OpenAI looking to reduce reliance on Nvidia's GPUs, plans to make its own AI chips at TSMC

OpenAI managed to weather a leadership crisis and is now trying to strengthen its position on the market, as it plans to enter the AI chip maker arena. The Financial Times reports CEO Sam Altman is currently in talks with UAE investors to cover the funding side, as well as with TSMC for production purposes. These efforts should help the company reduce reliance on Nvidia’s GPU hardware that is in very great demand and difficult to procure.

Such an endeavor would certainly require billions of USD, so Altman is turning to UAE investors like Sheikh Tahnoon bin Zayed-al-Nahyan, who is the brother of UAE’s president and chairman to another AI company called G42, closely partnered with Microsoft and OpenAI.

Altman eyes the Taiwanese foundries for production since they currently have the most advanced nodes. However, many other important players are fighting for production capacity at TSMC, which pushes costs higher, so OpenAI would have to negotiate a lucrative partnership.

In order to design proprietary AI chips, OpenAI is apparently planning to harness the power of its own AI algorithms. The process will take several years most likely if it is done from scratch, and OpenAI will also have to keep an eye on all the latest hardware developments from Nvidia, which intends to release new AI GPU families every year moving forward. It may prove quite the venture for OpenAI, but the company can definitely get some inspiration from Amazon and Google - two companies that already develop their own AI chips.