Broadcom CEO says funding will be the company’s biggest AI-growth constraint
Hock Tan paired confidence in AI assistants’ revenue potential with criticism of speculative investors, costly model training and enterprise experiments without clear returns.
Listen to this story
The audio brief
Story brief
3 key pointsTan's remarks shift the AI-growth question from chip supply to the ability to finance expansion: he says Broadcom is working through wafer and memory shortages, but capital could become its largest constraint, with implications for the wider industry. He also distinguishes demand from durable economics, citing OpenAI's $50 billion in annualized revenue in September while noting that training costs kept it loss-making. Tan credited OpenAI and Anthropic with demonstrating revenue potential for AI assistants, but did not call that proof of sustainable profitability.
- 01
Tan spoke at a Singapore panel on October 12, 2026, and warned that insufficient capital could constrain the wider AI industry.
- 02
He likened some AI investors to gamblers who set extreme expectations without understanding what they are backing, particularly large language models.
- 03
Tan compared companies' AI experiments without clear returns to quickly spending arcade tokens without learning anything.
Broadcom is overcoming wafer and memory shortages, but CEO Hock Tan expects funding to become its biggest constraint as AI development continues. Speaking at a Singapore panel on October 12, 2026, Tan put capital ahead of those hardware bottlenecks in his assessment of the company’s growth prospects. He also warned that insufficient capital could constrain the wider AI industry.
A casino for investors, an arcade for businesses
Tan also challenged the expectations behind AI spending. According to Investing.com’s account, carried by Yahoo Finance, he compared the investment climate to a casino and described some investors as gamblers. Many investors, he said, set extreme expectations without understanding what they were backing, particularly large language models.
His criticism extended beyond investors buying into the boom. Tan compared businesses experimenting with AI without clear returns to children rapidly spending arcade tokens and learning nothing. The analogy targeted the value companies get from their experiments, rather than the ability of AI providers to sell them technology.
Your parents give you $10 of tokens, you finish it in 10 minutes and you haven't learned anything.
Hock Tan, Broadcom CEO, on enterprise AI experimentation, as quoted by Investing.com
Revenue potential is not profitability
Tan nevertheless credited Anthropic and OpenAI with demonstrating significant revenue from their latest models deployed as AI personal assistants. His assessment acknowledged demand for those products, even as he warned about the capital needed to support continued growth.
For OpenAI, Tan cited $50 billion in annualized revenue in September, according to Investing.com. He said high training costs kept the company loss-making. He took the revenue as a sign that a viable business model could exist, but stopped short of treating it as proof of sustainable economics. His qualified conclusion was that providing advanced generative AI models offered substantial potential for revenue growth.
Sources
- rmb.reuters.comBroadcom CEO says biggest hurdle in AI-fuelled tech growth is capital, not memory chips
- finance.yahoo.comBroadcom CEO says many AI investors are gamblers
Reader comments
Newest comments first. Replies stay oldest first.