Broadcom and Intel CEOs say capital, not chips, is AI’s next bottleneck

Photo: Pixabay / Pexels

Broadcom and Intel CEOs say capital, not chips, is AI’s next bottleneck

Hock Tan and Lip-Bu Tan told a Singapore conference that financing costs could slow the AI infrastructure buildout more than hardware shortages

For the past few years, the AI buildout story has been about scarcity. Not enough chips, not enough memory, not enough power to plug it all into.

Now the CEOs of Broadcom and Intel say the shortage that matters most is money.

Speaking at the DBS Asia Future Forward Conference in Singapore on October 12, 2026, Broadcom CEO Hock Tan and Intel CEO Lip-Bu Tan warned that the rising cost of capital could become the biggest constraint on AI infrastructure expansion. In their view, financing could soon overtake the semiconductor and memory crunches that defined the last stretch of the AI boom.

Hock Tan kept his version short.

Advertisement

“The biggest constraint is capital.”

Lip-Bu Tan offered a slightly expanded take: “It’s capital and the cost of capital.”

Demand still outruns supply

Neither executive suggested that appetite for AI compute is fading. Both said demand for computing infrastructure is projected to exceed supply until at least 2029 or 2030.

The physical side of the equation looks healthier than it used to. According to the executives, earlier constraints in wafers, memory and power have eased over the last two years. Improvements in wafer capacity are expected by 2028-2029, and memory shortages are likely to resolve within the next year.

Broadcom is already building the financing pipes

Broadcom isn’t just naming the problem. The company has set up a $35 billion financing platform with Apollo and Blackstone, aimed at supporting more than 20 gigawatts of planned AI capacity by 2028.

The broader AI market is heading in the same direction. There have been reported discussions of more than $50 billion in debt financing for OpenAI and up to $42 billion for Anthropic.

Both CEOs stressed that investors pouring capital into AI will eventually expect a return on it.

What this means for the AI buildout

A few things are worth watching. One is how the Broadcom, Apollo and Blackstone platform deploys money against its 20-gigawatt target. Another is whether the reported OpenAI and Anthropic debt talks close, and on what terms. It will also be telling whether other hardware suppliers follow Broadcom in building their own financing arrangements. And the expected easing of memory shortages within the next year will show whether the physical constraints truly fade as projected.

Disclosure: This article was edited by Diego Almada Lopez. For more information on how we create and review content, see our Editorial Policy.
Broadcom and Intel CEOs say capital, not chips, is AI’s next bottleneck
Broadcom and Intel CEOs say capital, not chips, is AI’s next bottleneck

Hock Tan and Lip-Bu Tan told a Singapore conference that financing costs could slow the AI infrastructure buildout more than hardware shortages

Photo: Pixabay / Pexels

For the past few years, the AI buildout story has been about scarcity. Not enough chips, not enough memory, not enough power to plug it all into.

Now the CEOs of Broadcom and Intel say the shortage that matters most is money.

Speaking at the DBS Asia Future Forward Conference in Singapore on October 12, 2026, Broadcom CEO Hock Tan and Intel CEO Lip-Bu Tan warned that the rising cost of capital could become the biggest constraint on AI infrastructure expansion. In their view, financing could soon overtake the semiconductor and memory crunches that defined the last stretch of the AI boom.

Hock Tan kept his version short.

Advertisement

“The biggest constraint is capital.”

Lip-Bu Tan offered a slightly expanded take: “It’s capital and the cost of capital.”

Demand still outruns supply

Neither executive suggested that appetite for AI compute is fading. Both said demand for computing infrastructure is projected to exceed supply until at least 2029 or 2030.

The physical side of the equation looks healthier than it used to. According to the executives, earlier constraints in wafers, memory and power have eased over the last two years. Improvements in wafer capacity are expected by 2028-2029, and memory shortages are likely to resolve within the next year.

Broadcom is already building the financing pipes

Broadcom isn’t just naming the problem. The company has set up a $35 billion financing platform with Apollo and Blackstone, aimed at supporting more than 20 gigawatts of planned AI capacity by 2028.

The broader AI market is heading in the same direction. There have been reported discussions of more than $50 billion in debt financing for OpenAI and up to $42 billion for Anthropic.

Both CEOs stressed that investors pouring capital into AI will eventually expect a return on it.

What this means for the AI buildout

A few things are worth watching. One is how the Broadcom, Apollo and Blackstone platform deploys money against its 20-gigawatt target. Another is whether the reported OpenAI and Anthropic debt talks close, and on what terms. It will also be telling whether other hardware suppliers follow Broadcom in building their own financing arrangements. And the expected easing of memory shortages within the next year will show whether the physical constraints truly fade as projected.

Disclosure: This article was edited by Diego Almada Lopez. For more information on how we create and review content, see our Editorial Policy.