AI financing fueling a surge in U.S. convertible bond sales

U.S. convertible issuance reached about $34 billion in the first four months of 2026, more than double the same period a year earlier, and is on pace to exceed last year’s record of over $120 billion. Roughly half of issuance is tied to AI, with large deals from Oracle ($5 billion), CoreWeave ($4 billion), and IREN ($2.6 billion), as companies use cheap convertible funding for data centers, cloud expansion, and refinancing. Elevated Treasury yields and strong demand from hedge funds are keeping convertibles attractive despite higher-rate conditions.

Analysis

The convert market is functioning as a pressure valve for the AI capex cycle: it lets issuers finance multi-year infrastructure without immediately forcing either expensive straight-debt coupons or equity dilution. That matters because the buyers are effectively underwriting both credit and upside convexity, which means the cheapest capital is now flowing to the names with the highest expected volatility rather than the cleanest balance sheets. Second-order effect: this can keep AI spending elevated for longer than the cash flow cycle would normally allow, delaying the point where hyperscaler and infrastructure capex has to slow.

The hidden winner is not just the headline issuers, but the ecosystem that can repeatedly tap convert demand at attractive terms—data center operators, power-linked utilities, and semiconductor beneficiaries with high beta. However, this also creates a crowded financing channel: if convert issuance keeps accelerating, the market will likely force a widening in credit spreads for lower-quality AI-linked credits even as equity traders still chase the theme. That bifurcation should favor stronger balance sheets while punishing marginal names that are being valued more on narrative than on financing durability.

The biggest risk is that investors are extrapolating present volatility and AI enthusiasm into a permanent funding subsidy. If equity volatility compresses or AI capex growth slows over the next 3-6 months, convert demand can reprice quickly because the embedded option becomes less attractive, pushing issuers back toward harsher terms. A second risk is maturity-wall refinancing: the wave of pandemic-era paper rolling over may reveal which companies were only viable under near-zero rates, creating sudden credit differentiation.

Contrarian take: the market may be overestimating how much of this issuance is genuine growth funding versus balance-sheet extension. That distinction matters because refinancing-driven convert supply does not create incremental enterprise value; it only delays dilution or default. In the near term, the tradeable edge is to own the issuers with real AI operating leverage and short the lower-quality “convert-dependent” names that are using the market as a crutch.

Key Decisions for Investors

• Go long CRWV vs short a basket of weaker AI-funding names on any 1-2 week post-issuance drift lower; the relative trade favors names where capital raised translates directly into contracted compute demand rather than speculative capacity build-out.
• Buy ORCL on 1-3 month pullbacks only if convert spreads remain tight; upside is durable because it can finance growth cheaply, but the risk/reward worsens if the market starts demanding straight-debt economics again.
• Short WYFI on a 3-6 month horizon against a stronger infrastructure/AI beneficiary; this is the cleanest expression of financing-risk dispersion if convert appetite cools and investors stop paying up for growth-at-any-price.
• Use long TEM calls or call spreads over 2-4 months as a convex play on the market rewarding zero-coupon, equity-linked funding structures; risk is that the stock stalls before conversion math improves.
• Avoid chasing ON, MCHP, and DUK after the convert bid; these are lower-beta ways to express the theme, but the asymmetric upside is already being harvested by convertible buyers, not equity holders.