Big Tech's massive AI spending is putting one of its longtime strengths to the test

AI capex is pushing Big Tech’s hyperscalers toward outside funding: capital spending is now above 100% of operating cash flow for some firms, coinciding with sharp free-cash-flow deterioration (Meta -91% y/y to FCF $31.1B capex; Alphabet first-ever quarterly cash outflow since 2004; Amazon FCF outflow $8.8B with $54.21B capex). Rating risk is rising at the margin—S&P Global warns credit metrics are moving closer to downgrade thresholds (notably Meta’s lease commitments ballooning to ~$280B vs ~$180B), while S&P downgraded Oracle to the lowest investment-grade tier (BBB-) and flagged “threaten credit quality” risk from “unprecedented” AI investment. Investors are still focused on the funding path and the return on AI spend, with Wall Street expecting an AI payback inflection around 2028; failure to see that payoff could “create more serious credit issues.”

Analysis

The near-term loser is not just the hyperscalers’ equity multiples; it is their capital-return model. As free cash flow gets absorbed by buildout, the first-order market effect is lower buyback support, but the second-order effect is multiple compression because investors will start capitalizing AI spend like industrial capex rather than a software-like annuity unless monetization shows up quickly. Microsoft is the cleanest relative winner inside the group because its balance-sheet optionality lets it keep funding growth without forcing the market to price in a financing overhang.

The financing ecosystem is a secondary beneficiary. BLK, BX, and OWL can monetize the wave through origination, asset management, and structured-data-center exposure, but that only works while spreads remain tight and the assets are viewed as investment grade-like. If the AI cycle needs progressively more off-balance-sheet funding, that is usually a late-cycle tell: returns are not yet covering the cost of capital, and lenders will demand wider spreads or better collateral, which pushes risk back onto equity holders.

Contrarian view: the market may be too focused on leverage and not enough on timing. This looks more like a 12-24 month cash bridge than an imminent solvency event, so default risk is not the trade; dilution of equity compounding is. ORCL is the most vulnerable because it has the least room for a financing hiccup, while META/AMZN/GOOGL are more about FCF impairment than credit stress. What would falsify the bearish thesis is a visible step-up in external AI revenue or capacity rental that offsets capex by the next 2-3 quarters; absent that, the crowding-out of buybacks should stay a drag into the next earnings season.

Key Decisions for Investors

• Long MSFT / short ORCL as a 3-6 month relative-quality trade; MSFT has the best financing cushion, while ORCL is the cleanest expression of downgrade/refinancing risk if AI spend stays debt-funded. Stop if ORCL generates sustained positive FCF or bond spreads retrace materially.
• Trim META and AMZN on strength versus MSFT into the next earnings cycle; the equity risk is not bankruptcy, it is a slower buyback machine and a higher cost of capital. Falsifier: management guides capex growth below revenue growth for two consecutive quarters.
• Buy BLK or BX on pullbacks as a financing-pipeline beneficiary over the next 1-3 months; these platforms earn from underwriting/structured capital formation so long as hyperscaler issuance keeps coming. Reduce if AI issuance pauses or credit spreads widen sharply.
• Add OWL as a tactical long only if data-center financing announcements continue; it is a cleaner way to own the private-credit wedge to the buildout. Risk/reward worsens quickly if lenders start demanding higher equity checks from hyperscalers.