The AI bubble can wait, but your exit plan can't

Forrester, the analyst firm, believes a bubble in AI infrastructure spending is not imminent, though businesses should brace for a correction that could disrupt suppliers, contracts, and prices. The analyst is not convinced that the current spending on AI infrastructure will generate enough returns to justify the investments.

Forrester's co-head of emerging technology, Brian Hopkins, says that 90% of the money for AI computing infrastructure comes from large, healthy technology companies, reducing the risk. However, there are concerns about a few companies facing financial strain and circular financing deals with AI labs.

Forrester has developed a tool, the AI Bubble Model, to help businesses monitor risks and prepare for potential corrections. The company's latest report notes that hyperscale operators are committing a record share of revenue to infrastructure to support AI development and use, but this investment may not be justified by the productivity gains.

Forrester calculates that the infrastructure investments would break even only if AI increases US GDP by 5% by 2035. However, productivity improvements could only account for about two percentage points of this increase, leaving roughly $1.3 trillion to be generated from other sources of value.

Bain & Company's recent report echoed similar findings, estimating the AI industry needs to generate $6 trillion in annual revenue by 2031 to fund all the infrastructure required. This revenue target will likely require new uses for AI beyond simply enhancing productivity.

Forrester warns that the same forces that are making AI abundant could also make parts of the market unstable. Enterprise users should closely monitor the financial health of the companies building AI computing and be prepared for corrections. Companies should track whether their AI projects deliver measurable value and limit exposure to vulnerable suppliers.