Imagine a company that started in the shadows of the crypto mining world and suddenly finds itself at the heart of the biggest tech revolution since the internet. That’s CoreWeave in a nutshell. Their latest quarterly results just landed, and let me tell you, the numbers are nothing short of jaw-dropping.
I’ve been following the AI infrastructure space for years, and rarely do you see growth this explosive. It’s like watching a startup rocket from garage tinkering to Wall Street darling overnight. But is it sustainable? Let’s unpack what really happened in Q3 and why investors are buzzing.
The Headline Numbers That Turned Heads
Picture this: revenue more than doubling year-over-year. Not a modest bump, but a full-on surge. CoreWeave clocked in at $1.36 billion for the third quarter, easily topping the $1.29 billion that analysts had penciled in. That’s the kind of beat that sends shares twitching—even if they dipped a bit after hours.
Coming off $583.9 million in the same period last year, this 134% jump isn’t just growth; it’s acceleration. In my experience covering tech earnings, triples are rare, but consistent triples quarter after quarter? That’s the signature of a company riding a secular wave. And right now, artificial intelligence is that wave.
Losses narrowed too, from around $360 million to $110 million. Sure, they’re still in the red—building out gigawatts of capacity doesn’t come cheap—but the trajectory is clear. Cash burn is slowing as revenue ramps. Perhaps the most intriguing part? Their backlog ballooned to $55.6 billion. That’s not pocket change; that’s a multi-year runway locked in.
Breaking Down the Revenue Drivers
So where’s all this money coming from? Simple: Nvidia GPUs on steroids. CoreWeave doesn’t build chips or write models; they rent the plumbing that makes AI possible. Think of them as the premium landlord for the world’s most powerful computing real estate.
During the quarter, they inked some monster deals. One with a certain ChatGPT maker worth up to $6.5 billion in expansion. Another six-year pact with the social media giant formerly known as Facebook, potentially hitting $14.2 billion. And then there’s the mysterious “leading hyperscaler” handing over their sixth contract. My guess? One of the big three clouds needing overflow capacity.
The AI buildout isn’t slowing—it’s compounding. Every model gets bigger, every training run hungrier for compute.
Power tells the story too. Contracted capacity jumped from 2.2 gigawatts to 2.9 in just three months. For context, that’s like adding the output of two nuclear reactors dedicated solely to AI. I’ve found that power contracts are the ultimate leading indicator in this space—once secured, revenue follows like clockwork.
From IPO to 164% Gains: The Stock Story
Remember March? CoreWeave debuted at $40 a share in a market that was anything but welcoming. Volatility was crushing even the hottest names. They raised $1.5 billion—respectable, but a far cry from the billions some unicorns were chasing pre-2022.
Fast forward eight months, and those shares closed Monday at $105.61. Do the math: that’s a 164% return while the Nasdaq “only” gained 32%. In a year when many tech IPOs are still underwater, this performance stands out like a beacon.
But let’s be real—extended trading saw some profit-taking. Normal after a beat, especially with guidance coming. The question on everyone’s mind: can they keep signing these mega-deals, or was Q3 a peak?
- March IPO: $40/share, $1.5B raised
- Current price: $105.61 (pre-earnings close)
- Total return: 164% vs. Nasdaq’s 32%
- Market cap implication: Well north of initial valuation
The OpenAI and Meta Expansions: Game Changers?
Let’s zoom in on those headline deals, because they deserve more than a passing mention. The OpenAI expansion isn’t just another contract—it’s validation at the highest level. When the company behind the models everyone’s using chooses you for billions in spend, that’s a stamp of approval no marketing budget can buy.
Same story with Meta. Six years, up to $14.2 billion. That’s not a trial run; that’s a strategic partnership. I’ve always believed that in infrastructure, stickiness is everything. Once you’re woven into a customer’s roadmap, switching costs become prohibitive.
What’s fascinating is the diversification. Yes, they’re tied to Nvidia’s ecosystem—everyone in this space is—but their customer mix is spreading. Hyperscalers, AI natives, even enterprises dipping toes into custom models. This isn’t a one-trick pony betting on a single tenant.
The Core Scientific Deal That Wasn’t
Ah, the one that got away. Less than four months post-IPO, CoreWeave announced intent to acquire Core Scientific for $9 billion. Seemed like a masterstroke—vertical integration, instant scale, bitcoin mining assets repurposed for AI.
Then shareholders voted no. Ouch. In hindsight, maybe the timing was off. Markets hate complexity, and combining a fresh public company with a volatile crypto-adjacent target? Recipe for indigestion.
But don’t cry for CoreWeave. Organic growth is proving more than sufficient. Sometimes the best deals are the ones you don’t do. This rejection might have saved them from distraction at a critical scaling moment.
Power, Backlog, and the Real Moat
Let’s talk about what separates winners from also-rans in AI infra. Anyone can buy GPUs. The hard part? Securing power and data center space at scale. CoreWeave’s 2.9 gigawatts under contract isn’t just a number—it’s a defensive moat.
Every gigawatt translates to roughly 100,000 high-end GPUs. Do the multiplication: they’re positioned for hundreds of thousands of the latest chips. And with lead times stretching 12-18 months for new builds, this backlog is gold.
| Metric | Q2 2025 | Q3 2025 | Growth |
| Contracted Power | 2.2 GW | 2.9 GW | +32% |
| Revenue Backlog | ~$40B est. | $55.6B | +39% |
| Revenue | ~$1B est. | $1.36B | +36% QoQ |
See the pattern? Power leads backlog leads revenue. It’s a flywheel. And unlike some peers burning cash on speculative builds, CoreWeave’s capacity is largely spoken for. Lower risk, higher visibility.
Competition: The Elephant in the Room
No analysis is complete without addressing the giants. Google, Microsoft, Amazon—they’re all building AI clouds. Why rent from CoreWeave when you can DIY?
Fair question. But here’s the nuance: even hyperscalers hit capacity walls. Training the next GPT-5 or Gemini Ultra requires clusters that dwarf anything in production today. When your own fabs are maxed, you turn to specialists.
Plus, CoreWeave offers something the clouds don’t always prioritize: bare-metal access. No virtualization overhead, full control. For bleeding-edge AI shops, that performance edge matters. It’s like choosing a race car over a luxury sedan—different use cases.
Valuation: Expensive or Justified?
At 164% post-IPO, eyebrows raise. Is this another bubble stock? Let’s think in multiples. With $55.6 billion backlog and growing, they’re trading on future earnings, not trailing.
Rough math: assume 20% of backlog converts annually (conservative), that’s $11 billion run-rate by 2027. At current enterprise value, you’re paying maybe 8-10x forward sales. Pricey? Yes. Insane? Not in AI land.
Compare to Nvidia’s 40x sales or Super Micro’s wild ride. CoreWeave’s multiple feels almost reasonable. The key variable: execution. Any delay in data center commissioning, and margins compress fast.
What Guidance Might Tell Us
As executives hop on the call, all eyes are on forward commentary. Will they raise full-year numbers? Hint at Q4 linearity? The backlog gives confidence, but investors crave specifics.
In my view, the most watched metric won’t be revenue—it’s power deployment cadence. Every quarter of acceleration compounds their lead. Miss a site rollout, and suddenly that backlog looks less imminent.
Also watch capex guidance. They’re in full buildout mode. Any signal of efficiency gains (cost per GPU dropping) would be hugely bullish. Infrastructure margins are lumpy; smoothing them is the holy grail.
The Bigger Picture: AI’s Infrastructure Crunch
Step back, and CoreWeave is a proxy for a much larger story. Global AI spend is projected to hit trillions by 2030. But chips are useless without electrons and square footage. The bottleneck isn’t silicon anymore—it’s everything around it.
Utilities are scrambling. Data center developers can’t build fast enough. In this environment, companies with secured capacity are kings. CoreWeave didn’t just luck into position; they repurposed crypto mines when bitcoin crashed. Talk about turning lemons into AI lemonade.
Perhaps the most interesting aspect? They’re proving you don’t need to be a household name to dominate a critical layer of the stack. In AI, the picks and shovels often outperform the gold miners. History rhymes.
Risks That Keep Me Up at Night
No investment is risk-free, especially in hyper-growth tech. Customer concentration looms large—two big deals drove much of the quarter. What if training cycles peak? What if on-prem AI takes off?
- Concentration risk: Top customers = disproportionate revenue
- Capex intensity: Billions needed before profitability
- Competition: Hyperscalers could insource more
- Macro: Recession = delayed AI budgets
- Execution: Data center delays = backlog slippage
That said, the demand picture looks robust. Every earnings call from Big Tech mentions AI capex increases. This isn’t hype; it’s infrastructure catching up to ambition.
Where Do We Go From Here?
CoreWeave stands at an inflection. IPO behind them, mega-customers locked in, balance sheet fortified. The next 12 months will test whether they graduate from specialist to platform.
Watch for: international expansion (Europe/Asia power deals), software layer announcements (beyond raw GPU rental), and—dare I say—profitability timelines. If they start guiding to breakeven by 2027, the stock rerates hard.
In a market obsessed with efficiency, CoreWeave reminds us that sometimes the biggest moats are physical. Atoms still matter, even in a bits world. And right now, they’re stacking both at a pace that’s hard to ignore.
Infrastructure is the new oil. CoreWeave isn’t drilling—they own the pipelines.
– Anonymous AI fund manager
Whether you’re an investor, a tech enthusiast, or just curious about where the AI revolution is heading, this is a name to watch. The Q3 print wasn’t just good—it was a statement. The buildout accelerates, and CoreWeave is strapped in for the ride.
One final thought: in March, skeptics questioned the $40 valuation. Today, at $105+, some call it expensive. But if the backlog executes even halfway, we’re looking at a company that could define AI infrastructure for a decade. That’s not a trade—that’s a transformation.
Stay tuned. The conference call starts soon, and guidance could move markets. In AI, momentum compounds faster than anywhere else. CoreWeave just proved they have it in spades.
