- The AI industry could need $6 trillion in annual revenue, new report claims
- New AI products would generate trillions in commercial value
- Search, advertising, and autonomous systems could drive most of the future revenue
AI infrastructure spending is racing ahead, and by 2031 the industry will need to find much more commercial value to justify the bill, new research has claimed.
Bain & Company estimates that annual spending on facilities, processors, memory and networks alone will rise to $1.5 trillion by 2031.
To keep this investment going, the industry needs about $6 trillion in annual revenue, assuming infrastructure swallows about 25% of sales each year.
Latest Videos FromTechRadar
Where would all that revenue come from?
Bain calls this 25% assumption bold but fair, as it matches what cloud computing providers have spent on infrastructure in previous years.
Still, the math leaves AI businesses needing much more commercial activity than their current products bring in, so the hunt for new revenue begins.
The biggest chunk, about $4.2 trillion, would come from new products in search, advertising, autonomous systems and physical AI, many of which barely exist today.
Enterprise productivity comes in second, adding $1 trillion to $1.4 trillion as companies lean on AI for software development, sales, marketing, customer support and IT operations.
Consumer subscriptions and advertising are far behind, contributing only $200 billion to $400 billion even as providers push AI products to billions of users.
The speed of adoption is also important, so leading AI labs spend more than $9.75 billion on engineering that helps companies make AI work faster.
“The economics of AI infrastructure demand trillions in new revenue via productivity gains,” said David Crawford, president of Bain’s global technology practice.
Crawford adds that the industry needs a flood of ideas big enough to dwarf what mobile technology and cloud computing have unleashed in previous years.
Beyond that, Bain says future products could reach into drug discovery, mental health and energy generation, areas where AI has limited presence today.
The expansion of the data center compounds the financial pressure
The physical infrastructure supporting AI will become larger and more expensive as companies continue to increase their computing capacity.
Bain says the size and cost of the data center have increased at roughly double rates over periods that last about 12 to 16 months globally.
According to Epoch AI, Meta’s Prometheus facility in Ohio had 600MW capacity and an estimated $24 billion cost in 2025.
The facility could reach 2GW and $80 billion by 2027, before increasing to 5GW and $175 billion in 2029 under current projections.
By 2030, Epoch AI estimates the project could reach 9GW while requiring as much as $200 billion in total spending.
These facilities require additional electricity generation, grid connections, advanced semiconductors, skilled workers and equipment capable of supporting sustainable operations at scale.
The pace at which money is flowing into data centers has made securing sufficient computing capacity the industry’s most pressing concern.
“But the more important question is perhaps whether enough economic value can be created to justify it,” the report said.
About National News
Follow TechRadar on Google News in the add us as a preferred source to get our expert news, reviews and opinion in your feeds.