One day after Nvidia authorised a record $150 billion share buyback, Bain & Company published a number that puts the industry's optimism in perspective. In its Technology Report 2026, released Monday, the consultancy estimates that sustaining AI infrastructure spending will require an AI market approaching $6 trillion in annual revenue by 2031 — and that today's visible revenue lines cover only $1.2 trillion to $1.8 trillion of it.
The arithmetic runs backwards from capital expenditure. Hyperscalers Microsoft, Google, Amazon, Meta and Oracle could spend around $780 billion in 2026, nearly five times their level three years earlier. Leading-edge AI data centres are approaching 1 gigawatt of power capacity, with many expected to pass 2 GW by 2027 and 9 GW campuses emerging by the end of the decade. The biggest facilities roughly double in power and cost every 12 to 16 months: Meta's Prometheus opened in 2025 at about 0.6 GW and $24 billion and is projected at 4–5 GW and $120–175 billion for 2029, then roughly 9 GW and $200 billion by 2030.
By 2031, Bain estimates, annual spending on AI infrastructure could reach $1.5 trillion, including new capacity and upgrades to the installed base of GPUs, memory and networking equipment. Assuming capex equals about 25 percent of industry revenue — 'ambitious but reasonable', Bain writes, based on cloud-provider trends — the market has to reach roughly $6 trillion a year.
Consumer AI, through subscriptions and advertising, could contribute $200–400 billion by 2031; enterprise productivity gains could add $1 trillion to $1.4 trillion 'to providers alone'. That leaves a gap of about $4.2 trillion that must come from value that does not exist yet. Bain splits it across four categories: search and advertising monetised inside chatbots ($100–200 billion), 'autonomous everything' — self-driving cars, trucks, drones and industrial automation ($400 billion), physical AI including humanoid robotics and digital twins ($900 billion), and entirely new products such as AI-driven drug discovery, always-available mental-health support, next-generation batteries and autonomous scientific research.
Bain's framing lands in the middle of a growing argument about whether the infrastructure is being built ahead of the demand curve. German outlet Golem covered the report under the headline that the AI industry is missing a large share of the revenue it needs to reach profitability. Bain says funding the buildout sustainably would add roughly 1 percent to annual global GDP growth, and that the industry needs 'a wave of application innovation comparable with what mobile and cloud unlocked, not just productivity gains on existing workflows.'
