Man Group Sees ‘Bubble Risks’ as AI Bond Sales Break Records

Man Group says “bubble risks” are mounting as AI-related bond issuance surges, with particular concern in high-yield and leveraged loan markets. The firm warns that enthusiasm for AI infrastructure has encouraged significant overreach, especially among free-cash-flow negative borrowers. The note is a cautionary read-through for credit investors rather than an immediate catalyst for broad market moves.

Analysis

The market is rewarding AI capex as if all growth is created equal, but credit is finally starting to discriminate between hyperscaler-backed infrastructure and the lower-quality layer of the financing stack. The near-term winners are still equipment suppliers and contractors with hard assets and order visibility; the losers are the marginal borrowers funding buildouts before revenue ramps, where refinancing risk can become the real bottleneck rather than demand. That matters because once the funding window tightens, project completion risk feeds back into delayed compute deployment, which can compress expected returns across the broader AI ecosystem.

The second-order issue is that credit excess typically shows up first in secondary spreads, then in new issue pricing, and only later in defaults. If leveraged loan and high-yield appetite cools over the next 1-3 months, the most vulnerable names will be those with long-dated monetization, weak sponsor support, and asset-light business models that cannot be pledged as collateral in a downturn. Expect dispersion to rise inside AI-adjacent credit: top-tier issuers with contracted cash flows should still clear the market, while smaller, free-cash-flow-negative borrowers may be forced to pay up materially or delay issuance.

The contrarian view is that bubble language may be early for the core AI buildout but late for marginal credit. In other words, the technology demand curve may still be intact, yet the financing curve can break first. That creates a tactical opportunity to fade the weakest paper without shorting the entire AI theme, especially if equity investors are still extrapolating hardware demand while ignoring the cost of capital moving higher for non-strategic issuers.

The key catalyst is not an AI demand miss; it is a failed refinancing, widening CLO loan spreads, or a visible concession in one or two high-profile deals that resets the market's tolerance for leverage. Once that happens, issuance could slow for quarters, forcing a repricing of AI infrastructure timelines and increasing the bargaining power of cash-rich incumbents over capital-dependent entrants.

Key Decisions for Investors

• Short the weakest AI-adjacent high-yield / leveraged loan credits on any new-issue concession over the next 1-3 months; prioritize free-cash-flow-negative issuers with no near-term path to self-funding. Target 200-400 bps spread widening; stop out if primary market demand remains oversubscribed.
• Long a quality-vs-junk credit dispersion trade: buy bonds of hyperscaler or utility-backed infrastructure issuers and short lower-rated private-capital-funded AI borrowers. Best risk/reward if market starts to price refinancing risk rather than headline AI demand.
• Use downside hedges on broad credit proxies via CDX HY protection or HYG puts for a 1-2 quarter window. The convexity is attractive if one failed deal triggers a broader re-rating of the asset class.
• Stay long AI capex beneficiaries with strong balance sheets and hard-asset exposure, but avoid extending duration into names reliant on repeated capital raises. The market may punish financing risk before it slows end-demand.
• Set a trigger to add to shorts after the first visible deal failure or spread shock in the leveraged loan market; that is likely the point where investor sentiment shifts from 'bullish growth' to 'capital scarcity.'