In Singapore at the DBS Asia Future Forward Conference, Broadcom chief Hock Tan argued the tightest pinch point in the AI boom may be money itself. "[What] I think might be the biggest constraint is capital," he said. Intel CEO Lip-Bu Tan backed that up, zeroing in on the price of funding and the need to prove payback: "If there's any challenge in the front, it's capital and the cost of capital, and then the other part is in the longer run, people want to see the ROI."
Tan at Broadcom said the industry has run into one snag after another over the past two years, but he anticipates several of today's constraints will loosen as capacity expands. "Memory shortage today might not be that way a year from now," he noted, and he suggested wafer shortages could abate by 2028 or 2029. He added that the supply of power should get better as additional infrastructure is brought online.
For now, though, demand is still ahead of what suppliers can deliver. "There is more demand for compute infrastructure than we can supply," Hock Tan said, adding that the gap could last through 2029 and potentially into 2030.
Capital, not silicon, is becoming the real constraint on AI build-out. Market Briefs covers that shift free every weekday.
The executives' cautions are landing at a moment when climbing borrowing costs could make financing large AI buildouts more complicated. An increase in bond yields could raise the expense of financing data center builds and chip acquisitions, which might put strain on projects that could take years to show returns. Even so, Intel's Tan said he does not expect a pronounced slowdown in the next two to three years, pointing to broader adoption of AI and sustained demand for computing power.
The near term looks sturdy, but the longer runway depends on results. Broadcom's Tan warned that demand for computing infrastructure could cool if businesses are disappointed by advanced models or if capability gains start to level off. Lip-Bu Tan emphasized that companies will increasingly need to show concrete payoffs from AI through efficiency improvements and new business models. For your wallet, the signal is clear: keep an eye on financing costs, how quickly new power and wafer capacity arrives, and whether enterprises can turn AI spending into real-world returns.
When chip executives say funding is the bottleneck, listen. Join Market Briefs free and follow it.
