Broadcom and Intel CEOs Question AI’s ROI and Financing Model

Summary

Broadcom CEO Hock Tan and Intel CEO Lip-Bu Tan said at the DBS Asia Future Forward Conference that the economics of AI remain uncertain despite massive investment. They questioned whether frontier-model companies such as OpenAI can turn reported revenue growth into a sustainable, profitable business while their model and infrastructure costs remain high. Hock Tan said enterprises need to find applications that generate returns, rather than merely consuming tokens while experimenting. He expects the infrastructure buildout could continue through 2029 or 2030, but said that outlook depends on adoption and business ROI. The executives cited recurring bottlenecks in memory, wafers and power, while saying financing and the cost of capital may be the largest constraints. AI-related borrowing costs have risen, making the expected returns on infrastructure spending more difficult to achieve. Demand for compute currently exceeds supply, but it could weaken if advanced-model deployment disappoints or capability improvements plateau. Intel’s Lip-Bu Tan said companies will need to demonstrate efficiency gains or new business models to justify their AI investments, while Hock Tan stressed how little is still known about the technology’s development.