Apollo Flags a Quiet Repricing of Big Tech Credit Risk
TREE NEWS reports: Apollo Global Management’s chief economist Torsten Slok is warning that credit markets have begun to reprice the risk embedded in the debt-fueled artificial intelligence buildout, even as equity investors appear to be looking the other way. The spread between credit default swaps on hyperscale cloud providers and those on banks has widened from roughly zero in October 2025 to about 60 basis points, a move Apollo describes as a systematic deterioration in the sector’s credit fundamentals.
Slok explicitly ruled out the most benign explanation for the move — dealers hedging new bond issuance. If that were the driver, bank CDS spreads would have widened in tandem, since banks remain the single largest source of investment-grade supply. Instead, bank spreads have held near 40 basis points, barely budging. The divergence points to an independent repricing of hyperscaler credit risk rather than a technical supply effect.
The Cash Flow Picture Behind the Spreads
The widening spreads are not occurring in a vacuum. FactSet data on forward-looking balance sheet metrics shows three of the four largest hyperscalers running deeply negative free cash flow:
- Alphabet: forward debt-to-equity of 13%, forward free cash flow of negative $25.7 billion
- Amazon: debt-to-equity of 23%, free cash flow of negative $30.0 billion
- Meta: debt-to-equity of 34%, free cash flow of negative $25.7 billion
- Microsoft: debt-to-equity of just 7.34%, free cash flow of positive $33.4 billion — the clear outlier
The pattern supports Apollo’s core thesis: hyperscalers are leaning on debt to sustain extraordinarily high capital expenditure, while the return profile on AI infrastructure remains highly uncertain and depreciation on those assets is already a drag. When free cash flow turns negative, the funding mix shifts toward borrowed money, and credit investors begin to demand compensation for that risk.
Why This Matters for Markets
Credit markets are typically the more sensitive early-warning system. The independent widening of hyperscaler CDS means bond investors have effectively cast a vote on the sustainability of AI capex before equity markets have done so. Since late June, hyperscaler equities have decoupled from their credit risk trajectory — stocks have held up while spreads have widened. That gap is the crux of the warning.
If credit is right, equity multiples on the largest technology companies may be vulnerable to compression, particularly for names with the weakest cash flow profiles. A sustained widening in CDS could also raise the cost of new debt issuance, creating a feedback loop that makes the capex cycle more expensive to finance. For bondholders, the risk is straightforward: more leverage, negative free cash flow, and uncertain asset returns are a recipe for wider spreads and mark-to-market losses.
There is also a demand-side risk. Several leading developers of frontier large language models have recently signaled a desire to slow the pace of product iteration for safety reasons. If those intentions translate into action, the commercial outlook for the cloud providers hosting those workloads would weaken — deepening credit concerns and potentially validating the bond market’s caution.
For crypto and commodity markets, the transmission is indirect but real. A repricing of AI-related credit could weigh on risk appetite broadly, pressuring high-beta assets including digital tokens, while a slowdown in data-center buildout would soften demand for power, copper, and other infrastructure-linked commodities. Currencies could feel it through shifting rate expectations if tighter financial conditions spill into broader credit markets.
Key Takeaways for Investors
- Hyperscaler CDS spreads have widened to roughly 60 basis points over bank CDS, a move Apollo attributes to genuine credit repricing rather than issuance hedging.
- Three of the four largest cloud providers show negative forward free cash flow, with Meta carrying the highest debt-to-equity at 34%.
- Microsoft stands apart with positive free cash flow and the lowest leverage, making it the relative safe haven within the group.
- Equities have decoupled from credit since late June; if the bond market is correct, tech multiples face compression risk.
- Watch AI model developers’ stated plans to slow iteration — a real slowdown would hit hyperscaler revenue and reinforce the credit warning.
