OpenAI, the company behind ChatGPT, is moving to design its own artificial-intelligence chips rather than relying entirely on hardware from outside suppliers, according to coverage from Jon Peddie Research, The Globe and Mail, and Techerati.
According to The Globe and Mail, OpenAI is developing the custom chip in partnership with Broadcom, a major semiconductor firm. Jon Peddie Research frames the effort as OpenAI building "its own AI silicon," signaling that the company wants more control over the specialized processors that power its models.
Why would a software company want to make chips? The most powerful AI systems run on specialized processors, and those chips are expensive and, at times, in short supply. By designing a chip tailored to its own needs, OpenAI could reduce its dependence on a single supplier and better match the hardware to its software.
That raises an obvious question, and The Globe and Mail poses it directly: should Nvidia investors be worried? Nvidia's valuation rests heavily on demand for its AI chips, so any move by a major customer to build alternatives is closely watched.
Techerati describes the development as a signal of "a new phase in the AI infrastructure race," suggesting the competition among AI leaders is shifting from software and models toward the physical hardware and supply chains underneath them.
It is worth noting that designing a custom chip is a long, costly process, and these reports describe an effort still being built rather than a finished product on the market.
Why it matters: control over chips is becoming as strategic as the AI models themselves, and OpenAI's push to design its own silicon could reshape the balance of power among the companies racing to lead artificial intelligence.