Chip stocks slide after Broadcom boss likens AI investing to ‘casino’
Semiconductor shares fell sharply across Asia on Tuesday following comments by Broadcom's chief executive comparing many AI investors to "casino" gamblers, coupled with claims from Chinese rival CXMT about progress on new memory chip technology. The sell-off reflected mounting concerns that companies have been investing heavily in AI infrastructure based on unrealistic expectations rather than sound business fundamentals. Samsung Electronics shares dropped 7.4% and SK Hynix fell 10% on the Korean exchange, whilst South Korea's benchmark Kospi index tumbled 6.25%.
Broadcom chief executive Hock Tan said many investors "have no clue what they are investing in" and had set "very extreme expectations" only to discover they were essentially gambling. He noted that whilst OpenAI has annualised revenues of $50bn (£38bn), the company remains loss-making due to substantial spending on model training, with much of AI firms' current revenue coming from pilot projects rather than established, profitable operations.
- Semiconductor shares crash across Asia after Broadcom boss criticises AI investor expectations
- Chinese chipmaker CXMT claims technical progress that could challenge Samsung and SK Hynix
- OpenAI shows revenue growth but remains unprofitable due to high training costs
New here? Start with this
Investors have been pouring enormous amounts of money into artificial intelligence in recent years, betting that the technology will transform business and society. However, questions have emerged about whether this spending is based on realistic expectations or whether companies are investing heavily without fully understanding the risks. These doubts have begun to affect the computer chip industry, which manufactures the processors that power AI systems.
The concerns centre on a fundamental business problem: most AI companies are not yet profitable despite generating some revenue. A major chip manufacturer's chief executive recently highlighted that whilst an AI firm like OpenAI has substantial annual revenue, it continues to lose money because it spends heavily on developing and training its systems. Much of the income at present comes from small pilot projects and test deployments rather than established products that generate steady profits.
This matters because computer chip companies rely on sustained heavy investment from the technology industry to buy their products and build new factories. If technology companies have been investing in AI based on inflated expectations rather than sound business fundamentals, the demand for chips could eventually weaken, affecting an entire supply chain.
Both sides, in good faith
The strongest fair case each way — we don't pick a winner.
The case for
The critique identifies a genuine and serious risk in market dynamics. When investors concede they lack understanding of what they are funding whilst deploying enormous capital into AI infrastructure, this suggests speculative behaviour rather than disciplined investment. The evidence proves telling: OpenAI's reported $50bn in revenue accompanied by substantial losses indicates that current business models cannot sustain the vast capital deployment occurring across the sector. This pattern—extreme expectations, limited profitability, and heavy wagering on transformative potential—closely resembles bubble dynamics that historically precede market corrections. Prudent scepticism about whether these investments reflect genuine commercial viability rather than collective overconfidence is entirely reasonable.
The case against
Dismissing AI infrastructure investment as casino gambling conflates rational long-term strategy with speculation. Every genuinely transformative technology required substantial upfront capital before profitability—cloud computing, semiconductors, and the internet all faced identical scepticism in their early phases. OpenAI's losses reflect deliberate strategic choices to invest heavily in model development and competition, not fundamental business failure; the $50bn revenue itself demonstrates viable commercial demand exists. Pilot projects represent a natural adoption phase, not proof of market delusion. Sophisticated companies making these investments are conducting serious analysis of AI's economic potential, and characterising this as casino behaviour unfairly dismisses their reasoning. Transformative technologies cannot be built through conservative, immediately-profitable strategies.