The CEOs of Broadcom and Intel see the cost of capital as the next bottleneck for AI

In Singapore, Hock Tan and Lip-Bu Tan believe that funding could hold back AI more than chips or energy, despite demand outstripping supply until 2029.

At a conference hosted by DBS in Singapore on 12 October, Hock Tan, CEO of Broadcom, said access to capital could become the main constraint of AI's growth, even before the shortages of semiconductors, memory or electricity. Lip-Bu Tan, head of Intel, abounded in pointing to the rising cost of money and the question of the return on investment that markets will eventually demand.

Shortages to be overcome

For Hock Tan, the hardware bottlenecks follow but are eventually overcome: the tension on memory could ease within a year, the tension on silicon pads around 2028-2029, and the electrical supply could increase with the new infrastructure. Increasing bond yields, on the other hand, increased the cost of financing data centres whose profitability is measured over several years.

Both leaders remain optimistic in the short term: the demand for calculation exceeds supply and could continue to do so until 2029, even 2030. Beyond this, growth will depend on the real adoption of AI by companies and the continued progress of the models, a stagnation that can slow down orders.