BCA Research estimates the artificial-intelligence industry may need close to $10 trillion in annual revenue to justify the capital spending flooding into data centers. The estimate lands as history buffs point to the dot-com crash, when the S&P 500 lost more than 45% of its value once an earlier tech boom went bust.
The last major technology bubble cost the S&P 500 more than 45% of its value and took Cisco a quarter-century to recover from.
The dot-com playbook
The S&P 500 fell more than 45% after the dot-com bubble burst, while the technology-heavy Nasdaq-100 lost more than 80% of its value. Cisco, the poster child of that crash, took roughly a quarter-century to recover, and the Nasdaq-100 took around 15 years to get back to even. Wall Street's pattern repeats: early investors profit, more pile in fearing they'll miss out, and prices get pushed past reasonable valuation. Nvidia is now subsidizing customers in ways that are bolstering demand for its AI chips, and market watchers are already questioning those arrangements.
A $10 trillion revenue bar
BCA Research quantifies just how large the AI payoff needs to be. Assuming a 15% pre-tax return on invested capital and EBITDA margins of 30% rather than Wall Street's more optimistic 50%, hyperscalers alone would need $7.4 trillion in annual revenue. Add non-hyperscaler AI spending across China, "neocloud" providers and private ventures such as SpaceX, and the global AI industry would need roughly $10 trillion in annual sales — an amount BCA Chief Economist Peter Berezin says equals total global annual spending on either healthcare or food.
Margins masked by capex accounting
The S&P 500 trades at roughly 19 times forward earnings, in line with its 10-year average, but that rests on record forward profit margins of 16.7%. If margins normalized to 2019 levels, equities would trade at a forward P/E of 26.7x instead.
Current margins are propped up partly because hyperscalers — Microsoft, Amazon, Alphabet, Meta and Oracle — book AI hardware purchases as capital spending to depreciate over time rather than expense immediately. BCA projects those depreciation charges will more than double from $255 billion in 2026 to $581 billion by 2029, as capex hits $1.16 trillion; including off-balance-sheet spending, total hyperscaler capex could approach $1.4 trillion annually by the end of the decade.
History offers one consolation: oversupply from a burst bubble usually cuts the cost of new technology, broadening its use even after the correction that gets there.
Sources: The Motley Fool, Investing.com
Trading involves risk.
Risk Disclaimer
