Introduction
CoreWeave (stock ticker: CRWV), a GPU cloud service provider focused on AI workloads, recently launched the CoreWeave Ventures fund to invest in early-stage AI startups. This move comes at a time of rapid expansion for CoreWeave: its Q2 2025 revenue reached $1.213 billion (a 207% year-over-year increase), with a total backlog of $30 billion. One might ask: in such a capital-intensive expansion period, why risk investing in startups? Is CoreWeave’s leverage getting too high? This article will delve into the logic behind CoreWeave’s investment strategy and assess the risks and rewards of its high-leverage operations.
Why Invest in AI Startups?
CoreWeave’s core business is providing high-performance GPU cloud computing resources (like NVIDIA H100 and GB200 chips) to help AI companies train models and run applications. However, relying solely on providing “hardware” infrastructure makes it easy to be surpassed by AWS, Microsoft Azure, or emerging competitors (like Nebius). Therefore, CoreWeave is investing in startups through CoreWeave Ventures not just for financial returns, but to build a solid AI ecosystem. The specific reasons include:
Locking in Long-Term Customers and Creating Demand: CoreWeave Ventures adopts a “compute-for-equity” model, providing startups with free or discounted cloud computing resources in exchange for equity. This allows startups to quickly test and scale their products on CoreWeave’s platform, creating “stickiness”—as these companies grow, they naturally become long-term customers of CoreWeave. For example, CoreWeave invested in the generative video AI company Moonvalley (participating in an $84 million seed round in July 2025), providing cloud resources to accelerate its scaling and secure future GPU usage demand.
Seizing the Initiative in AI Innovation: The AI market is in a phase of explosive growth, with US AI startups raising $104.3 billion in the first half of 2025. By investing, CoreWeave gains exposure to emerging technologies (like generative AI, distributed systems, and high-impact applications) and integrates them into its own platform. For example, investments in Chai (an AI chat platform) and Wombo (a lip-sync AI application) allow CoreWeave to position itself in emerging application scenarios ahead of time, strengthening its platform’s competitiveness. The 2025 acquisition of OpenPipe (a reinforcement learning AI company) directly enhanced CoreWeave’s AI development tool capabilities.
Building an Ecosystem Moat: By partnering with startups, CoreWeave is building a tightly integrated ecosystem, similar to the early fund strategies of OpenAI or Perplexity AI. These investments not only bring potential equity appreciation but can also enhance CoreWeave’s product line through technology licensing or acquisitions. CEO Michael Intrator stated that the goal of CoreWeave Ventures is to “accelerate innovation to market” while consolidating CoreWeave’s leadership position in the AI infrastructure market.
Why Still Invest During the Expansion Phase?
CoreWeave is in a period of rapid expansion: Q1 2025 revenue was $982 million (a 420% year-over-year increase), and Q2 reached $1.213 billion, with the full-year ARR expected to exceed $5 billion. They are heavily investing in data centers (33 facilities, 420MW of active power, 1.6GW of contracted power), with capital expenditures for 2025 projected to be $21-23 billion, including a $6 billion AI data center plan in Pennsylvania. This capital-intensive expansion might seem like a “distraction” from investing in startups, but it is a strategic necessity:
Creating Its Own Demand: Large-scale data center expansion means CoreWeave needs more customers to fill the new computing capacity. Investing in startups is like “planting seeds”; the future growth of these companies will bring stable and continuous GPU demand, reducing the risk of idle capacity. For example, a $11.2 billion contract with OpenAI shows that CoreWeave has already locked in major customers, but startup investments can further diversify its revenue sources.
Keeping Up with the AI Wave: The AI infrastructure market is expected to reach $1.5 trillion by the end of 2025, with the number of global data centers increasing to 1,136. If CoreWeave doesn’t build an ecosystem through investments, it could be overtaken by competitors (like Nebius, which secured a $19.4 billion contract with Microsoft). Investing in startups allows CoreWeave to work more closely with innovators, ensuring a technological lead.
This strategy is common in high-growth industries, especially when the AI infrastructure market is still immature. CoreWeave’s investments are for long-term competitiveness, not short-term speculation.
Is the Leverage Too High?
CoreWeave is indeed employing a high-leverage strategy, which is the norm in the AI infrastructure industry—high growth requires huge capital support. But is it “too high”? Here is an analysis:
Financial Leverage Overview: Since 2024, CoreWeave has raised a cumulative $25 billion (debt and equity), with total debt of about $11 billion and cash reserves of only $1.1 billion. The debt-to-equity ratio in March 2025 was 3.88 (2.60 in Q1), and the Altman Z-Score was only 0.98, indicating financial pressure. The projected interest expense for 2025 is $1 billion, accounting for nearly 20% of revenue, with financing rates as high as 15%. Additionally, $2.6 billion in operating lease liabilities (for 32 leased data centers) constitutes hidden leverage.
Reasons for High Leverage: CoreWeave’s business model is extremely capital-intensive: purchasing NVIDIA chips (spending $100 million on H100s in 2022) and building data centers require huge amounts of capital. Capital expenditures in Q1 2025 reached $1.9 billion, mainly repaid through contract financing. Although the leverage is high, revenue growth is rapid (up 737% in 2024), and cost growth (617%) is slightly lower, showing a preliminary leverage effect.
Risks and Buffers: High leverage comes with risks: high customer concentration (a single customer, likely Microsoft, accounts for 72% of revenue), and if AI demand slows, idle capacity could drag down cash flow. However, CoreWeave’s $30 billion backlog (including the OpenAI contract) provides strong support, and the backing of investors like NVIDIA also boosts confidence. Since its IPO in March 2025 (at $40), the stock price has risen over 150%, but it is highly volatile (peaking at $187 in June, now around $55). The average analyst target price is $90.20, indicating that the market remains optimistic.
Conclusion
CoreWeave’s investment in AI startups through CoreWeave Ventures is not a blind risk but a move to build an ecosystem, lock in long-term customers, and seize the initiative in innovation. This is particularly important during the expansion phase to ensure stable demand for new GPU capacity and consolidate its market leadership. Although high leverage (a debt-to-equity ratio of 3.88) brings financial pressure, strong revenue growth and a substantial order backlog provide a buffer. Investors should pay attention to the IPO lock-up expiration on August 14 (unlocking 83% of Class A shares, which could pressure the stock price) and the Q3 earnings report to assess its long-term potential.
CoreWeave’s strategy is a classic example of high risk, high reward. They are trying to stand out in the AI infrastructure “gold rush,” and CoreWeave Ventures is a key step in this race.
About Grok
由 X AI 開發的大型語言模型。