Debt-Funded AI Is Here. That's A Signal The Easy Tech Trade Is Fading
Debt is creeping into AI's buildout-and that's not a small footnote. Morgan Stanley Wealth Management CIO Lisa Shalett says the shift from cash-funded AI to private credit is a tell: the simple bull story in mega-cap tech is losing momentum.
"What was a very simple story is suddenly getting a lot more complex," she said. Expect investors to push harder for proof that these billions are earning their keep.
The Meta-Blue Owl deal: a new template for AI capex risk
Consider Meta's roughly $30 billion, debt-financed AI data center in Louisiana. According to reporting, it's the largest private credit deal on record. An SPV will own the site, with Blue Owl Capital controlling 80% and Meta keeping 20%. Off-balance-sheet. Morgan Stanley was the bookrunner.
That's a departure from the recent past, when Big Tech paid for AI infrastructure with free cash flow. Now private credit sits in the capital stack, and the bar for returns just went up.
Why this matters to finance teams and investors
- Cost of capital is now a core input to AI ROI. Debt service meets uncertain monetization timelines.
- Off-balance-sheet structures mask leverage. Read the SPV and lease terms, not just the parent's financials.
- Utilization risk is real. Data center capacity, GPUs, and power contracts must translate into durable revenue.
- Vendor concentration and cross-deals can distort pricing, margins, and reported growth.
- Private credit terms and covenants can force hard choices if assumptions slip.
Shalett's take is blunt: once you bring in private-credit players and shadow banking money, "it just gets harder to track, and it gets more and more unwieldy," which increases pressure to actually earn the return. For equity holders, that means fewer free passes on "build now, monetize later."
The circularity problem: cross-deals and vendor financing
Shalett also flagged the tight web between AI platforms, chipmakers, cloud providers, and startups. Recent deal flow "starts to feel and look and smell like circular dealing, like vendor financing." She's not alleging bad actors-just raising a practical issue: circular flows can inflate activity without improving underlying economics.
- Watch for prepayments, credits, and take-or-pay obligations that lock in capacity without revenue certainty.
- Scrutinize related-party exposure, side letters, and multi-year purchase commitments.
- Model scenarios where subsidies or credits fade and pricing normalizes.
What could break: the "accident" scenario
"The accident could be that somehow OpenAI doesn't actually develop a true revenue model to pay for all this capacity that they've committed to buy from everybody," Shalett said. If a major buyer under-monetizes, capex-heavy vendors and lenders feel it quickly.
In that case, she sees a possible 10% to 20% drawdown in the S&P 500. Not a doomsday call-just acknowledging that one weak link can ripple through a tightly connected chain.
Her market view: up, but slower-and choppier
Shalett still expects stocks to "grind higher," but not in a fresh, accelerating boom. The legs are "getting weaker and weaker" as debt, cross-dealing, and profitability questions pile up. Translation: positive trend, rising dispersion, and shrinking room for sloppy capital allocation.
What to track next (and how to underwrite it)
- Unit economics: gross margin impact per AI feature or product; attach and retention rates; incremental ARPU.
- Capacity utilization: GPU hours, inference/compute mix, data center fill rates, and power costs versus pricing.
- Contract quality: take-or-pay minimums, step-up clauses, and termination rights across SPVs and leases.
- Cash conversion: free cash flow after capex; maintenance vs. growth capex split; debt service coverage.
- Supplier risk: concentration to a handful of chip or cloud vendors; potential for price resets.
- Accounting signals: consolidation triggers, off-balance-sheet obligations, and disclosure quality.
Tactical implications for portfolios
- Favor cash-rich operators with clear AI monetization over capex-first stories funded with leverage.
- Reward proof, not promises: tie valuation work to ROIC on AI deployments within defined timeframes.
- Use catalysts: earnings with AI revenue details, contract disclosures, and power procurement updates.
- Risk controls: consider selective hedges around names with heavy capacity commitments and unclear demand.
Morning market snapshot
- S&P 500 futures: +0.62% (prior close +0.58%).
- STOXX Europe 600: flat early.
- FTSE 100: flat early.
- Nikkei 225: +1.35%.
- CSI 300: +1.13%.
- KOSPI: +2.5%.
- NIFTY 50: +0.08%.
- Bitcoin: $111K.
Sources and further reading
- Report on the Meta-Blue Owl private credit deal (Wall Street Journal)
- SEC guidance on off-balance-sheet arrangements
For finance teams building AI fluency
If you're evaluating AI ROI and want a quick scan of practical tools used in finance, here's a curated list: AI tools for finance.
Your membership also unlocks:
