Broadcom CEO warns AI capital constraints could outlast chip shortages

Broadcom chief executive Hock Tan has warned that financial constraints, rather than hardware shortages, may become the primary bottleneck for artificial intelligence development. Speaking at the DBS Asia Future Forward Conference in Singapore, Tan suggested that the current investment climate resembles a casino, with many participants lacking clarity on the underlying economics of the technology they are funding.

According to Bloomberg, Tan stated that many investors have no clear understanding of what they are backing. He observed that these stakeholders set extreme expectations, only to discover that the reality of the market feels akin to gambling. This sentiment was echoed during a panel discussion on the future of AI funding, where Tan shared the stage with Intel chief executive Lip-Bu Tan.

What we know

Tan argued that most enterprises have yet to identify AI applications that deliver a clear financial return. He described the current spending habits of many companies as wasteful, comparing them to a child spending money quickly at an arcade without learning anything. According to The Business Times, he noted that businesses are simply burning through tokens in an attempt to understand the technology, rather than deriving tangible value from it.

The executive highlighted the financial situation of OpenAI as an example of the sector’s economic challenges. While OpenAI’s annualised revenue reached $50bn in September, Tan noted that this figure alone does not indicate long-term profitability. The company remains loss-making due to heavy spending on model training, suggesting that high revenue does not equate to sustainable economics for AI labs.

Both Tan and Intel’s Lip-Bu Tan agreed that capital could become a more significant constraint than chips, memory, or power. CNBC reported that Tan identified capital and the cost of capital as the biggest challenges facing the industry. Lip-Bu Tan added that investors will eventually demand a return on their investment, creating pressure on the financial models of AI developers.

Despite these concerns, demand for computing infrastructure continues to outstrip supply. Tan expects this imbalance to persist through 2029 and possibly into 2030. He anticipates that wafer shortages will ease by 2028 or 2029. Intel’s Lip-Bu Tan stated he saw no significant slowdown in demand over the next two to three years. Broadcom is raising $60bn in debt to help customers like Anthropic pay for AI chips and infrastructure.

Lip-Bu Tan also commented on the rise of Chinese frontier models such as DeepSeek, Kimi, and GLM. According to Reuters, he described these models as “good enough” and noted that their lower token costs have driven rapid adoption. This competitive pressure adds another layer of complexity to the global AI landscape.

The remarks sparked immediate market reaction. Chip stocks fell in Asia on Tuesday following the comments. Elon Musk responded on X, writing “No crying in the casino” in reply to a post about Tan’s remarks. The exchange highlights the tension between speculative enthusiasm and economic reality in the AI sector.

Defender's checklist

• Review current AI spending against measurable business outcomes to ensure ROI.
• Monitor capital allocation strategies for AI infrastructure investments.
• Assess the sustainability of vendor financial models and pricing structures.
• Evaluate alternative AI models that offer lower token costs and comparable performance.

What to do and how to stay safe: Broadcom

• Audit your organisation’s AI experiments to identify projects with no clear path to revenue or efficiency gains.
• Establish strict budget caps for AI token usage and infrastructure spending to prevent uncontrolled costs.
• Diversify your AI vendor portfolio to mitigate risk associated with any single provider’s financial instability.
• Once the vendor provides an update on long-term pricing models, reassess your total cost of ownership for AI services.

Step-by-step guide: AI Governance for Small Teams: Practical Controls and Cost

General security guidance from the TechWire Beat newsroom. It is not confirmed advice from the organisations named in this story.

Frequently asked questions

Why does Hock Tan compare AI investors to gamblers?

He argues that many investors lack understanding of what they are backing and set unrealistic expectations, similar to casino gambling.

What is the biggest constraint facing AI development according to Tan?

Tan believes that capital and the cost of capital will be the biggest constraints, surpassing hardware or power issues.

When does Tan expect wafer shortages to ease?

He expects wafer shortages to ease by 2028 or 2029, though infrastructure demand may outstrip supply until 2030.