This AI company surged 250% and left doubters baffled

Ever wondered how elite investment banks structure a board deck? 
I have pulled together 400 real M&A presentations you can learn from — you can explore them here.

Hi there,

This week, we break down CoreWeave, a once-obscure AI infrastructure player that has quietly become central to the generative AI supply chain.

You wouldn’t guess it from the $76 billion market cap, but CoreWeave started as a crypto mining outfit. Fast-forward a few years, and they’re now a go-to name for GPU-based compute — essentially renting out high-end chips to AI developers who can’t (or won’t) rely on cloud giants for capacity.

The IPO back in March 2025 priced at $40. Since then? It’s run up more than 250%, with $1.5 billion in fresh equity raised.

Let’s dive in.

A Message from StartEngine

StartEngine’s $30M Surge — Own a Piece Before June 26

StartEngine is the investing platform providing exposure to pre-IPO companies like OpenAI, Perplexity, and Databricks.

After doubling their revenues YoY in 2024 ($23M to $48M), StartEngine’s now tripled first quarter revenue YoY to a record $30M, based on its unaudited Q1 2025 financials. Now you can join 45K+ shareholders across all offerings before this round closes next month.

Invest in StartEngine

_{Reg A+ via StartEngine Crowdfunding, Inc. No BD/intermediary involved. Investment is speculative, illiquid & high risk. See OC and Risks on page.}

📚 From Ethereum to Enterprise Contracts

CoreWeave was founded in 2017 — not exactly the classic Silicon Valley story. The team came from commodity trading, not software engineering, and spotted an opportunity in the crypto crash: buy up discarded GPU rigs on the cheap.

While everyone else was liquidating, they retooled the hardware for more sustainable workloads such as visual effects rendering, machine learning training, and eventually, gen-AI compute.

But hardware was never the endgame.

The company morphed into something tighter: a specialized infra platform offering on-demand access to GPUs, built for customers who need speed, scale, and predictability — and don’t want to wait in line behind the cloud hyperscalers.

That pivot, from generic mining capacity to tailored compute infrastructure, laid the foundation. Today, the business runs on long-term contracts, just-in-time delivery, and elastic scaling that actually flexes with customer needs.

👾 The Nvidia Axis

CoreWeave's competitive positioning centers on preferential access to Nvidia's hardware rather than simply owning GPUs. The company has secured priority allocation for advanced GPUs, providing meaningful time-to-market advantages while competitors rely on standard procurement processes.

This extends beyond conventional vendor relationships. Nvidia's $250 million IPO investment validates CoreWeave as a preferred distribution channel, enabling faster hardware deployment than traditional hyperscalers with their longer procurement cycles.

The business operates on a self-reinforcing cycle:

• Early hardware access pulls in AI-native workloads

• Workloads convert into long-term contracts

• Contracts support additional debt financing

• Credit funds more datacenter construction

• And the cycle reloads

💰 How the Financial Model Works

CoreWeave’s 2024 financials are hard to ignore:

• $1.9 billion in revenue (+737% YoY)

• ~74% gross margins

• Adjusted EBITDA north of 60%

But let’s be clear: this isn’t a software company. The economics lean closer to project finance than cloud SaaS.

CoreWeave doesn’t build speculatively. It secures multi-year AI infrastructure contracts upfront, then uses those commitments as collateral to raise debt for new datacenter buildouts.

By year-end 2024, Remaining Performance Obligations (RPO) hit $15.1 billion — holding strong at $14.7 billion in Q1 2025, with average contract terms of four years.

That contractual base becomes the engine:

• Long-term RPO underwrites new credit lines

• Credit funds accelerated Nvidia capacity purchases

• Faster access wins more high-intensity workloads

• Workloads convert to more contracts

The loop works, but it’s capital-fueled and Nvidia-dependent. One key win: OpenAI signed an $11.2 billion multi-year deal in early 2025, and took ~8.75 million shares in the process. That added to backlog and sent a strong signal to the rest of the market.

⚖️ Key Risk Factors to Watch

Despite CoreWeave’s impressive growth and success, several structural vulnerabilities could impact its business model:

Customer Concentration: Microsoft is 62% of revenue. If Azure shifts strategy, and for instance builds internal capacity or secures its own Nvidia pipeline, it would be a major hole in the revenue base.

Supplier Reliance: CoreWeave’s positioning depends on Nvidia’s goodwill. Any change in allocation priority, delivery cycles, or chip mix could squeeze margins and delay deployments.

Competitive Pressure: There’s no deep IP moat here. The real advantage is execution speed. If hyperscalers solve for procurement delays or alternative chipmakers start to catch up, CoreWeave’s differentiation narrows fast.

Balance Sheet Stress: The debt load is heavy, and built for growth, not flexibility. If workloads soften or utilization dips, covenants kick in. In that case, refinancing could get messy, especially in a higher-rate environment.

📅 Key Catalysts to Monitor

A few key developments could swing things in either direction over the next 12–18 months:

Customer Behavior

• Microsoft’s internal GPU strategy: If Azure ramps internal capacity or secures priority access directly from Nvidia, CoreWeave’s anchor tenant could shift spend away. Watch for CapEx disclosures, Azure workload commentary, or GPU inventory in future earnings reports.

• New logos or contract expansions: Management hinted at material new bookings for H2 2025 — especially from GenAI firms seeking dedicated capacity. Confirmation would reinforce the demand-side of the model.

Supplier Dynamics

• Nvidia allocation announcements: Upcoming hardware cycles will test whether CoreWeave retains preferential access or if hyperscalers secure earlier allocations.

• Export controls or yield constraints: Any regulatory or supply-side bottlenecks affecting Nvidia shipments could directly disrupt CoreWeave’s deployment cycle.

Financing Conditions

• Refinancing window in 2026: With $8.8B+ in debt, spreads matter. If credit markets tighten (e.g. >300bps vs 2024), the model compresses. Monitor refinancing timelines and disclosures on interest expense sensitivity.

• RPO-to-net-debt ratio: This is the metric to watch. If it slips below 1.0×, leverage constraints could force CoreWeave into expensive equity issuance or datacenter pullbacks.

✒ Final Thoughts

CoreWeave’s meteoric rise underscores both the promise and the fragility of today’s AI infrastructure landscape.

The company’s ability to secure early access to Nvidia’s most advanced GPUs and lock in long-term contracts has fueled extraordinary growth, but these same factors create acute dependencies.

Looking ahead, CoreWeave’s future will hinge on its ability to maintain privileged supplier relationships and diversify its customer base beyond Microsoft. Any shift in Nvidia’s allocation strategy or a change in Microsoft’s internal GPU buildout could fundamentally alter the company’s trajectory.

At the same time, the broader market is watching to see if CoreWeave’s capital-intensive, contract-backed model becomes the blueprint for next-generation compute infrastructure — or if it proves vulnerable to the same forces that have upended previous hardware cycles.

Still, it’s a fascinating real-time test case: Is this the playbook for next-gen AI infra? Or just a high-leverage bet that worked in one cycle?

Either way, the coming quarters will reveal whether its model can withstand tightening credit, evolving supply chains, and intensifying competition, or if its rapid ascent will be matched by equally swift challenges.

If you’re enjoying these mid-week editions, let me know what you’d like to see next — investor profiles, firm histories, or deep-dives on strategic shifts. I’m always open to ideas.

Until next time,
PE Bro

P.S. Was this forwarded to you? Subscribe here.

Disclaimer: This overview is based on publicly available sources. While accuracy is a priority, not all details are independently verified. This is not an official statement from CoreWeave or its affiliates.

Sources:

[1] Reuters
[2] Investors.com (I)
[3] Investors.com (II)
[4] Wikipedia
[5] Esofund
[6] CoreWeave (I)
[7] CoreWeave (II)
[8] Financial Times