Bain: AI market must make $6T a year by 2031 to fund its infrastructure

The AI industry must generate $6 trillion in annual revenue by 2031 to keep funding the infrastructure that expected demand requires. The figure comes from the 2026 Global Technology Report by management consultancy Bain & Company.

Where $6 trillion comes from

Bain says demand for AI compute has revived the hardware industry. High-bandwidth memory (HBM) used in GPUs, advanced packaging, custom silicon and application-specific integrated circuits (ASICs) are all scaling fast.

The report points to an "arms race" among hyperscalers. Capital expenditure by Microsoft, Google, Amazon, Meta and Oracle could reach $780 billion in 2026, nearly five times the level of just three years earlier. By 2031, annual AI infrastructure spending could hit $1.5 trillion, close to the $1.6 trillion Omdia forecasts for 2030.

The $6 trillion follows from Bain's assumption that capex will equal roughly 25 percent of industry revenue, a ratio it calls ambitious but reasonable given recent trends among cloud providers.

The revenue gap

Existing AI uses, from consumer subscriptions and advertising to enterprise software, sales, marketing, customer service and IT operations, should bring in between $1.2 trillion and $1.8 trillion. That leaves about $4.2 trillion to be found elsewhere.

Bain lists four categories. AI models replacing search engines and selling ads could unlock $100 billion to $200 billion. Autonomous cars, drones and industrial automation could be worth about $400 billion. Physical AI, including simulations, digital twins and robotics, as much as $900 billion. Together that is $1.5 trillion, still leaving a $2.7 trillion shortfall that Bain says must come from "new products and uses that don't exist today", such as AI-driven drug discovery, mental health support, materials science and faster scientific research.

The doubts

"The debate today is fixated on employee productivity. What the industry needs is a wave of innovation that will dwarf what mobile and cloud unlocked," said David Crawford, chairman of Bain's global Technology practice.

Not everyone expects the target to be met. A Jefferies report found only half the US datacenter capacity scheduled for 2026 is actually under construction, with work yet to begin on as much as 80 percent of the 2028 pipeline. A second Jefferies report says chip manufacturing constraints will cap how many planned AI server farms can come online. A year ago Bain itself put the required revenue at $2 trillion by 2030, a number that has now trebled.