AI credit markets at risk of 'violent' correction and investors must stay clear-eyed, Man Group says

Man Group warns that rapid AI-related bond issuance is creating mispriced credit risk, especially in high-yield and leveraged loan markets where many borrowers remain free-cash-flow negative. The firm favors rigorous credit selection and diversification into Europe, emerging markets, Japan, and Hong Kong, while remaining moderately constructive on EM currencies. Key risks include crowded speculative positioning and a more hawkish-than-expected Federal Reserve.

Analysis

The key second-order issue is not simply “AI capex is rising,” but that credit is now financing an equity-duration bet without owning any of the convexity. That creates a structurally poor underwriting mix: lenders are left with downside from delayed monetization, while the strongest balance sheets can still outspend peers and widen competitive moats. In practice, this tends to bifurcate the market into a small set of AI-adjacent winners with cheap access to capital and a larger tail of infrastructure/adjacent borrowers whose spreads stop compensating for technological obsolescence risk.

The more interesting setup is in the transmission from public credit weakness to private-market valuation resets. If leveraged borrowers need to refinance into a less forgiving tape over the next 6-18 months, the first casualties are likely to be sponsors and growth equity owners who underwrote to uninterrupted funding and fast deployment curves. That matters for cloud, networking, power, and data-center supply chains: vendors with concentrated exposure to a few hyperscaler customers can see order deferrals before default rates show up, so the equity market may initially misread slowing spend as a benign digestion phase rather than a funding constraint.

The EM comment is a useful portfolio clue: if U.S. rates stay sticky and AI credit absorbs more global risk budget, the relative value case for non-U.S. carry and credit improves, especially where spreads still reflect idiosyncratic fundamentals rather than crowded AI enthusiasm. Japan and Hong Kong screening well likely reflects cleaner balance sheets and less commodity beta, while China/Indonesia remain vulnerable to being “cheap for a reason” if growth or commodity signals deteriorate. The contrarian point is that AI bond issuance could remain a funding advantage for the strongest platforms for longer than skeptics expect; the market may be early on credit stress but late on dispersion, so the right trade is not a blanket short-credit stance but a long-quality/short-funding-fragility expression.

Key Decisions for Investors

• Short the weakest AI-adjacent credit via HYG or JNK puts over the next 3-6 months; best payoff is if issuance keeps clustering in low-quality borrowers and spreads gap wider on a single refinancing scare.
• Pair trade: long HYG-quality defensive cash flows via senior secured industrial/utility credits, short CCC-heavy loan exposure through SJNK or a leveraged-loan ETF; target a 100-150 bps spread widening differential if AI capex funding tightens.
• Add relative-value exposure to Japan/Hong Kong credit over China/Indonesia through regional credit funds or sovereign/corporate mix; hold 6-12 months, as cleaner balance sheets should outperform if U.S. rates stay restrictive.
• Long U.S. dollar vs select EM FX basket if Fed hawkishness re-prices; use EM FX proxies or local-currency sovereign debt hedges, with a 2-3 month catalyst window around U.S. inflation and Fed communication.
• For equity hedging, buy downside on AI supply-chain enablers with concentrated customer exposure instead of the hyperscalers themselves; this is the highest beta way to express a slowdown in AI-funded capex without fighting the winners directly.