The real risk for AI companies is running out of cash

This article summarizes an episode of Invest Like The Best’s video series featuring Ben Thompson, founder of Stratechery.

Photo credit: [eamesBot] / Shutterstock

Believing in the future of AI will not pay for the servers required to run it. Stratechery founder Ben Thompson

Ben Thompson, founder of tech analysis newsletter Stratechery, warns leaders that the current tech boom masks a cash flow problem, leaving companies at risk of going bankrupt before they can generate the returns needed to survive.

AI infrastructure costs threaten corporate survival

The rush to secure computing resources masks a cash flow problem where companies could exhaust funding before generating profits.

Thompson explains,”I believe in AI. It’s real. I think the economic impact is going to be astronomical… You can believe all that and still be worried about whether we make the bridge to generating the level of returns necessary to continue to fuel [AI investment].”

Surviving this expensive building phase requires boards to execute a clear timeline:

• Past planning: Today’s hardware shortages are the result of decisions made long before the current boom.
• Capital runway: Companies must monitor when they stop using their cash and require outside loans.
• Market consolidation: A market crash would transfer ownership of data centers to wealthier competitors.

Compute demands break standard software economics

To secure that capital runway, software companies must abandon flat-rate subscriptions and adapt to unpredictable computing expenses:

• Charge based on complexity: Group customers by computing needs to separate casual queries from complex AI tasks.
• Track corporate usage: Avoid arguments over employee licenses by monitoring utilization.
• Use advertising models: Cover the infrastructure costs of consumers who refuse monthly fees by improving ad tracking.

This shift reveals that free access expands the user base, but subscriptions rarely cover the infrastructure required for a global audience.

The risk of hardware constraints forcing missed opportunities is high. Thompson observes that “Risk doesn’t disappear. It just moves. Right now, the risk is that every single big tech company realizes, ‘If we had more compute, we could be making more money.’”