Microsoft is leaning even more on OpenAI, this time not for AI models, but for the chips that power them.
According to Bloomberg, the tech giant plans to use its partner’s custom chip designs to strengthen its own lagging semiconductor program, a move that feels both pragmatic and overdue, given how far behind Microsoft is compared to rivals like Google and Amazon, who’ve already built their own AI-focused chips.
Here’s how the deal works: OpenAI is developing AI chips with Broadcom, and Microsoft gets full access to everything that comes out of that partnership.
“As they innovate even at the system level, we get access to all of it,” Microsoft CEO Satya Nadella said in a recent interview with podcaster Dwarkesh Patel, describing how the company plans to adopt and expand on OpenAI’s chip designs for its own infrastructure.
Related: Microsoft Releases Magentic Marketplace to Test AI Agent Safety
Under their revised partnership agreement, Microsoft has secured intellectual property rights to OpenAI’s chip designs and will maintain access to OpenAI’s AI models through 2032. The only thing off-limits? OpenAI’s upcoming consumer hardware products, which the ChatGPT maker wants to develop and sell independently.
This latest move highlights a growing truth in the AI race: building chips is hard and expensive. Even for a company the size of Microsoft, creating competitive AI silicon from scratch is a massive challenge. Instead of going solo, Microsoft is betting that tapping OpenAI’s expertise (and Broadcom’s production capabilities) will help it close the gap faster.
In an era where hardware defines who leads the AI pack, this partnership could be Microsoft’s most strategic shortcut yet. The question is, will borrowing brilliance be enough to catch up to Google’s Tensor and Amazon’s Trainium chips?
