.png){kind=logo}
Broadcom CEO Hock Tan said on Monday, October 12, that access to capital is now the biggest limit on AI growth. Building and expanding AI systems costs more each year, and funding is getting harder to find than chips.
Broadcom makes custom AI chips and networking parts for large computing centers. Its customers include the companies that build and run advanced AI models. Demand for these products is strong.
The company earlier raised its AI chip revenue outlook to about USD 115 billion for the year ending October 2027. It expects revenue to reach around USD 230 billion in fiscal 2028.
Data centers need a lot of money before they earn anything. Companies must pay for buildings, computers, and networking gear up front. The returns may take years to arrive.
Many firms borrow to cover these costs. Higher interest rates make that borrowing more expensive. Investors are also watching more closely. They want to see profit, steady cash flow, and a clear plan to earn back the spending. AI developers, cloud providers, and data center operators all face the same question: will future revenue justify the cost?
Also Read: Personal AI Agent Startup Instinct Lands USD 1 Billion Funding Round
Getting chips is not enough to add capacity. A data center also needs reliable power, a good site, and a finished building. A delay in any of these can slow a project, even when the chips are ready.
The timing matters. Tech companies are about to report their latest earnings, and AI spending is expected to stay a major part of their investment plans. Rising rates and more corporate debt could add pressure to those plans.
Tan's comments point to a change in what holds the industry back. For a while, the main worry was whether enough chips could be made. Now the worry is whether companies can pay for the growth they want.
The firms that can raise money at a fair cost and turn their spending into steady returns will likely grow fastest. Those that cannot may have to slow down, no matter how strong the demand for AI becomes.
