OpenAI is getting into the chip business. The Sam Altman-led startup has unveiled Jalapeño, a custom processor built for AI "inference" — the work of actually running AI models to answer user queries, as opposed to training them.
According to OpenAI, the company designed the chip from scratch in just nine months. It did not do so alone. As reported by MSN, Broadcom handled the silicon implementation, networking, and connectivity, while Celestica developed the physical computing hardware that houses it.
The spicy codename aside, the move carries real strategic weight. TechCrunch frames Jalapeño as "Big Tech's spiciest move away from Nvidia," the company that has dominated the AI chip market for years. By designing its own silicon, OpenAI joins a growing club — TechCrunch lists Google, Apple, and SpaceX — of large technology firms building their way out of total dependence on Nvidia's processors.
Why build your own chip? Custom processors can be tuned for a company's specific workloads, potentially cutting costs and easing reliance on a single supplier whose chips are expensive and in fierce demand. The question of what this means for Nvidia is already being asked directly: Moneywise published an analysis titled "OpenAI's new chip and what it means for Nvidia."
None of the sources here detail Jalapeño's performance, pricing, or when it will be widely deployed — only that OpenAI has announced its plans and named its hardware partners.
Why it matters: If one of the world's most prominent AI companies can credibly design its own inference chip in well under a year, it signals that Nvidia's grip on the AI hardware market — long treated as unshakable — may finally be loosening.