Particle.news

CEOs Warn Capital Crunch Could Become AI Buildout’s Main Bottleneck

Rising borrowing costs are prompting a shift toward large lending and leasing platforms to fund data‑centres and frontier model builds.

Overview

  • Broadcom chief Hock Tan and Intel boss Lip‑Bu Tan said on Monday that access to capital and the cost of financing may soon limit AI infrastructure growth more than chip or memory shortages.
  • Broadcom is already moving to underwrite that risk by arranging large financing vehicles with private equity and banks, with reports of a $35 billion platform tied to Apollo and Blackstone and talks of multi‑billion syndicated debt for major model developers.
  • Investors have shown sensitivity to private revenue signals after a Financial Times figure put OpenAI’s annualised run rate near $50 billion, and chip and data‑centre stocks dropped as concern over funding and customer concentration rose.
  • Analysts offer competing near‑term views: UBS forecasts sharply faster hyperscaler cloud revenue growth while Goldman Sachs warns grid and high‑voltage power availability could become the practical constraint on new data‑centre builds.
  • Upcoming earnings from TSMC, ASML and major cloud providers are being treated as immediate tests of whether firms can monetise AI services enough to justify continued heavy capex and complex financing structures, with possible knock‑on effects for hiring, lease terms and regional power projects.