6 Hyperscalers Driving the AI Revolution Are Projected to Spend $1.3 Trillion on Capex in 2027. Only 1 Is Forecast to Have Positive Free Cash Flow.

S&P Global forecasts that AI-driven capex will rise from $870B this year to $1.3T next year (+~50% vs 2025) across six major hyperscalers, but only Microsoft is projected to generate positive free cash flow (FCF) in 2027. The model implies severe FCF pressure for peers, with Alphabet projected -$82.7B, Amazon -$60.1B, Meta -$3.5B, Oracle -$41.6B, and SpaceX -$114.4B in 2027. While Microsoft’s $329B in future lease obligations and AAA credit rating may mitigate concerns, investors remain skeptical that 2028 capex flattening will restore FCF and returns fast enough.

Analysis

This is less a fundamental sell signal on AI demand than a liquidity-tax on the equity story. The market has been willing to pay for growth while ignoring that the marginal dollar of AI infrastructure now earns lower near-term cash conversion than the rest of the business; that tends to hit valuation multiples first, then buybacks, then sentiment. The cleanest second-order winners are the suppliers that monetize spend immediately while deferring ROIC scrutiny — especially the GPU / network / power-chain complex — while the losers are the hyperscalers with the weakest ability to mask depreciation and lease drag in reported cash flow.

The setup becomes more dangerous if rates stay sticky or credit spreads widen, because these balance-sheet-heavy buildouts are duration assets funded with very short investor patience. Over the next 1-3 months, the catalyst is not capex itself but management commentary on 2027 spend, lease accounting, and incremental revenue-per-unit economics; any hint that spend is still accelerating faster than monetization should compress multiples again. Over 6-18 months, the real question is whether this becomes a buyback famine: if free cash flow stays pinned, equity support from repurchases fades exactly when the market is asking for proof of return on capital.

The consensus is missing that this may be more uneven than a blanket bearish call on "AI." Microsoft still has the best cushion because its capital structure gives it the most flexibility to hide cash burn, whereas the more aggressive builders are implicitly financing optionality at the expense of near-term shareholder yield. That makes the better trade a relative one: own the infrastructure beneficiaries and avoid the names where the market is paying a premium multiple for worsening cash conversion; the thesis breaks if capex inflects down sooner than expected or if AI monetization visibly reaccelerates enough to restore FCF inflection by mid-2027.

Key Decisions for Investors

• Go long NVDA against a basket short of AMZN/GOOG/META on a 3-6 month horizon; thesis is that suppliers still monetize the spend while hyperscaler equity gets penalized for cash conversion deterioration. Size modestly and trim if hyperscaler managements cut 2027 capex or show faster-than-expected AI revenue payback.
• Prefer relative long MSFT vs short AMZN or GOOGL into strength; MSFT has the cleanest cash-flow optics and the most room to preserve buybacks, while the others face higher risk of multiple compression as free cash flow narrows. Falsify if MSFT guidance implies lease-adjusted cash burn is worsening faster than peers.
• Watch ORCL CDS / credit spreads rather than chasing the equity; if spreads widen, it would confirm the market is finally repricing lease-adjusted leverage in the AI buildout and could create a cleaner entry to short the more leveraged software/cloud stack.
• If you want a lower-beta expression, buy the suppliers to data-center power and connectivity on any 5-10% pullback over the next 1-3 months; the first-order order book impact should arrive before investors fully digest the ROIC debate.
• Avoid outright shorting MSFT unless the stock breaks on a confirmed guide-down in FCF or capex normalization fails to appear by the next earnings cycle; this is the least asymmetric way to express the theme because the balance sheet and AAA rating reduce forced-deleveraging risk.