OpenAI, the company behind ChatGPT, is reportedly set to begin large-scale production of its own artificial intelligence (AI) chips through a partnership with Broadcom (NASDAQ:AVGO).
Experts in the space see the move as a bid to cut reliance on chip giant NVIDIA (NASDAQ:NVDA) and ease the global shortage of processors driving platforms like ChatGPT.
The news came after Broadcom CEO Hock Tan told analysts in a September 4 call that the company had secured a fourth major customer that has committed to a US$10 billion order.
Tan said the deal would bring “immediate and fairly substantial demand” beginning next year. Although he did not name the customer, sources have since identified it as OpenAI. Both companies have declined to comment publicly.
OpenAI has long relied on NVIDIA’s graphics processing units (GPUs) for its AI needs, but securing enough GPUs has become increasingly difficult as demand for large-scale computing surges.
OpenAI CEO Sam Altman has been blunt about the challenge. In February, he wrote on X: “It is a giant, expensive model. We really wanted to launch it to plus and pro at the same time, but we’ve been growing a lot and are out of GPUs.”
Altman added at the time that OpenAI was preparing to add “tens of thousands of GPUs next week … (hundreds of thousands coming soon, and I’m pretty sure y’all will use every one we can rack up).”
However, shortfalls have slowed the rollout of new models, including GPT-4.5 earlier this year, and have pushed OpenAI to explore alternatives. The Broadcom deal marks its clearest step yet in building in-house hardware.
Broadcom has been developing custom processors it calls XPUs, or accelerated processing units, which are designed for specific high-intensity applications such as AI training. A person close to the project told the Financial Times that the new chips are intended to serve OpenAI internally rather than be sold to outside customers.
The prospect of custom silicon entering the AI infrastructure market has lifted Broadcom, which has climbed more than 30 percent this year. After Tan’s announcement, shares rose nearly 11 percent in New York trading.
For OpenAI, the investment is both a hedge and a necessity. The company has been racing to scale up computing capacity as businesses and consumers adopt ChatGPT in greater numbers. Last month, Altman said OpenAI planned to double its compute fleet “over the next 5 months” to meet demand from its newest model, GPT-5.
Alongside the Broadcom chips, OpenAI has made other major moves as well: it has signed a multibillion-dollar data center contract with Oracle (NYSE:ORCL) in July, struck a smaller cloud deal with Alphabet’s (NASDAQ:GOOGL) Google in June and announced its own data center initiative called Stargate at the beginning of the year.
The deal with Broadcom also comes as Altman has been working to frame OpenAI not only as a leader in cutting-edge research, but also as a driver of economic opportunity. In a message earlier this month, he said that while AI will be disruptive, it also has the potential to “unlock more opportunities for more people than any technology in history.”
He pointed to initiatives such as the OpenAI Academy and a new jobs platform designed to help workers and companies connect over AI skills, with the goal of certifying 10 million Americans in AI fluency by 2030.
Securities Disclosure: I, Giann Liguid, hold no direct investment interest in any company mentioned in this article.