CoreWeave's 15% Yield Puts Its $35 Billion Debt in Focus

CoreWeave shares tumbled on Oct. 14, 2026, as yields on its bonds neared 15% on concern about AI borrowing. CNBC Yield here means the yearly return bond buyers demand to hold the debt.

That move put the focus on its borrowings. CoreWeave had $35 billion in debt, with interest expense driving steep losses. The Globe and Mail

Price action spilled into social media. Investor Chamath Palihapitiya posted a chart on X about CoreWeave's stock tumble. The Globe and Mail Elon Musk reacted with "Yikes" as CoreWeave's bond yield hit 15% amid Palihapitiya's private credit bubble warning. Yahoo Finance

Capital stack repricing

CoreWeave built that debt stack in public in April. On April 9, 2026, it announced a proposed $3.0 billion Convertible Senior Notes offering. CoreWeave Investors These are bonds that pay interest and can later convert into stock.

It then priced an upsized $3.5 billion Convertible Senior Notes offering with 1.75% annual interest payable semiannually beginning October 1, 2026. CoreWeave Investors Semiannually means in two payments per year.

Days later it added straight high-coupon debt. CoreWeave announced pricing of $1,000 million of 9.750% Senior Notes due 2031. CoreWeave Investors The notes have a maturity date of October 1, 2031. The company expected closing to occur on April 21, 2026.

That April 16 pricing meant CoreWeave tapped the US junk-bond market for a second time in a week. Bloomberg Junk bonds are lower-rated loans that pay higher interest for higher risk.

In my view, the structure explains the pain for savers and investors. A bond priced at 1.75% or 9.750% in April now trading at 15% means new lenders demand far more interest, like a mortgage resetting higher. That higher rate lowers the present value of future profits and raises the bill on $35 billion of debt.

A crowded primary

CoreWeave is not borrowing in isolation. Hyperscalers, data centers and chip companies are borrowing hundreds of billions of dollars to finance the AI buildout. Bloomberg Hyperscalers are the largest cloud providers.

The pipeline keeps printing. A CoreWeave-tied data center developer sponsored by affiliates of Blue Owl Capital raised $1.1 billion through a junk-bond offering. Bloomberg

Goldman Sachs data show gross debt issuance from hyperscalers is expected to hit a record $420 billion next year, up 60% from 2026 estimates. Reuters That supply meets a buyer base that has turned selective on AI paper.

The broader context here is how quickly price adjusts when supply surges. Borrowing windows can shut fast. The most indebted companies and longest building projects feel it first.

Footprint still expanding

Alongside the financing, CoreWeave kept adding capacity and credentials. It was selected for inclusion in the Nasdaq-100 Index 15 months after its IPO. CoreWeave

It entered India with AdaniConneX, planning a 240 MW deployment in Navi Mumbai. CoreWeave It announced new facilities extending its global buildout into Southeast Asia. CoreWeave It also announced a collaboration with Leidos to deliver secure, mission-ready AI capacity for national security. CoreWeave

In my view, those items read as background, not offset. Index inclusion changes the holder base. It does not change leverage. New megawatts add future revenue potential and near-term funding need in equal measure.

The broader context here is a classic late-cycle credit test. Lenders funded a capex supercycle with a mix of converts, high-yield and structured vehicles tied to specific data center assets. That works while utilization stays high and contracted cash flow covers interest. It reprices fast when investors start to ask who funds the next phase and at what coupon.

In my view, watch three transmission channels. First, refinancing. Paper priced at 9.750% in April trading toward 15% in October changes the marginal cost of the buildout. Second, covenants and collateral. Asset-level raises like the $1.1 billion Blue Owl-sponsored deal show how much of the AI trade now clears as secured project risk rather than parent unsecured risk. Third, equity sensitivity. A levered AI infrastructure model has high operating leverage. Small changes in funding cost feed directly into free cash flow math.

In my view, none of that resolves whether 15% is clearing level or overshoot. Credit rarely moves in straight lines into a heavy supply calendar. With hyperscaler issuance projected 60% higher next year, dispersion is likely to persist. Strong contracted cash flows will still print. Weaker structures will pay up.

In my view, for professionals, the signal is narrow. CoreWeave compressed a full credit cycle into six months. April was upsized converts and a second junk tap in a week. October is double-digit yields and a social-media-driven equity selloff. The debt funds the growth. The debt now prices the growth.