Intel and Broadcom CEOs have pointed to funding as the biggest constraint on the artificial intelligence industry.

CNBC reported on Oct. 12 that the two chiefs warned capital and the cost of capital could become a bigger bottleneck than chips or power, even as demand for computing infrastructure outstrips supply.

The remarks were made at the DBS Asia Future Forward Conference in Singapore. Hock Tan (호크 탄), Broadcom's CEO, called capital the biggest constraint holding back AI expansion. "The biggest constraint may be capital," he said. As AI models become more advanced, they require huge sums for data centres, advanced chips and power infrastructure, meaning funding conditions could determine the pace of industry growth.

Lip-Bu Tan (립부 탄), Intel's CEO, also highlighted the issue. "If there is an immediate challenge ahead of us, it is capital and the cost of capital," he said. "In the long run, the market will want to see whether companies can generate real returns from their AI investments," he added. He sees a risk that expansion could slow if returns on investment do not support continued funding.

The CEOs' concerns tie into the recent structure of AI infrastructure investment. The industry has kept up large-scale investment in data centres, advanced semiconductors and power facilities to prepare for surging demand, but rising borrowing costs are increasing the burden of such investment. Higher bond yields can raise the cost of financing data centre construction and chip purchases, and pressure the economics of projects that take time to recoup returns.

Still, the two companies expressed strong confidence in near-term demand. Hock Tan said demand for computing infrastructure exceeds what the company can supply. He also mentioned the possibility that this supply-demand imbalance could continue through 2029 and even into 2030. Lip-Bu Tan also said the industry was unlikely to face a major slowdown over the next 2 to 3 years, citing the spread of AI applications and rising computing demand.

They also said existing supply bottlenecks could ease over time. Hock Tan said the industry has faced supply constraints several times over the past 2 years but ultimately moved past them, adding that the current memory shortage may not be the same issue a year from now. He also said wafer shortages could ease around 2028 or 2029, and power supply and demand could improve once new infrastructure comes online.

The key issue is what comes after that. The two CEOs said AI's long-term growth requires faster adoption of AI in businesses and proof that the technology delivers real economic effects. Broadcom warned that demand for computing infrastructure could cool if adoption of advanced AI models falls short of expectations or if the pace of performance improvements slows. Intel stressed that companies will need to show AI investment results more clearly through efficiency gains and new business models.

As a result, the AI industry's focus appears to be shifting from simply expanding semiconductor supply to financing structures and verifying investment returns. In the short term, infrastructure demand will support the market, but over the medium to long term, competitiveness is emerging as a question of who can secure funding at lower cost and prove investment performance.