CoreWeave CEO Mike Intrator: We are very confident in our revenue for 2026

Key Concepts

• Unit-Level Economics: The profitability and efficiency of individual business units or service contracts, which the company uses as a primary indicator of long-term health.
• GAAP EPS (Earnings Per Share): The accounting standard-based measure of profitability, which the company argues is currently distorted by tax impacts and scaling costs.
• Inference vs. Training: The shift in data center compute usage from "training" AI models to "inference" (running the models to provide answers/results).
• Operating Leverage: The ability of the company to increase revenue at a faster rate than costs as they scale, expected to accelerate significantly by late 2026.
• Debt Market Underwriting: The process by which debt investors evaluate the company’s "building block" deals, viewing them as high-quality, investment-grade assets.

---

Financial Performance and Market Outlook

The company addressed concerns regarding its GAAP EPS loss of $1.40 (compared to the expected $1.18), emphasizing that investors should not "overindex" on this single figure.

• Accounting vs. Reality: Management argues that the reported loss is largely a function of tax impacts and the physical costs of scaling, rather than poor unit-level economics.
• The 2026 Inflection Point: The company reaffirmed its 2026 revenue and operating margin targets. They anticipate an "enormous acceleration" of operating leverage starting in late 2026, with a noticeable inflection point occurring in Q3 and Q4 of the current year.
• Debt vs. Equity Markets: A key argument presented is that debt markets are currently more "dialed in" to the company’s success than equity markets. Because debt investors focus on the underlying "building block" deals—which are projected to hit mid-20s margins—they view the company’s debt as high-quality, investment-grade paper.

Data Center Evolution and Compute Demand

The discussion highlighted a significant shift in how compute resources are utilized within data centers.

• The Shift to Inference: More than 50% of the company’s compute capacity is now dedicated to inference rather than training.
• Workflow Complexity: The transition to generative AI tools has changed compute requirements. Modern AI workflows often involve "chains" of requests that are significantly more compute-intensive than a "single-shot completion."
• Demand Vectors: The company is experiencing a "wall of demand" from major industry players, including Anthropic, OpenAI, and Perplexity. This demand is driven by existing customers returning to expand their capacity, citing the superior quality of the company’s product.

Strategic Perspective

The company maintains a high level of confidence in its growth trajectory. The core strategy involves:

1. Scaling Infrastructure: Accepting short-term GAAP losses to build the necessary physical infrastructure to meet the massive, multi-vector demand for compute.
2. Focusing on Unit Economics: Prioritizing the profitability of individual contracts, which serves as the foundation for the company's long-term financial stability.
3. Leveraging Market Confidence: Using the validation from debt markets to signal to equity investors that the underlying business model is robust and capable of delivering significant future returns.

Conclusion

The company is currently in a capital-intensive "ramp" phase. While GAAP earnings appear negative, management asserts that this is a temporary byproduct of rapid scaling. By focusing on unit-level economics and the shift toward high-demand inference workloads, the company expects to see a major financial inflection point by the end of 2026, supported by strong demand from leading AI organizations.