The chief executives of Intel and Broadcom have identified capital — not compute — as the single biggest constraint on the growth of artificial intelligence.

In a rare point of agreement between two of the semiconductor industry's most prominent leaders, both executives said the cost of financing AI buildouts and the pressure to demonstrate returns on those investments now pose a greater risk to the sector's expansion than shortages of chips or electricity.

The assessment marks a shift in how the industry frames its bottlenecks. For the past two years, public discussion of AI's limits has centred on physical constraints: advanced chips, fabrication capacity and grid power. The two CEOs are now pointing to a financial constraint instead.

What does the warning actually say?

Both executives named two specific pressure points:

• Financing costs. The expense of raising and deploying capital for AI infrastructure has risen to the top of the risk register.
• Return on investment. Companies pouring money into AI capacity face mounting demands to show that the spending generates measurable returns.

Their joint view is that these two financial factors outweigh supply-side concerns. Chips and power, the classic constraints of the AI hardware race, now rank as lesser risks in their assessment.

Why does the bottleneck matter for decision-makers?

The signal from two semiconductor CEOs carries weight because both companies sit at the centre of the AI supply chain. Intel manufactures processors; Broadcom supplies chips and networking technology that underpin large-scale AI infrastructure. When leaders at that level of the stack say capital is the binding constraint, they are describing the economics their own customers face.

The warning reframes the question boards and investors are asking about AI. The issue is no longer only whether the industry can produce enough hardware. It is whether the economics of AI investment hold up under current financing conditions — and whether deployed capital earns its keep.

What changes if capital is the constraint?

If chip and power shortages are secondary risks, then the pace of AI growth depends on financial discipline rather than physical supply. That places the burden on companies committing capital to AI projects: they must show returns, not merely secure capacity.

It also implies a different set of watch-points for the industry. Interest-rate conditions, the cost of capital and the willingness of investors to fund successive rounds of AI infrastructure spending become the variables that determine how fast the sector scales.

The bottom line

Intel and Broadcom's chief executives have moved the AI constraint question from the fab and the power plant to the balance sheet. Their stated position is direct: financing costs and the need to prove returns on AI investment represent a greater risk to the industry's growth than any shortage of silicon or electricity.

For executives allocating AI budgets, the message is to treat capital efficiency as the primary planning assumption — and to expect investors to demand evidence of returns sooner rather than later.