AI bubble could hit U.S. and global growth: dot-com echoes, record leverage, trillions at risk

Could an AI unwind hit the U.S. and global economy?

AI-linked stocks have powered indexes to record highs. Valuations now echo the late 1990s, which is why some policy voices are warning about bubble dynamics. Others argue the capex is rational and the productivity math works.

Both can be true in different time frames. The near-term question for finance teams: if AI equity multiples compress, does it stay an equity story or spill into credit and the real economy?

What's priced in

Bullish cases hinge on sustained productivity gains. Some sell-side views expect U.S. productivity to trend from roughly 1.5% to ~1.9% in the early 2030s, which would justify today's data center buildout and software spend.

The hurdle: those returns must clear rising capital costs and massive depreciation. If unit economics disappoint or adoption slows, multiples and capex plans get marked down fast.

Equity shock vs. credit shock

History splits here. The dot-com bust primarily hit equities, the Fed eased, and the recession was mild (real GDP fell ~0.3%). The housing collapse impaired collateral across the system and credit contracted; GDP declined ~4.2%.

Today's AI cycle looks less credit-heavy on the surface-large cash piles and strong balance sheets help. But there are pressure points worth watching.

• Related-party deals: Circular investments between big platforms and customers can inflate perceived demand and valuations. If those unwind, revenue quality gets questioned.
• Debt for infrastructure: Data centers, power, and networking are capital intensive. If rates back up, debt service eats into free cash flow and capex flexibility.
• Nonbank fragility: Insurers, pensions, hedge funds, and private credit now hold roughly half of global financial assets with uneven oversight. Liquidity can vanish when redemptions hit.
• Leverage in the system: Reported margin debt hit a record ~$1.1T in September, up ~34% year over year. Forced deleveraging can turn a pullback into a slide.

FINRA's margin debt series is worth a permanent spot on your dashboard.

Macro context matters more this time

Policy friction is higher: tariffs are back in play, monetary policy independence is debated, and dollar confidence is a live topic. In that backdrop, a sharp equity drawdown would lean harder on sentiment and global funding markets.

One scenario analysis suggests a dot-com-sized correction could erase roughly $20T of U.S. household wealth (near 70% of U.S. GDP) and more than $15T for foreign investors (~20% of ex-U.S. GDP). That doesn't ensure a crisis-but it's a meaningful hit to balance sheets and spending.

For a broader view of nonbank risk and liquidity mismatches, see the IMF Global Financial Stability Report.

Three plausible paths

• Orderly repricing: -10% to -15% in AI leaders as earnings catch up, limited credit impact. Curves bull-steepen, credit spreads widen modestly, USD mixed.
• Bubble deflation: -20% to -30% broad market drawdown. Confidence shock cools capex; a moderate recession becomes likely. Duration rallies, quality outperforms, high beta and CCCs lag.
• Disorderly unwind: -40%+ in speculative tech with spillovers to private credit and NBFIs. Liquidity gaps, spread blowout, cross-border funding stress. Policy response becomes the swing factor.

Risk playbook for finance teams

• Map exposures: Direct AI equity, vendor concentration, partner equity, and venture links. Flag any revenues tied to related-party spending.
• Stress test: 20-30% equity drawdown, 150-250 bps HY spread widening, 75-125 bps IG widening, 100-150 bps rally in 10-year yields. Add a private credit liquidity stress.
• Liquidity stack: Stagger maturities, pre-fund near-term needs, expand committed lines, and keep high-quality collateral unencumbered.
• Leverage discipline: Tighten margin and prime financing haircuts for hard-to-value AI names. Monitor counterparty concentration in NBFIs.
• Hedge overlays: Index puts or put-spreads, collars on single-name high flyers, and dynamic vol targeting. Size to cash-flow-at-risk, not conviction.
• Quality barbell: Favor profitable, cash-generative tech and enablers (semis, power, networking) on one side; defensives with pricing power on the other. Keep pure-concept exposure small and funded.
• Credit hygiene: Re-underwrite covenants, EBITDA add-backs, and off-balance-sheet commitments linked to AI infrastructure. Scrub customer financing risk.
• Vendor and capex diligence: Validate data center utilization, power contracts, and delivery timelines. Watch for capex deferrals and canceled orders.
• Governance and disclosure: Track related-party transactions, revenue concentration, and capitalization of software costs.
• Communication plan: Align stakeholders on trigger levels, hedge usage, and cash priorities before volatility hits.

What would validate the bull case

• Clear productivity lift in non-tech sectors, not just Big Tech margins.
• Improving data center utilization and ROI as energy and cooling constraints ease.
• Evidence of cost takeout and revenue expansion from deployed AI in Fortune 1000 P&Ls.
• Capex efficiency: more throughput per dollar, not just bigger budgets.

Signals to monitor

• AI capex run-rate vs. free cash flow and net leverage at mega-cap leaders.
• Spread behavior: IG/HY OAS, CCC share of issuance, primary deal reception.
• Private markets: NAV marks, fund gates, amendment activity, warehouse line usage.
• Market plumbing: equity vol term structure, skew, dealer gamma positioning.
• Policy shocks: tariff headlines, fiscal standoffs, central bank guidance, dollar funding stress.
• Household balance sheets: wealth effect proxies and savings rates if equities slide.

Bottom line

AI can still be a long-term growth story and a short-term source of market friction. Position for both. Keep leverage contained, build liquidity, hedge asymmetry, and be picky on cash flows and governance.

If you're evaluating practical AI tooling for finance while you tighten risk, this curated list is a useful starting point: AI tools for finance.