Key Moments
- Broadcom Inc. CEO Hock Tan compared parts of the AI investment boom to gambling, saying many investors set extreme expectations without understanding large language models.
- Tan highlighted that OpenAI reached annualized revenue of $50 billion in September, yet is still losing money because of high model training costs.
- Industry leaders on the same panel warned about capital constraints and hardware shortages, with one noting that AI-related hardware could remain tight for two to three years.
Investor Behavior in the AI Boom
Broadcom Inc. Chief Executive Officer Hock Tan told an audience in Singapore on Monday that the surge in artificial intelligence investing has attracted market participants who behave more like gamblers than informed investors.
Tan said many investors are placing outsized bets on AI, particularly on large language models, without fully understanding the technologies or business models involved. He likened the environment to a casino, with investors chasing extreme outcomes while lacking a clear grasp of what they are funding.
Enterprises Seen ‘Burning Tokens’ on AI Experiments
The Broadcom CEO also addressed worries that corporations are pouring money into AI initiatives without clear commercial payoffs. To illustrate the point, he compared current enterprise experimentation with AI to children spending tokens at video arcades without gaining anything meaningful.
“Your parents give you $10 of tokens, you finish it in 10 minutes and you haven’t learned anything,” Tan said. “That’s what happening in enterprises today.”
AI Economics: Revenue Growth vs. Heavy Losses
Tan pointed to the financial performance of leading AI developers as evidence of both the promise and strain within the sector. He noted that AI companies such as ChatGPT-maker OpenAI are reporting billions of dollars in losses as they work to build sustainable business models that can support the high cost of advanced technology.
The sector requires substantial capital to fund advanced chips, data centers, and the development of increasingly complex models. Against that backdrop, Tan highlighted one key milestone: OpenAI reached annualized revenue of $50 billion in September, a figure he said indicates that a workable business model exists.
However, he added that OpenAI continues to lose money because of high training expenses tied to its frontier models. “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,” Tan said.
Capital Constraints and Market Repricing
Tan warned that a lack of sufficient capital could emerge as the next major brake on AI industry expansion. He cited the example of Australian data center operator Firmus Grid Ltd., which is backed by Nvidia Corp., withdrawing its planned $5.5 billion initial public offering last week.
Before Tan’s remarks, DBS Group Holdings Ltd. CEO Tan Su Shan told the same event that the market is now reassessing risk in AI-related financing in light of the heavy investment that has already flowed into the space.
| Company / Executive | Key Point | Figure / Event |
|---|---|---|
| Broadcom Inc. – Hock Tan | Warned of speculative, casino-like AI investing and questioned enterprise ROI | Commented on large language models and enterprise AI spending |
| OpenAI | Reported strong top-line trajectory but continued losses | Annualized revenue of $50 billion in September; still losing money |
| Firmus Grid Ltd. (backed by Nvidia Corp.) | Faced capital market pushback | Withdrew planned $5.5 billion IPO last week |
| DBS Group Holdings Ltd. – Tan Su Shan | Observed shifting risk assessment in AI financing | Noted the market is repricing risks after heavy investment |
| Intel Corp. – Lip-Bu Tan | Forecast persistent hardware tightness and potential slowdown in a key segment | Hardware shortages expected for two to three years; memory chips seen as early to lose momentum |
Hardware Bottlenecks and Segment Risks
Intel Corp. CEO Lip-Bu Tan, who participated in the same panel discussion, said the industry will continue to face hardware constraints as AI infrastructure is built out. He expects hardware to remain in short supply for the next two to three years, driven by rapid deployment of AI systems.
Looking across the semiconductor landscape, Lip-Bu Tan indicated that the memory chip market may be one of the first areas within the AI hardware ecosystem to see a loss of momentum as conditions evolve.
TradingPedia.com is a financial media specialized in providing daily news and education covering Forex, equities and commodities. Our academies for traders cover 
