Dan Nathan: CoreWeave IPO is the 'Exact Wrong Time' to Test AI Stock Demand

CoreWeave IPO Concerns

• ⚠️ Dan Nathan expresses extreme caution, stating he "wouldn't buy this with my worst enemy money," highlighting significant risks with the CoreWeave IPO.
• 💡 The company's business model involves long-term leases for data centers and raising capital to purchase GPUs, primarily the previous generation.
• 📉 These GPUs are depreciated over six years but are expected to be obsolete in approximately two years, creating an accounting risk if faster depreciation is required.

Financial and Customer Risks

• 🏦 CoreWeave has substantial debt and relies on continuous debt raises for growth, with limited success in raising equity during the IPO.
• 🎯 A major concern is customer concentration, with Microsoft accounting for 62% of sales, posing a risk if Microsoft decides to cancel leases due to overcapacity in their own data centers.
• ⚡ Microsoft's leasing from CoreWeave is likely not for their most critical workloads, making them a potential early canceller.

Investing in the Generative AI Trade

• 🚫 Nathan argues that investing in CoreWeave is the least innovative way to participate in the generative AI trade within public markets.
• 🚀 He contrasts CoreWeave with innovative private companies like OpenAI, Anthropic, and Cohere, which are customers of data centers and likely to gain more pricing power.
• 📈 Nathan believes there is currently no real capacity constraint in the market and that CoreWeave may have overbuilt.

Market Sentiment and Shorting Potential

• 📉 Public market sentiment for AI-related stocks is negative, with hyperscalers down at least 20% and Nvidia down nearly 30%.
• 💰 Nathan has heard from equity capital markets groups that investors are not buying the stock, and some are looking to short it if it pops.
• 📊 He anticipates that major investment banks like Goldman, JP Morgan, and Morgan Stanley will likely initiate coverage with buy ratings, but he would be shocked if this were the case.