Intel and Broadcom CEOs Warn That Capital Access May Be AI's Biggest Bottleneck
· TexaHub · Semiconductors, AI in Industry, Technology Trends
Speaking at a financial conference in Singapore, Intel's Lip-Bu Tan and Broadcom's Hock Tan said the cost of capital and unrealistic investor expectations could constrain the AI buildout, even as demand for computing infrastructure continues to outstrip supply.
Capital as the Limiting Factor
Intel CEO Lip-Bu Tan and Broadcom CEO Hock Tan addressed a financial conference in Singapore on 12 October, painting a cautious picture of AI-sector economics despite strong demand signals [1]. Hock Tan said access to capital "might be the biggest constraint" facing the industry, while Lip-Bu Tan described funding and its cost as the near-term challenge, adding that investors ultimately want to see returns [2].
Hock Tan likened parts of the AI equity market to a casino, warning that many investors have set extreme expectations for AI stocks. He pointed to OpenAI as an example, noting the company "is still losing money because they're spending a lot on training," even though its revenue potential remains large [2].
Enterprise Revenue and the Pilot Problem
The Broadcom chief also questioned the depth of current enterprise AI adoption, saying much of AI labs' business revenue comes from pilot projects rather than committed production workloads [2]. He argued that labs should focus on winning customers who can demonstrate a return on their inference spending, a shift from the current land-grab dynamic in which companies compete mainly on training compute.
Lip-Bu Tan offered a complementary view, calling Chinese open-source models "good enough" for budget-conscious users and suggesting that not every enterprise will need the most expensive frontier systems [2]. The remark implicitly acknowledged a growing competitive pressure from Chinese AI developers whose models are available at no licensing cost.
Supply Shortages and Demand Outlook
On the hardware side, Hock Tan predicted that data-centre chip shortages would ease by 2028 or 2029, but cautioned that the gap between computing demand and supply could last into 2030 [2]. Intel disclosed last month that it can currently meet only about 50 percent of incoming orders, a constraint Lip-Bu Tan attributed to capacity ramp-ups still in progress [2].
Hock Tan warned that demand for computing infrastructure could slow if adoption of advanced AI models disappoints or if improvements in their capabilities begin to plateau [1]. The comment reflected a broader concern in the semiconductor industry that the current investment cycle depends heavily on sustained performance gains at the model layer.
Market Context
The remarks came as semiconductor stocks faced a turbulent week. South Korean chip shares fell sharply on 13 October, with Samsung Electronics and SK Hynix each declining more than 8 percent on concerns about Chinese competition and AI overheating [3]. Reuters described the session as the strongest sign yet that sentiment on what some analysts have called the "trade of the decade" is shifting, citing rising interest rates as a risk to AI capital spending [3].
Sources
[1] Intel, Broadcom CEOs warn capital crunch could become AI's biggest bottleneck — CNBC, 12 Oct 2026
[2] Intel, Broadcom CEOs flag AI industry challenges at financial event — SiliconANGLE, 12 Oct 2026
[3] Korean Chip Stocks Sink on Chinese Rival Worry, Broadcom Remarks — Bloomberg, 13 Oct 2026