It was the latest move for OpenAI in its effort to build a homegrown stack to support its growing range of AI models, products and hardware. It could also reduce OpenAI’s dependence on semiconductors from Nvidia, as more tech companies peel off to design their own chips that challenge the dominance of the most valuable publicly-traded company.
OpenAI President and co-founder Greg Brockman said the Jalapeño chip would pave the way for quicker deployment of reliable AI that’s more affordable for consumers and businesses.
“By designing more of the stack ourselves, we can serve more intelligence with greater efficiency and keep pushing advanced AI toward broader access,” Brockman said in a news release.
Last October, OpenAI inked a deal with the California-based Broadcom chipmaker to partner in developing in-house chips that could power its slate of products like ChatGPT. OpenAI said at the time it planned to use enough homegrown chips to consume ten gigawatts of electricity, enough energy to power millions of homes. To better understand the scale of energy consumption in play: Each nuclear reactor, for example, produces roughly one gigawatt of electricity, per the Department of Energy.
Broadcom President and CEO Hock Tan said the partnership with OpenAI would be a critical piece in building out the physical infrastructure to sustain AI for the next decade.
“By co-developing our industry-leading silicon directly with OpenAI, we are enabling the deployment of gigawatt scale data centers with Microsoft and other partners beginning in 2026,” Tan said. The announcement didn’t list a release date for the Jalapeño chip.
OpenAI’s dependence on Nvidia
Late last year, Nvidia said it was exploring a $100 billion megadeal to invest into OpenAI and allow the ChatGPT-maker to use its semiconductor chips within its expanding fleet of data centers. The preliminary agreement was one part of tech companies’ sprint to spend hundreds of billions on data centers.
It was never finalized. The Wall Street Journal reported in January the deal between the pair of tech behemoths was on ice for the time being.
OpenAI CEO Sam Altman defended the strength of the companies’ relationship. “We love working with NVIDIA and they make the best AI chips in the world. We hope to be a gigantic customer for a very long time,” Altman wrote on X in early February.
