CoreWeave expanded its OpenAI agreement in September 2025, which brought the total contract value to approximately $22.4 billion. This expansion added $6.5 billion to the existing commitments. Before this, the companies signed an $11.9 billion agreement in March 2025 and a $4 billion expansion in May 2025. These specific OpenAI contracts help bolster the massive $104 billion revenue backlog CoreWeave reported as of June 30, 2026. You already know that CoreWeave rents GPU clusters to AI companies, but these specific OpenAI commitments anchor the company’s massive growth. The company reported Q2 2026 revenue of $2.575 billion, which marked a 112% increase over the $1.212 billion reported a year earlier. This revenue growth accompanied a net loss of $626 million for the same period. Furthermore, the company added more than $25 billion of net new commitments in early Q3 2026. This influx of demand pushes the total revenue backlog toward unprecedented levels, well beyond the $99.4 billion reported at the end of Q1 2026.
The financing terms for these massive commitments changed significantly when CoreWeave closed its $2.6 billion delayed draw term loan facility (DDTL 5.5) in August 2026. Lenders no longer provide the investment-grade terms seen in the $8.5 billion DDTL 4.0 facility from March 2026. In that March transaction, the facility received ratings of A3 from Moody’s and A (low) from DBRS. Instead, the new DDTL 5.5 facility received lower ratings of Ba2 from Moody’s and BB+ from Fitch. Lenders demanded a cash lockbox and a 1.35x debt-service-coverage covenant to ensure they receive payments before CoreWeave uses its cash. The interest rate for the new facility hit SOFR + 550, and the deal requires full amortization, which indicates that lenders demand much tighter control over the cash flows generated by these AI clusters. This transaction follows a $3.1 billion DDTL 5.0 facility completed earlier this year. While the $2.6 billion deal appeared oversubscribed, it only closed after CoreWeave accepted more restrictive terms and a higher cost of capital.
| Facility Type | DDTL 4.0 (March 2026) | DDTL 5.5 (August 2026) |
|---|---|---|
| Total Amount | $8.5 billion | $2.6 billion |
| Moody’s Rating | A3 | Ba2 |
| Fitch Rating | A (low) | BB+ |
| Interest Rate | 5.9% fixed tranche | SOFR + 550 |
| Maturity | March 2032 | ~5 years |
I consider the mismatch between the five-year loan maturity and the three-year average length of customer contracts the most significant risk. If customers do not renew their contracts, CoreWeave must still pay back the debt without the guaranteed income required to cover it. Concentration risk also looms large because Anthropic provides 40% of the collateral for the loan and Jane Street accounts for 35%. This concentration remains a concern for lenders who must underwrite the credit of unrated private companies. The company reported a net loss of $626 million for the second quarter of 2026 as it spent $9.4 billion on capital expenditures. This massive spending helps the company build out the capacity needed for its $104 billion backlog. Additionally, the company reported $1.393 billion in depreciation and amortization alongside $640 million in interest expense during the same quarter. Total scheduled debt principal maturities reached $35.551 billion as of June 30, with approximately $10.597 billion due in 2026 and 2027. The company also reported an adjusted EBITDA of $1.510 billion, or a 59% margin. Can CoreWeave successfully transition from these high-cost, high-risk debt facilities to more stable, long-term financing as it scales? I recommend watching the relationship between the interest expenses and the revenue recognition from these OpenAI commitments.
