Intel and Broadcom CEOs warn: capital, not chips, may slow AI growth

The Quartz article, written by Cris Tolomia and updated on October 12, 2026, reports remarks by the heads of Broadcom and Intel on what could be the biggest brake on AI growth. According to the article, which in turn cites CNBC, both believe financing matters more than shortages of semiconductors, memory or electricity. What follows is based solely on the text received.

Broadcom CEO Hock Tan and Intel CEO Lip-Bu Tan spoke at the DBS Asia Future Forward conference in Singapore. Hock Tan said that "what I think could be the biggest constraint is capital." Lip-Bu Tan agreed, pointing to borrowing costs and the pressure on companies to show returns. According to the quote reported, the immediate challenge is capital and what it costs, and over the longer term people will want to see a return on investment (ROI).

Despite those concerns, both agree that demand for AI computing infrastructure remains strong in the short term. Hock Tan said demand for compute capacity exceeds what the industry can supply, and that the gap could last until 2029 and possibly 2030. Lip-Bu Tan said a significant slowdown in the next two or three years is unlikely, given the growing adoption of AI applications.

Beyond that horizon, both see less certainty. Hock Tan warned that spending on compute infrastructure could slow if adoption of the most advanced AI models falls short of expectations or if the pace of improvement in their capabilities begins to level off. Lip-Bu Tan added that companies will have to show tangible returns, whether through efficiency gains or new business models.

Hock Tan also said some of the supply constraints of the past two years should ease. He expects wafer limitations to loosen around 2028 or 2029, and noted that access to electricity will expand as new infrastructure comes online.

The article notes that both companies have raised large sums to finance AI infrastructure. Broadcom is arranging $60 billion in debt to help Anthropic and other companies access AI chips and infrastructure. Banks are preparing a $42 billion senior secured tranche, and Blackstone is leading an $18 billion subordinated tranche. Under a loan commitment disclosed in Anthropic's IPO prospectus, Broadcom agreed to lend it up to $42 billion to cover infrastructure spending, an amount that could convert into Anthropic equity. Anthropic is expected to be Broadcom's largest compute customer in 2027.

Intel, for its part, expanded a common stock offering in August to $20,000,000,000, with 210,526,315 shares at $95 each, to fund the expansion of its contract chipmaking business. In July it raised its 2026 capital expenditure target to $20 billion from $18 billion, and committed to reaching high-volume production on its 14A process node in 2028.

Taken together, the message is that AI's bottleneck could shift from the physical (chips, wafers, power) to the financial: who can fund the infrastructure, and how long investors will keep demanding proof of profitability. The text is short and sticks to the remarks and figures cited; it offers no further analysis.