Broadcom CEO: Capital, Not Memory Chips, Is the Biggest Bottleneck for AI Growth
nashnova research
Broadcom CEO Hock Tan said the biggest constraint on AI expansion is shifting from wafer and memory shortages to capital itself — a sign that chip-supply pressure is easing, but the ability to raise money will decide who keeps building.
Why did Broadcom's "biggest bottleneck" shift from chips to cash?
Hock Tan said at a Singapore industry forum on October 12 that Broadcom is overcoming wafer-supply tightness and memory shortages.
Yet he identified capital — not hardware supply — as the most critical constraint in the next phase.
This means → the pressure point for AI infrastructure is migrating: from "can't buy the chips" to "can't raise enough money to build."
Is the memory-chip supply crunch really easing?
Markets had been most worried about shortages of wafers — the base material for chips — and memory — the chips that temporarily store data while AI models run.
Tan's statement signals that Broadcom sees these supply bottlenecks being gradually resolved.
In plain terms = before, money alone couldn't get you the parts; now the parts can arrive, but the money may run short.
Can AI actually generate revenue?
Tan noted that Anthropic and OpenAI have shown their latest AI personal-assistant models can produce meaningful revenue.
That provides demand-side support for continued AI compute investment — someone is paying, so there is a return to chase.
This reflects a broader shift: AI is moving from "burn cash and tell a story" to "prove the spend with product revenue."
市场有风险，内容仅供研究参考，不构成投资建议。