Bain estimates annual revenue of $6 000 billion is needed to fund AI infrastructure in 2031

Bain & Company estimates that $6 000 billion in annual revenue could be needed in 2031 to fund global artificial intelligence infrastructure in its seventh Technology Report, published on September 29, 2026. The firm bases this estimate on annual infrastructure spending that could reach $1 500 billion and a hypothetical investment-to-revenue ratio.

In the press release accompanying its report, Bain describes infrastructure being built ahead of demand. The firm thus links the expansion of computing capacity to the revenue that future services will need to generate to fund it.

The calculation depends on two assumptions

The $1 500 billion in annual spending envisaged for 2031 covers new data centers, additional computing capacity and the replacement of equipment already installed. This includes GPUs, the graphics processors used for AI computations, as well as memory and networking equipment.

Bain assumes that these investments represent approximately 25% of the sector’s revenue, an assumption inspired by the economics of cloud providers. Dividing 1 500 billion by 0.25 gives $6 000 billion in annual revenue. With infrastructure spending unchanged, the calculated revenue requirement falls with a higher ratio and rises with a lower ratio. The amount therefore depends as much on this assumption as on the projected spending.

Between $4 200 billion and $4 800 billion in revenue to generate

For the categories already identified, Bain projects between $1 200 billion and $1 800 billion in annual revenue in 2031. Consumer subscriptions and advertising would account for $200 billion to $400 billion. Enterprise AI providers could generate between $1 000 billion and $1 400 billion through services used in software development, sales, marketing, customer service and IT operations.

At the high end of this projection, the expected $1 800 billion would leave an additional $4 200 billion in annual revenue to be generated to reach the model’s $6 000 billion. With the low-end projection of $1 200 billion, simple subtraction gives a gap of $4 800 billion. This latter amount is an arithmetic deduction. These gaps concern future revenue in Bain’s scenario; they do not measure a current loss or an observed cash shortfall.

Bain is counting on new products and services

Bain proposes new opportunities in search and advertising, as well as in autonomous machines, including vehicles and drones. The firm also cites physical AI, which includes robotics, simulations and digital twins, virtual models of equipment or processes.

Drug discovery also features among the possible new products. These avenues could expand AI providers’ sales if they find customers. Their contribution to projected revenue remains to be demonstrated.

Cost savings reduce the expenses of a business using AI. They must be distinguished from the revenue recorded by providers for the products or services supplied. Bain’s calculation links investments to that annual revenue; it leaves open the questions of profit after costs and the timing of cash receipts.

If applications and their commercialization progress too slowly, capacity could be underused and harder to fund. This risk stems from the mismatch Bain describes between the infrastructure being built and expected demand. The firm warns that its projections do not guarantee any outcome.

To assess this scenario, AI revenue from external customers will need to be compared with capital expenditure and margins. Actual infrastructure usage will also help assess whether the capacity built attracts the expected demand.