Broadcom has ridden the AI surge to about $68 billion in AI semiconductor revenue since late 2023, according to its earnings reports. In its fiscal third quarter, which ended Aug. 2, AI semiconductor revenue more than tripled from a year earlier to $16.7 billion.
That boom has made Broadcom one of the clearest beneficiaries of the race to build out data centers and frontier AI systems. Hyperscalers are pouring billions into those facilities, and Broadcom designs custom AI chips for them.
But CEO Hock Tan is sounding a warning about the market behind the boom. On Monday, he said many investors chasing AI have 'no clue' what they are investing in, according to Bloomberg. Speaking at the DBS Asia Future Forward Conference in Singapore, he said they 'set very extreme expectations only to find out that it’s almost like they are in a casino.'
Tan’s concern is not limited to investors buying stocks. He also questioned the economics of the AI labs themselves. Broadcom sells the hardware frontier AI labs need, but those labs still have to find enough paying customers to cover the cost of developing and running their models.
OpenAI’s reported $50 billion in annualized revenue at the end of September suggests there is real demand, Tan said, but he warned that figure does not prove profitability. Annualized revenue measures the current sales pace extrapolated over a year, not cash collected over 12 months. OpenAI expects to spend over $25 billion on training alone this year.
'For now it’s still losing money because they’re spending a lot on training,' Tan said. 'So frankly, it doesn’t tell anything at this point except that there is a lot of potential revenue growth out of providing frontier generative AI models.'
He also said a shortage of capital could become the industry's biggest obstacle, more than any shortage of memory, semiconductors or electricity. Some analysts expect AI capital expenditure growth to slow next year. Investors are already signaling more caution: Firmus, an Australian data center company backed by Nvidia, withdrew its planned IPO on Oct. 9 after weak investor demand for its marketed share prices, according to Bloomberg.
Tan extended the critique to enterprises spending on AI as well. He said companies are 'wasting money' on tokens to experiment with AI instead of figuring out where it produces returns. He compared that to children feeding arcade tokens into games and ending up with nothing to show for it.
The market has already voted on the basic direction of AI. Capital is flowing into chips, data centers and model training at a massive pace, and Broadcom is collecting the fees. But Tan’s comments make the central question harder to avoid: whether the current spending spree is building durable earnings, or just inflating a very expensive bill for investors and the taxpayer-adjacent economy of cheap money and easy capital.
For free enterprise, the lesson is simple. Innovation rewards risk when returns are real, not when expectations outrun economics. If AI is going to justify this scale of spending, it will need customers, profits and discipline, not just hype and leverage.
