intelligence processors. 

The deal was designed as a long-term strategic tie-up between the world’s leading AI chipmaker and one of its largest customers. But the intent never materialised into a formal contract. By January this year, The Wall Street Journal reported that the agreement was effectively on hold because of broader investor caution around artificial intelligence investments. US technology stocks have fallen 17 per cent since the start of the year, reflecting concerns about valuations and the pace of infrastructure spending.

 

What does the new $30bn Nvidia investment in OpenAI involve?

The revised arrangement is structured as a direct equity investment, according to the FT, in which Nvidia will invest up to $30 billion in OpenAI in exchange for shares in the company, replacing the earlier staggered investment model that was tied to compute expansion. 

This investment forms part of a larger funding round expected to raise more than $100 billion for OpenAI. The round would value OpenAI at $730 billion, excluding the fresh capital being raised. 

OpenAI is expected to reinvest much of the new funding into expanding its computing infrastructure, which relies heavily on specialised AI chips. The funding will support the build-out of gigawatts of additional compute capacity over time. 

The funding round also includes other major investors, the FT said, with SoftBank in the final stages of negotiations to invest $30 billion, while Amazon could invest up to $50 billion as part of a broader partnership involving the use of OpenAI’s models.  ALSO READ | Tata signs OpenAI as 1st customer for data centre under Stargate initiative

Why has OpenAI diversified beyond Nvidia chips?

The collapse of the earlier partnership deal coincides with OpenAI expanding its hardware supplier base beyond Nvidia. 

In October last year, OpenAI announced a major partnership with AMD to secure up to six gigawatts of computing capacity using AMD’s Instinct MI450 chips. The agreement includes warrants for 160 million AMD shares worth more than $26 billion. 

Reuters reported that OpenAI has explored alternative chip providers such as Cerebras and Groq since 2025. These suppliers are expected to handle about 10 per cent of OpenAI’s inference workload, which involves running trained models rather than building them. 

Reports of these moves had fuelled speculation about tensions between Nvidia chief executive Jensen Huang and OpenAI chief executive Sam Altman, although both have publicly denied any friction, with Altman saying OpenAI hopes to remain a major Nvidia customer.

How important is computing power to OpenAI’s growth?

OpenAI’s revenue has closely tracked its computing capacity, with the company reporting annualised revenue of more than $20 billion earlier this year. 

According to the FT, OpenAI’s access to computing power and its revenue have each roughly tripled annually, which reflects the central role of specialised AI hardware in training and operating large language models such as ChatGPT.