Two of the biggest names in AI chips are selling into the same boom but describing it very differently

Broadcom CEO Hock Tan has a message for AI investors. Many of them, in his view, are acting less like analysts and more like gamblers in a casino.

That is a striking thing to hear from a man whose company just posted a 221% jump in AI chip revenue. It is even more striking next to Nvidia CEO Jensen Huang, who has spent the same stretch telling anyone who will listen that the AI build-out is only getting started.

Tan made his remarks on October 12, 2026, at the DBS Asia Future Forward Conference in Singapore. His core argument was that many investors do not fully understand the technology or the business models they are betting on.

He also warned that returns are uncertain for enterprises jumping into generative AI. Training large AI models is enormously expensive, and that bill has to be paid long before any profit shows up.

To make the point, Tan cited OpenAI . He put the company’s annualized revenue run-rate at approximately $50 billion, which he framed as clear evidence that demand is real. The catch, in his telling, is that demand and profitability are two different things. Operational costs, especially training, could still produce heavy losses.

Tan then flagged a constraint that has not gotten much airtime. As supply chain bottlenecks ease, he suggested that access to capital and the cost of that capital could become the next limiting factor for AI infrastructure.

Broadcom reported AI semiconductor revenue of $16.7 billion in its fiscal third quarter, up 221% from a year earlier. The company guided for $21.7 billion in AI semiconductor revenue in the fourth quarter.

Daily. Free. Join 34,000+ readers across crypto, finance, and policy.

Tan has set Broadcom’s AI revenue targets at $115 billion for fiscal 2027 and $230 billion for fiscal 2028, which would mean doubling in a single year.

Tan’s comments led to slight declines in some chip stocks, according to the research summary of the episode.

Jensen Huang has argued that spending on AI infrastructure will reach trillions of dollars before the end of the decade. He has also said that every company will eventually become an AI company.

Huang has brushed off concerns about a bubble or a slowdown in the sector, and maintained that the US will keep its leadership position as the AI landscape evolves.

A Bloomberg newsletter dated October 16, 2026, laid out the split between the two executives, contrasting how each views investor worries and the durability of the AI boom.

Both companies are benefiting from heavy demand from major AI developers, including OpenAI and Anthropic . Both CEOs have reaffirmed that demand for AI infrastructure and semiconductors should persist through at least 2028. Where they diverge is on what happens to the people placing those orders.

Huang’s framing treats AI spending as an inevitable, economy-wide transition. Tan’s framing treats it as a real but expensive bet, one where the payoff for buyers is far from guaranteed.