The AI industry may need $2.6 trillion in annual spending by 2030 to support its infrastructure and development, according to a BCG Institute analysis cited by Fortune. The striking question is whether businesses will actually spend enough, and get enough value back, to make that build-out pay.
Watch Desk analysis
What happened
Fortune reports that BCG’s estimate covers data centres, chips and model development, as well as returns for investors. It excludes China and reflects infrastructure costs concentrated mainly in the US. Enterprises would account for $1.8 trillion of the annual total, according to the analysis.
That $1.8 trillion is a measure of what the AI ecosystem would need from companies, not a forecast that they will spend it. In one illustrative scenario, BCG says enterprises could reach the figure by shifting $857 billion from existing technology budgets and adding $937 billion in new spending. The scenario would put those budgets about 13 per cent above an IDC IT-spending forecast, with AI rising from 5 per cent of enterprise technology budgets today to 25 per cent in 2030.
Why it matters
This is a useful counterweight to the endless parade of announcements about new chips and data centres: the build-out ultimately depends on customers paying for it. BCG’s estimate is a spending requirement, not evidence that demand or returns are guaranteed.
The analysis also offers numbers to test that demand against. It estimates that each gigawatt of capacity coming online in 2026 needs to generate about $15 billion in revenue to cover its costs, assuming AI chips last five years. BCG identifies two signals to watch: whether infrastructure financing shifts persistently towards debt, and whether deployed compute generates enough revenue. Weaker demand could mean price wars and losses for model providers and data-centre operators.
Our read
The headline figure is huge, but the useful story is the bill’s proposed payers. The enterprise share is a demanding target, not a cheque already written. Companies weighing new AI spending should be able to point to productivity gains, cost reductions, new revenue or a competitive advantage, BCG’s Clark O’Niell told Fortune. He also warns that some strategic benefits may not fit neatly into traditional return calculations. “We might fall behind” is a consideration; it is not a business case all by itself.
For buyers, one practical suggestion from O’Niell is to avoid long-term price-per-token commitments while AI prices continue to fall. Flexibility could let companies benefit as costs change, rather than locking in yesterday’s price for tomorrow’s technology.
What to watch
- Whether enterprise AI spending approaches BCG’s $1.8 trillion estimate, rather than merely appearing in ambitious budgets.
- Whether AI infrastructure investment leans more heavily on debt, and whether deployed compute produces sufficient revenue.
- Whether buyers keep AI contracts flexible as token prices fall.
Discussion spark: If AI’s promised strategic value is hard to capture in conventional ROI, what evidence should boards require before approving another major investment?
Sources and evidence
- Source update (3 November 2026, 12:59 UTC)
Watch Desk is operated by WittyWires as an independent cross-cutting AI news tracker. It does not speak for the organisations or people it covers.