Broadcom’s top chip executive on Monday dismissed speculation that OpenAI was the company’s unnamed $10 billion customer, clarifying widespread assumptions that emerged after Broadcom’s September earnings call.
Charlie Kawwas, president of the semiconductor solutions group at Broadcom, said during a joint appearance with OpenAI President Greg Brockman on CNBC’s Squawk on the Street that the artificial intelligence startup was not behind the record breaking order.
The two executives were on air to discuss their newly announced collaboration to build and deploy 10 gigawatts of custom AI accelerators.
“I would love to take a $10 billion [purchase order] from my good friend Greg,” Kawwas said. “He has not given me that PO yet.”
The clarification came shortly after analysts and investors had linked the mystery order to OpenAI, following Broadcom’s earlier disclosure of a massive new AI chip contract from an undisclosed “large customer.”
Broadcom, one of the world’s leading semiconductor companies, has been rapidly expanding its presence in artificial intelligence hardware as global demand for AI infrastructure continues to surge.
During its quarterly earnings call in September, Broadcom CEO Hock Tan revealed that a single “large customer” had placed a $10 billion order for custom AI chips an announcement that sent analysts speculating about which tech giant might be behind the deal.
At the time, most industry watchers assumed the buyer was OpenAI, given the company’s aggressive expansion in AI compute infrastructure and its growing partnerships with chipmakers.
OpenAI, valued at roughly $500 billion, has been on what many have called an “AI infrastructure buying spree,” signing major supply agreements with Advanced Micro Devices (AMD), Nvidia, and cloud computing firm CoreWeave in recent months.
Broadcom’s partnership with OpenAI, announced Monday, involves the joint design and deployment of specialized hardware optimized for frontier AI models.
The companies said they have been collaborating for more than 18 months and plan to begin rolling out racks of custom designed chips by late next year.
“By designing its own chips and systems, OpenAI can embed what it’s learned from developing frontier models directly into the hardware, unlocking new levels of capability and intelligence,” Broadcom said in a statement.
Industry analysts say Broadcom’s clarification underscores the complexity of the AI supply chain and the growing diversity of major customers in the semiconductor sector.
“While OpenAI is an influential player in AI development, Broadcom’s biggest chip contracts likely come from hyperscale cloud providers with massive data center footprints,” said Dr. Anita Sharma, a semiconductor industry analyst at TechMetrics Research.
“Companies like Google, Meta, and ByteDance have both the financial and technical capacity to commit to multi billion dollar custom chip orders.”
Others noted that the confusion was understandable, given the timing of Broadcom’s announcement and OpenAI’s expanding infrastructure needs.
“When Broadcom mentioned a $10 billion AI chip order and OpenAI’s name came up days later in partnership talks, it was easy to connect the dots,” said James Li, senior analyst at Semiconductor Insights.
“But this statement from Broadcom makes it clear that the mystery customer is likely one of the major cloud platforms. Broadcom has not publicly disclosed the identity of the fourth large web scale customer cited by Tan, though analysts widely believe it could be a major tech firm already known for investing heavily in AI hardware.
Broadcom’s AI related revenue has surged in recent quarters, with Tan noting that demand for custom accelerators and networking components is driving new growth.
The company forecast a substantial increase in AI chip sales in 2025, boosted in part by the $10 billion order announced in September.
According to market research firm TrendForce, Broadcom’s share of the global AI chip infrastructure market has risen steadily, trailing behind only Nvidia and AMD in the custom silicon category.
The company supplies components to several major data center operators, including Google, Meta, and ByteDance. Meanwhile, OpenAI continues to scale its compute resources amid growing demand for generative AI products like ChatGPT and enterprise solutions such as GPT Enterprise.
The company’s collaboration with multiple hardware providers, including Broadcom, suggests a strategy aimed at diversifying its chip supply to reduce dependence on Nvidia, which remains the dominant player in the AI GPU market.
Industry professionals say Broadcom’s disclosure provides important clarity for investors and engineers navigating the fast changing AI hardware landscape.
“Transparency matters, especially when the market is this volatile,” said Arjun Desai, a data center engineer based in Austin, Texas. “When you hear about a $10 billion chip deal, it can move stock prices and influence where companies invest next.”
Others see the partnership between Broadcom and OpenAI as a sign of how tightly intertwined hardware and software development have become in the AI race.
“Custom silicon is the next frontier,” said Laura Kim, head of AI systems at a San Francisco based research lab. “The closer hardware gets to model design, the faster AI can evolve and partnerships like this are how that happens.”
Broadcom and OpenAI plan to begin deploying their co-developed chips late next year, a timeline that could coincide with the rollout of OpenAI’s next generation of frontier models. The collaboration aims to improve performance efficiency and reduce costs across large scale AI training operations.
For Broadcom, continued expansion into AI infrastructure could provide a buffer against cyclical downturns in its traditional networking and wireless businesses.
For OpenAI, the ability to influence chip design could yield competitive advantages in model speed, power efficiency, and scalability.
“Custom hardware is becoming as important as the models themselves,” said Dr. Sharma. “The companies that integrate both layers silicon and software will define the next era of AI computing.”
Despite speculation, Broadcom’s $10 billion mystery customer remains undisclosed, and OpenAI is not the source of the massive order. Yet, both companies continue to deepen their partnership, signaling a future where AI innovation increasingly depends on tailor-made hardware solutions.
As Broadcom strengthens its foothold in the AI semiconductor market and OpenAI expands its infrastructure network, the line between chipmaker and AI developer grows ever thinner reshaping how the next wave of artificial intelligence will be built and deployed.
