OpenAI is investing $10bn in a deal with US semiconductor firm Broadcom to develop its own artificial intelligence chips, according to reports in the Financial Times and Wall Street Journal.
The partnership is aimed at reducing OpenAI’s reliance on Nvidia, whose processors currently dominate the AI industry. The new custom chips are expected to be available from next year and will be used internally by OpenAI to train and run models such as ChatGPT, alongside other AI products.
Read More: Tesla board proposes pay deal that could make Musk world’s first trillionaire
Broadcom had hinted at a major new contract last week, though details only emerged later. The agreement would represent a significant boost for the California-based chipmaker, with OpenAI providing one of its biggest customers.
Nvidia has until now been the primary supplier of high-performance processors to cloud providers such as Amazon Web Services, Google, Microsoft and Oracle. Oracle recently announced plans to buy more than $40bn worth of Nvidia chips to power a new data centre tied to the so-called “Stargate Project”, an initiative by AI companies to expand computing infrastructure.
Also Read: Chinese tech giants pursue Nvidia AI chips despite Beijing’s warning
Hints of OpenAI’s in-house chip ambitions surfaced earlier this year, as pressure mounted on major players to secure more processing power. Google, Amazon and Microsoft are all developing their own AI chips in an effort to diversify away from Nvidia.
Despite this, Nvidia remains in high demand. The company reported a 56% rise in quarterly sales last week, underscoring its market strength. Meanwhile, reports suggest the Trump administration could ease some trade restrictions, potentially reopening major international markets to Nvidia’s latest chips.
