← Back to Hardware Hardware

Intel and Broadcom CEOs warn capital is becoming AI's biggest bottleneck

Intel CEO Lip-Bu Tan and Broadcom CEO Hock Tan each told the DBS Asia Future Forward Conference in Singapore that financing costs and ROI pressure, not chip or power shortages, are likely to be the biggest constraint on AI infrastructure growth.

Two CEOs, one warning

When two of the semiconductor industry’s most consequential executives converge on the same point, it deserves attention. At the DBS Asia Future Forward Conference in Singapore on October 12, 2026, Broadcom CEO Hock Tan and Intel CEO Lip-Bu Tan each identified capital—not chips, not memory, not power—as the most likely constraint on the next phase of AI infrastructure growth.

It is a striking reframing. Industry coverage of the AI buildout has fixated on supply: wafer capacity, HBM memory allocation, data center power, and the physical limits of how many racks can be switched on at once. Both CEOs now argue that the scarcest resource may be the money to buy all of it, and the mounting pressure to demonstrate that the money earns a reasonable return. In their framing, the constraint sits in the ledgers and boardrooms, not on the factory floor.

The quotes

Hock Tan was direct: “What I think might be the biggest constraint is capital.” He framed financing, not components, as the brake on an industry whose demand for compute already exceeds what can be supplied—an imbalance he expects to persist through 2029 and possibly into 2030. His comments come as Broadcom sits on one of the largest and fastest-growing order books in the industry, tied to custom accelerators and networking for major AI platforms.

Lip-Bu Tan echoed the view through a different lens. “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,” he said. Rather than a parts shortage, he pointed to borrowing costs and the expectations of investors as the risks for companies pouring money into AI infrastructure. The cost to fund the buildout, and the return required to justify it, are what he sees as the emerging constraint.

Why the timing matters

The warning lands as both companies are writing their own enormous capital checks. Intel upsized an August common stock offering to $20 billion, pricing more than 210 million shares, to fund expansion of its chip contract manufacturing business. The company raised its 2026 capital expenditure target to $20 billion from $18 billion and has committed to high-volume production with its 14A process node by 2028.

That is the practical tension behind the CEOs’ words: the companies closest to the silicon are also the deepest into debt and equity markets to keep the buildout moving. Intel’ move raises a real question about what a $20 billion stock issuance, stacked on top of an expanded spend plan, does to returns as the company races to bring its foundry business to scale. Broadcom’s exposure is even more levered to AI demand, with Anthropic expected to become its largest compute customer in 2027 ahead of a late-2026 IPO, concentrating its near-term growth in a single, extremely large relationship.

A capital crunch, or a confidence test

The CEOs did not paint a picture of collapse. Both signaled that today’s demand remains powerful and that the physical supply lines—wafer capacity and power delivery—will improve over time. Their concern is narrower and more financial: whether the capital market will continue to fund the pace of expansion cheaply enough, and for how long investors will tolerate long payback periods before demanding proof of ROI. If financing becomes more expensive or ROI scrutiny tightens, the buildout that depends on it could slow even while demand stays strong.

Beyond the bottleneck

Despite the caution, both executives remained bullish on near-term demand. Hock Tan said there is more demand for compute infrastructure than the industry can supply, while Lip-Bu Tan said a significant slowdown over the next two to three years is unlikely. He even suggested wafer shortages could ease by 2028 or 2029 and that power availability will improve as new infrastructure comes online. The constraint, in his framing, is the financing to bridge the gap.

What this means for you

What this means for you

If you are buying, building, or betting on AI infrastructure, the next bottleneck is likely to be money rather than parts. Rising cost of capital and investor pressure to demonstrate returns may slow expansion before chip or power supply does—so watch financing terms and ROI expectations, not just supply chains.

Sourcing and context

This report draws on CNBC, Quartz, and Benzinga coverage of the October 12, 2026 remarks at the DBS Asia Future Forward Conference in Singapore. Both executives remained publicly bullish on demand while flagging capital, the cost of capital, and ROI as the emerging constraint. The views are forward-looking judgments by two sitting CEOs, not confirmed market forecasts, and neither company detailed a specific point at which financing pressure would translate into a slowdown. Readers should weigh the remarks as executive risk assessment rather than an analyst call.

Sources: CNBC; Quartz; Benzinga.

Corrections: Report an error at corrections@wireshaman.com.

More from the wire