Bain puts a $6 trillion revenue test on the AI data center buildout

Bain's projection of $5 trillion to $6.5 trillion in data center spending through 2030 would require an AI market approaching $6 trillion in annual revenue by 2031. The link is a ratio the firm applies. If capital expenditure represents roughly 25% of industry revenue, then roughly $1.5 trillion in annual AI infrastructure spending Bain expects by 2031 implies an industry of that scale. Peter Hanbury, a partner at the Boston-based consulting firm, described the price tag as "conceivable" in an interview with Construction Dive published Oct. 8, while questioning whether AI companies will eventually produce enough revenue to keep construction going at that pace.

Hanbury says the largest hyperscalers could spend roughly $780 billion in capital expenditure in 2026, nearly five times their spending three years earlier. The five are Microsoft, Google, Amazon, Meta and Oracle. Hanbury is explicit that not all of that total is data center capex — the figure measures how much capital the buyers of compute already command, not a buildout forecast. "The near-term issue is not necessarily whether enough capital exists," he said.

Bain's immediate constraints are physical rather than financial. Hanbury points to power constraints and electrical labor as the headwinds contractors face now, and to lead times on key materials long enough that decisions made today determine capacity several years from now. That is grid access becoming an underwriting variable rather than a queue position, with the megawatt rather than the building as the asset.

Bain says local opposition blocked or delayed at least 75 projects worth $130 billion in the first quarter of 2026, nearly matching the impact across all of 2025. Opposition at that scale has already pushed some site strategy toward buying power that already exists. Construction Dive reported Oct. 5 that Blackstart Digital agreed to buy IBM's Almaden campus, whose substation provides 25MW, as a case of a developer paying for power a previous owner had already secured instead of waiting in a queue.

Hanbury frames the buildout as several conditions that must hold at once, and the first is that AI has to move beyond use cases focused on efficiency and productivity.

Five balance sheets that can commit roughly $780 billion in capital expenditure in 2026 can carry anchored leases; capacity without that counterparty is a different credit. If AI revenue arrives more slowly than the $6 trillion the arithmetic requires, the repricing would likely land first on merchant and non-anchored capacity, where no hyperscaler lease stands behind the offtake.