OpenAI unveiled Jalapeño, its first proprietary AI inference chip co-developed with Broadcom, positioned as competitive with Nvidia Blackwell and Google TPU. Chip targets gigawatt-scale data centers starting 2026.
OpenAI has entered the AI chip market in earnest, unveiling a custom AI semiconductor developed over the past year. OpenAI and Broadcom jointly revealed the chip "Jalapeño" on Wednesday, with plans to deploy it in data centers later this year. Broadcom CEO Hock Tan and President Charlie Kawas handed a prototype directly to OpenAI CEO Sam Altman and President Greg Brockman.
The companies note that final performance evaluation is ongoing, but early tests show Jalapeño outperforms current state-of-the-art AI chips in performance per watt. Notably, the chip advanced from initial design to tape-out—when the design transfers to a foundry for production—in just nine months, the fastest development cycle ever recorded for a custom application-specific integrated circuit (ASIC). OpenAI's own AI models powered the accelerated design and optimization process throughout.
Tan told Reuters that Jalapeño "is on par with Nvidia's Blackwell chip and Google's tensor processing units (TPUs)." TSMC in Taiwan will handle mass production, while Samsung Electronics and SK Hynix are supplying memory chips to Broadcom.
The two companies have mapped a roadmap for future chips, Bloomberg reported. The next version is slated for 2028, with new chips following annually. While Jalapeño focuses on inference, Tan said future variants may address other areas.
"The world is moving toward a compute-based economy," said Brockman. "Jalapeño will make compute more abundant, enabling faster, more reliable and more affordable AI for individuals and businesses."
OpenAI, widely regarded as Nvidia's largest GPU customer, now faces competitive pressure in the AI accelerator market with its own inference chip. Google pioneered this path with TPUs, and OpenAI has now followed suit, while Anthropic explores chip development as well. All three leading AI model companies are working to reduce reliance on Nvidia.
Nvidia's share price fell more than 1 percent from the previous close by 1:50 p.m. Eastern time Wednesday, trading around $197 during the session.