Broadcom, Intel CEOs say capital, not chips, may be AI's biggest bottleneck

The heads of two of the world's biggest chip companies said on Monday, October 12, 2026, that money, not chips, may now be the tightest limit on the growth of artificial intelligence. Broadcom chief executive Hock Tan and Intel chief executive Lip-Bu Tan were speaking on a panel at the DBS Asia Future Forward Conference in Singapore, moderated by Tan Su Shan of DBS. Hock Tan said financing could become a bigger obstacle to AI expansion than shortages of semiconductors, memory chips or electricity. Lip-Bu Tan made a similar point, saying: "If there's any challenge in the front, it's capital and the cost of capital."

For the past two years the story of the AI boom has been one of physical shortages. Companies could not get enough advanced chips, enough high-bandwidth memory or enough electricity to run new data centres. Chipmakers such as Broadcom and Intel sit at the centre of that supply chain, so when their leaders say the main constraint is shifting, it is worth noting.

Their argument is straightforward. Building AI computing capacity costs enormous sums, and much of that money is borrowed or raised from investors who expect a return. If interest rates stay high, or if investors begin to doubt that AI products will earn enough to repay the outlay, the flow of money slows. In that case, projects could be delayed even when chips and power are available. Hock Tan balanced this with optimism, saying demand still exceeds supply. But he also named the two risks that could change the picture: weaker than expected adoption, and a slowdown in how fast models improve.

Lip-Bu Tan's comment adds a second layer. It is not only whether capital is available, but what it costs. Higher financing costs make every data centre and every AI project harder to justify. That pushes companies to prove that AI is saving money or creating new revenue, rather than spending on it because competitors are.

For India, the warning is relevant. The country is trying to build its own AI computing base, through government programmes that give start-ups and researchers access to graphics chips and through large private data centre projects. Capital in India generally costs more than in the United States or Singapore. If global investors become more careful about AI spending, Indian projects that depend on foreign money may face sharper questions on returns. At the same time, a steady demand picture into 2029, as Hock Tan described it, suggests the global need for computing capacity, including capacity in India, is not expected to fade soon.

Short-term: Investors and lenders are likely to look more closely at whether AI projects can pay for themselves. Companies raising money for data centres may face tougher questions on returns and financing terms.

Long-term: If capital becomes the main limit, the pace of AI expansion may depend more on interest rates and investor confidence than on factory output. Companies that can show clear earnings from AI are likely to find funding easier than those that cannot.

Who is affected: Chipmakers, cloud and data centre companies, AI labs that rely on outside funding, banks and investors who finance AI infrastructure, and countries such as India that are building national AI computing capacity.

Questions and Answers