Bain says AI needs $6T a year by 2031 to justify data centres
Bain and Company, the US consultancy, says the AI industry needs to earn about $6 trillion in annual revenue by 2031 to justify the capital now going into data centres. The claim comes from its latest technology report series, published on Tuesday. Bain argues the industry will have to become creative with new propositions to get there.
The arithmetic starts with infrastructure. Bain forecasts that annual spending on AI infrastructure might hit $1.5 trillion by 2031, covering new facilities, more capacity and upgrades to the installed base of GPUs, memory and networking equipment. If capital expenditure amounts to about a quarter of industry revenue, which Bain calls "an ambitious but reasonable percentage based on trends among cloud providers", then sustaining that level of investment would require an AI market approaching $6 trillion a year.
Bain splits the needed revenue into segments. New product development is expected to be the biggest contributor, at about $4.2 trillion. That segment would include innovations in search, advertising, autonomy and physical AI. Enterprise productivity would require $1 trillion to $1.4 trillion in revenue, to support gains in software development, sales, marketing, customer service and IT operations. Consumer-focused services, meaning subscriptions and advertising revenue, are seen contributing $200 billion to $400 billion. The article describes that consumer segment as widely acknowledged to be crucial, as providers push AI products to billions of users.
Bain also names a "new competitive variable": absorption speed, defined as the pace at which companies can put AI to work. Leading AI labs are investing upwards of $9.75 billion in engineering models to help companies assimilate AI faster.
David Crawford, chairman of Bain's global technology practice and lead author of the report, said: "The debate today is fixated on employee productivity. The economics of AI infrastructure demand trillions in new revenue beyond productivity gains. What the industry needs is a wave of innovation that will dwarf what mobile and cloud unlocked." The report itself says the buildout has focused attention on the challenge of building capacity, "But the more important question may be whether enough economic value can be created to justify it." Bain adds that new products and uses that do not exist today may include drug discovery, mental health and energy generation.
On scale, Bain says the size and cost of AI data centres is accelerating, doubling approximately every 12 to 16 months. As an example it cites Meta Platforms' Prometheus data centre in Ohio, using data from San Francisco-based Epoch AI. Prometheus had a capacity of 600MW at an estimated cost of $24 billion in 2025. That is projected to reach as much as 2GW and $80 billion by 2027, then 5GW at up to $175 billion by 2029, and 9GW at $200 billion by 2030.
The report lists obstacles as more data centres are built: adding grid capacity, securing GPUs and other infrastructure components, a skilled workforce that must be retained "far above historical rates", and public opinion and regulation, such as pushback on resource use and noise pollution. On the other side, several governments support the growth of AI and data centres, including those of the UAE, Saudi Arabia, the EU, South Korea and the US. Bain notes that data centres are now central to technology innovation, economic growth and national sovereignty. It adds that bottlenecks in power, semiconductors and other inputs carry large capital needs of their own, opening additional entry points for investors, and that partnerships offer both a way in and geographic diversification as sovereign infrastructure becomes part of national strategies.
Key facts
- Bain and Company says the AI industry needs about $6 trillion in annual revenue by 2031 to justify data centre capital.
- The biggest piece is new product development at about $4.2 trillion, ahead of enterprise productivity ($1 trillion to $1.4 trillion) and consumer services ($200 billion to $400 billion).
- The $6 trillion follows from Bain's forecast that AI infrastructure spending might hit $1.5 trillion a year by 2031, assuming capex is about a quarter of industry revenue.
- Bain says AI data centres are doubling in size and cost approximately every 12 to 16 months; Epoch AI data has Meta's Prometheus in Ohio going from 600MW and $24 billion in 2025 to 9GW and $200 billion by 2030.
- Bain calls absorption speed, the pace at which companies can put AI to work, the new competitive variable.
Why it matters
The report shifts the question from whether capacity can be built to whether enough economic value exists to pay for it. Bain says the current debate is fixated on employee productivity, while the economics of AI infrastructure demand trillions in new revenue beyond productivity gains. The scale is the point: with spending that might reach $1.5 trillion a year by 2031 and data centres doubling in size and cost every 12 to 16 months, the revenue bar rises alongside the buildout.
Who it affects
Data centre builders and their backers, such as Meta with Prometheus in Ohio, face the capital numbers directly. AI labs are affected through the push to help companies adopt AI faster, where Bain says they are investing upwards of $9.75 billion in engineering models. Enterprises are affected because adoption pace is now a competitive variable. Bain also says bottlenecks in power and semiconductors open entry points for investors, and that sovereign infrastructure makes partnerships a route in for governments and investors. Governments named as supporting the sector are the UAE, Saudi Arabia, the EU, South Korea and the US.
How to use it
This is an analytical framework, not a product. It gives a way to read AI investment claims: take projected infrastructure spending, apply the assumption that capex is about a quarter of industry revenue, and see what market size that implies. It also gives the revenue mix to watch: new products (search, advertising, autonomy, physical AI), enterprise productivity in areas such as software development, sales, marketing, customer service and IT operations, and consumer subscriptions and advertising.
How solid is it
The figures come from a named consultancy report, with David Crawford as lead author, and the article reports them directly. The Prometheus numbers come from Epoch AI data, not from Bain's own estimates or from Meta. The quarter-of-revenue assumption is Bain's own; it calls the percentage ambitious but reasonable, based on trends among cloud providers. The article does not state current annual AI industry revenue, and the segment figures are not shown to sum to $6 trillion. The $9.75 billion figure is given without naming the labs or a timeframe.
Risks and caveats
The $6 trillion target rests on a forecast ($1.5 trillion of spending by 2031) and an assumed capex ratio, so a different ratio changes the answer. Much of the required revenue, about $4.2 trillion, depends on new products and uses that Bain says do not exist today. Bain lists practical constraints: grid capacity, GPU supply, skilled workforce retention, and public pushback on resource use and noise. The report does not say the target will be missed; it says the question may be whether enough economic value can be created. The article reports no reaction from Meta, AI companies or investors.
“What the industry needs is a wave of innovation that will dwarf what mobile and cloud unlocked”
— David Crawford, chairman of Bain's global technology practice and lead author of the report