CoreWeave's IPO: A House of Cards Built on GPUs and Microsoft?

The AI revolution has turned cloud infrastructure into a gold rush, and CoreWeave stands at the epicenter. The company's S-1 filing for its upcoming IPO reveals staggering revenue growth—surging from $16 million in 2022 to $1.9 billion in 2024—as it rides the wave of demand for AI compute power. Yet beneath the hype lies a precarious reality: its $35 billion IPO target assumes a near-perfect future where Microsoft remains its savior, hyperscalers don't undercut it, and NVIDIA's GPU supply never falters. For investors, the question isn't whether CoreWeave is part of the AI era—it's whether its valuation can survive the next storm.

The Illusion of Growth

CoreWeave's revenue explosion—1,346% in 2023 and 737% in 2024—has been fueled almost entirely by one customer: Microsoft. The tech giant contributed 62% of 2024 revenue, with two customers accounting for 77% of total sales. This extreme concentration creates a single point of failure. If Microsoft pivots to build its own data centers (as it has hinted) or switches to cheaper providers, CoreWeave's revenue could collapse overnight.

The IPO's valuation multiples only amplify the risk. At $35 billion, CoreWeave's market cap would be 18.4x its 2024 revenue. To justify this, the company must sustain exponential growth and achieve profitability—both unlikely bets. Even in a bull-case scenario (20x forward revenue), the math requires CoreWeave to nearly triple its revenue by 蕹2026, a pace no cloud provider has ever sustained for long.

The Hidden Liabilities

Beyond customer dependency, CoreWeave's financials are a minefield. Despite $1.4 billion in cash, it carries $7.9 billion in debt and $15.1 billion in Remaining Performance Obligations (RPOs)—long-term commitments to data centers and GPU purchases. These RPOs act as off-balance-sheet liabilities, locking CoreWeave into costly infrastructure even if demand evaporates. Compare this to WeWork's pre-IPO collapse, where similar contractual obligations sank its valuation.

The company's net losses, while shrinking as a percentage of revenue, remain massive: $863 million in 2024, or 45% of revenue. Operating leverage is nascent, and any slowdown in AI adoption—or competition from AWS or Google—could send losses soaring again.

The Investor Trap

CoreWeave's dual-class shares give founders 83% voting control, insulating them from shareholder pressure even as they cashed out $500 million pre-IPO. Meanwhile, preferred shareholders (including Blackstone and Magnetar) hold ratchets that could dilute common stockholders if the IPO price falters. After the 180-day lock-up period, insiders may flood the market, depressing the stock.

The base-case valuation—$20B–$30B (10–15x revenue)—still demands CoreWeave maintain high growth amid rising competition. NVIDIA's stock price, a proxy for GPU supply health, could also destabilize the narrative.

Verdict: Proceed with Caution

CoreWeave's IPO is a bet on two unproven assumptions: that AI compute demand will never wane, and that Microsoft will never turn its back. History suggests both are risky wagers. For retail investors, the IPO price embeds a “best-case” scenario that leaves little room for error. Institutional buyers with deep pockets might gamble on the upside, but the risks—customer concentration, hyperscaler competition, and a fragile balance sheet—are too great to ignore.

In the AI gold rush, CoreWeave's IPO may look like a shiny claim, but beneath the surface, the veins of its valuation are riddled with cracks. Unless you're prepared to double down on the idea that the AI boom is permanent, this stock is best left in the dust.