Research Finder
Find by Keyword
Putting the CoreWeave IPO Under the Microscope
What the largest GPU centric IPO means for the trajectory of the AI trade and the market as a whole.
The News
CoreWeave, an AI cloud computing startup backed by Nvidia, raised $1.5 billion in its IPO on March 27, 2025, pricing shares at $40—below the anticipated $47-$55 range—valuing the company at $23 billion. The downsized offering, initially targeting up to $2.7 billion, reflects investor caution amid stock market volatility and concerns over CoreWeave’s $8 billion debt and reliance on Microsoft for 62% of its revenue. Nvidia anchored the IPO with a $250 million order, and OpenAI secured an $11.9 billion deal. Shares begin trading on Nasdaq under "CRWV" on March 28, testing AI market enthusiasm. Read MarkeWatch’s coverage of the IPO here.
Analyst Take
The initial public offering (IPO) of CoreWeave, has been billed as one of the most anticipated market debuts of 2025. With heavyweight backers like NVIDIA and marquee clients such as Microsoft and IBM, the company’s pivot from cryptocurrency mining to GPU-powered AI infrastructure has captured Wall Street’s imagination. CoreWeave’s ambitions have hit a speed bump. Just days before its Nasdaq debut under the ticker “CRWV” on March 28, 2025, reports indicate a downsized raise of $1.5 billion at $40 per share, a notable retreat that has sparked questions about its prospects. What’s driving this shift, and should investors be excited or wary? Let’s dissect the CoreWeave IPO under the microscope.
A High-Stakes AI Play
CoreWeave’s journey began in 2017 as Atlantic Crypto, mining Ethereum until the 2018 crypto crash forced a strategic pivot. By 2020, it had relaunched as a cloud provider, leveraging its GPU expertise to serve the burgeoning AI market. Today, with 32 data centers, over 250,000 GPUs, and a client roster including Meta, Cohere, and Mistral AI, CoreWeave positions itself as a purpose-built AI hyperscaler. Its revenue reflects this transformation: a modest $16 million in 2022 ballooned to $228.9 million in 2023 and hit $1.9 billion in 2024, a staggering 737% year-over-year growth. The AI infrastructure market, projected to swell from $79 billion in 2023 to $399 billion by 2028 (Bloomberg Intelligence), offers a tantalizing backdrop for this growth story.
Yet, beneath the surface, CoreWeave’s financials reveal a company sprinting to keep pace with its own ambitions. Net losses have widened alongside revenue, reaching $863 million in 2024, with an accumulated deficit of $1.5 billion. While adjusted EBITDA paints a rosier picture at $1,219 million, the “D” (depreciation) and “I” (interest) loom large, hinting at a business model heavily reliant on accounting adjustments rather than cash profitability. This tension—explosive growth versus persistent losses—sets the stage for both opportunity and risk.
The Dependency Dilemma
CoreWeave’s business hinges on a web of dependencies that amplify its potential but also its fragility. First, there’s NVIDIA, its primary GPU supplier and a key investor. CoreWeave’s $12.9 billion in debt commitments, including a $7.6 billion GPU-backed facility led by Blackstone and Magnetar, is a masterstroke of financial engineering—borrowing from NVIDIA to buy its GPUs, then using those assets as collateral. This creates extraordinary operating leverage, but it’s a double-edged sword. Any disruption in NVIDIA’s supply chain or a drop in GPU value could unravel this delicate structure.
Then there’s Microsoft, which accounted for 62% of CoreWeave’s 2024 revenue. For any MSP this represents huge concentration risk. This concentration stems from a 2023 partnership to support OpenAI’s ChatGPT-driven demand, but it ties CoreWeave’s fate to a single client’s whims. If Microsoft shifts workloads to its own Azure platform or alternative XPUs (e.g., AMD or Intel GPUs), CoreWeave could face a revenue cliff. Add to this its reliance on third-party data centers and the scarcity-driven GPU market, and the company’s foundation looks less like bedrock and more like a house of cards.
Financial Model: Genius or Insanity?
CoreWeave’s financial strategy is as bold as it is unconventional. Its S1 filing reveals a company that “makes money” on an EBITDA basis—$1,219 million in 2024—but the devil is in the details. Depreciation, tied to its massive GPU fleet, and interest expenses from its debt load dominate the picture. The company’s $15.1 billion in remaining performance obligations signals robust future revenue, but its $16.5 billion in liabilities dwarf its $17.8 billion in assets, leaving a $413 million stockholders’ deficit. This isn’t a traditional cash-flow-positive tech darling; it’s a high-stakes bet on sustained AI demand and asset value stability.
The IPO itself reflects this tension. The original $2.7 billion target, valuing CoreWeave at up to $32 billion, has shrunk to $1.5 billion—a 44% haircut that suggests market jitters. Analysts like have flagged red flags, including limited disclosure under emerging growth company status and a valuation that may not hold at $27 billion post-IPO. CoreWeave’s ability to raise capital—$1.1 billion in Series C funding in May 2024, plus billions in debt—shows confidence from players like Coatue and Blackstone, but it also underscores a need for constant cash infusions to fuel growth.
The AI Trade: Boom or Bubble?
CoreWeave’s IPO arrives amid a broader debate about AI’s trajectory. The market’s enthusiasm for generative AI has driven valuations skyward, but whispers of an “AI bubble” are growing louder. We have seen recent concerns about Microsoft’s AI spending, while shifts toward non-NVIDIA XPUs could disrupt CoreWeave’s GPU-centric model. If AI demand softens or GPU values crater, CoreWeave’s debt-laden balance sheet could buckle under the weight of expectations. The company’s 96% reliance on committed contracts offers some stability, but it’s a thin buffer against a market downturn.
On the flip side, CoreWeave’s purpose-built platform is a genuine differentiator. Its 20% improvement in model flops utilization (MFU) over benchmarks and MLPerf records 29 times faster than competitors in 2023 showcase technical prowess. With NVIDIA’s GB200 NVL72 instances and a footprint dwarfing smaller rivals like Crusoe or Lambda, CoreWeave has the scale and partners—NVIDIA, Microsoft, IBM—to lead the AI infrastructure race. International expansion, like a £750 million UK investment, and acquisitions like Weights & Biases further bolster its ambitions.
CoreWeave is a darling in the AI infrastructure space onfluential and respected semiconductor analyst firm, SemiAnalysis giving CoreWeave the top ranking amongst the entire GPU-aaS provider space on the eve of the IPO and while I respect and admire SemiAnalysis’ depth of expertise and analysis the timing of the report drop did raise my eyebrows.
Looking Ahead
CoreWeave’s IPO is a litmus test for the AI economy. Its growth is undeniable, and its niche, GPU cloud computing for AI, taps into a secular trend with trillion-dollar potential (IDC estimates AI will add $20 trillion to global GDP by 2030). Yet, the downsized IPO signals investor caution, and for good reason. Dependency on NVIDIA and Microsoft, a financial model teetering on leverage, and an unproven path to profitability create a risk profile that’s as bold as its vision.
For optimists, CoreWeave is a front-runner in a market poised for exponential growth, with the right tools and clients to dominate. For skeptics, it’s a “time bomb” (as one analyst dubbed it), vulnerable to supply shocks, client loss, or an AI hype correction. The truth likely lies in between. If GPU values hold and AI demand accelerates, CoreWeave could justify its $32 billion aspirations. But if either falters, it may face a cycle of raising more debt or equity, or collapsing under its own weight.
Investors peering through the microscope will see a company at a crossroads: a high-stakes gamble on AI’s future, with equal parts brilliance and brinkmanship. As CoreWeave steps onto the public stage, the market will decide whether it’s a visionary trailblazer or a cautionary tale. For now, the lens reveals a tantalizing, if treacherous, opportunity.
Steven Dickens | CEO HyperFRAME Research
Regarded as a luminary at the intersection of technology and business transformation, Steven Dickens is the CEO and Principal Analyst at HyperFRAME Research.
Ranked consistently among the Top 10 Analysts by AR Insights and a contributor to Forbes, Steven's expert perspectives are sought after by tier one media outlets such as The Wall Street Journal and CNBC, and he is a regular on TV networks including the Schwab Network and Bloomberg.