What Was Said In Singapore

Broadcom chief executive Hock Tan said on Monday 12 October that financing could become a bigger obstacle to AI expansion than shortages of chips, memory or electricity. He was speaking at the DBS Asia Future Forward Conference in Singapore. CNBC reported him saying the biggest constraint might be capital.

Intel chief executive Lip-Bu Tan echoed the concern. If there is a challenge up front, he said, it is capital and the cost of capital. Over the longer run, people will want to see a return on the money spent.

Both stayed bullish on demand. Hock Tan said there is more demand for compute infrastructure than the industry can supply. Lip-Bu Tan did not expect a significant slowdown over the next two to three years. CNBC ties the remarks to rising borrowing costs: higher bond yields can make data centre construction and chip purchases dearer, on projects that may take years to pay back.

The Constraints And Their Stated Horizons

Hock Tan gave each limit a different end date. Capital was the only one he left open.

ConstraintHock Tan's view on 12 October
CapitalMay be the biggest constraint today, no end date given
MemoryToday's shortage may not persist a year from now
WafersShortages could ease by 2028 or 2029
PowerAvailability could improve as new infrastructure comes online
Compute demandExceeds supply, possibly through 2029 and into 2030

That ordering is Tan's own, drawn from one conference appearance and not from a forecast model. He also cautioned that demand could slow if adoption of advanced AI models disappoints or their capabilities stop improving. Both executives said growth beyond the next few years depends on businesses seeing real economic benefit. Lip-Bu Tan expects companies to prove returns through efficiency and new business models.

How Financing Strain Reaches A Buyer

When capacity is built with borrowed money, the price of that money ends up inside the price of renting the capacity. That is our reading of the two CEOs' point. Neither of them said it in those words.

A vendor paying more for money can raise prices at renewal or ask for longer commitments to reassure its own lenders. Both are plausible while demand exceeds supply, because the vendor then holds the stronger hand.

Hock Tan's other remarks show the other side of the ledger. Bloomberg reported that he likened many AI investors to casino gamblers, said some have little understanding of what they are betting on, questioned whether businesses are wasting money on AI initiatives and doubted the longer-term economics of money-losing AI labs. Investing.com adds his picture of enterprises spending $10 of arcade tokens in ten minutes and learning nothing and his note that OpenAI reached $50 billion of annualised revenue in September while still losing money on training costs.

Questions To Put To An AI Vendor Before You Commit

Ask the vendor how its capacity is financed and note how specific the answer is. Equity, long-term debt, customer prepayments and short-term borrowing each carry a different risk for you.

Then ask in writing what happens to your price if its borrowing costs rise, whether the contract allows repricing mid-term and what commitment length it expects in return for guaranteed capacity.

On memory, Tan's view is that today's shortage may be gone in a year. That is one executive's opinion and not a forecast of retail prices, so leave the memory line in your hardware budget as it stands and revisit it when contract prices actually move.