New Constructs Warns AI Compute Spending Is Not Profit

New Constructs said investors are overvaluing AI compute capacity after Morgan Stanley estimated that five major cloud operators could spend $1.2 trillion on AI infrastructure in 2027, according to the firm's press release.

David Trainer, CEO of New Constructs, said chips, servers, data centers, power, and access to AI models do not automatically create pricing power, customer value, or economic profit. He said investors should focus on whether AI spending produces measurable customer value, sustainable cash flow, and returns above the cost of capital.

The firm said proprietary data and workflows are more important than access to general AI models. Trainer cited closed loop datasets and application layers that organize business data, evaluations, and rules around models as areas where companies may create value.

New Constructs pointed to Palantir as an example of a company built around customer data and operational workflows. It also said its FinSights product uses Google Cloud technology with New Constructs research data, with the cloud infrastructure serving as a delivery layer rather than the source of the investment analysis.