Article Directory
CoreWeave's AI Dream: Built on Billions, Undercut by... a Missing Building?
Alright, let's talk CoreWeave. You know, the AI infrastructure darling that’s supposed to be riding this generative AI wave all the way to the bank. They just dropped their Q3 numbers, and if you squint real hard, you might think it’s all sunshine and rainbows. Revenue more than doubled, hitting a cool $1.36 billion, beating estimates. Net loss narrowed, too, from a whopping $360 million last year to a "mere" $110 million. Big whoop, right? Still a loss. Still burning cash like it’s going out of style, which, offcourse, it isn't in this market.
But here’s the kicker, the part where the corporate spin machine starts to sputter: the stock tanked. Six percent in extended trading, some folks even saw it hit 16.3% down. Why? Because these geniuses, fresh off their monster revenue beat, decided to give us a peek at their 2025 outlook, and it was… underwhelming. They’re now predicting $5.05 billion to $5.15 billion, which somehow trails the average analyst estimate of $5.29 billion. So, you beat this quarter, only to tell everyone you're probably gonna miss next year's expectations? That's not how you keep the hype train rolling, folks. That's how you hit a pothole at 100 mph.
The Emperor's New Data Center
So what’s the excuse? CEO Mike Intrator, looking all prim and proper at the Semafor World Economy Summit – probably still basking in the glow of the last funding round – blamed a "third-party data center developer" being behind schedule. Oh, just one data center, he chirped, out of 41 in their portfolio. Yeah, and that one tiny little hiccup is enough to shave hundreds of millions off your projected revenue? Give me a break. This isn't just a hiccup. No, scratch that—it's a flashing red light on the dashboard, and Intrator's trying to cover it with a Post-it note.
They’ve got this massive backlog, right? $55.6 billion in contracted power, up to 2.9 gigawatts. Sounds impressive. Like a kid showing off his mountain of Halloween candy. But here’s the thing about candy mountains, or massive backlogs for that matter: they don't mean jack if you can't actually eat them. Or, in CoreWeave's case, if you can't get the damn powered-shell data centers to put your shiny Nvidia GPUs in. Intrator admits they're "supply-constrained," not on power, but on the buildings. You're telling me a company worth billions, with deals like a $6.5 billion expansion with OpenAI and a $14.2 billion, six-year Meta deal—not to mention a sixth contract with some "leading hyperscaler"—can't secure a few buildings? What kind of operation are they running here? Are we supposed to believe they're just hoping someone builds them a place to put their gear? It’s like buying a Ferrari but forgetting to buy a garage.
Analyst Spin & My Own Damn Doubts
And then there's Morgan Stanley's Keith Weiss. Bless his heart. Top Morgan Stanley Analyst Hikes CoreWeave Stock (CRWV) Target after Q3 Earnings. A "Hold" rating, mind you, which in analyst-speak usually means "don't buy, but we don't want to piss off the company." He sees "strong demand" and "reducing customer concentration risk." Oh, really, Keith? Because last I checked, relying on OpenAI and Meta for billions still sounds like putting a lot of eggs in a few very large baskets. He admits "consistent execution remains a challenge." Ya think? The CEO himself just admitted they can't even get buildings built on time.
Intrator's big promise? "The overwhelming majority of the delay that you're seeing should be taken care of within Q1 of next year." Heard that one before. It's the classic "just around the corner" line. And let's not forget the little tidbit buried in the corrections: the CEO misstated the number of data centers. Can't even get the basic facts straight about their own operations? That just screams confidence, doesn't it? It’s like a chef telling you his restaurant has 41 locations, then later correcting himself to say it's actually 20, but don't worry, the food's still great!
They went public at $40 a share, now it's $105.61, a 164% return. The Nasdaq only gained 32% in that time. Yeah, it's a hell of a run. But when you’re talking about massive capital expenditures for 2026 – "well in excess of double" 2025's $12-$14 billion – and you're still reporting net losses because you can't get your infrastructure built, it makes you wonder. How long can this AI gold rush keep covering up these cracks? Are we witnessing the inevitable slowdown of a market that flew too close to the sun, or just a few clumsy stumbles on the way to actual profitability? Or, then again, maybe I'm the crazy one here for thinking companies should actually, you know, make money and deliver on promises before we crown them kings.
The AI Bubble's First Pimple
Look, the demand for AI infrastructure is real. No one's denying that. But CoreWeave's story right now feels like a classic case of overpromising and under-executing. They're trying to build Rome in a day, and they're hitting all the bottlenecks you'd expect. A massive backlog is great for headlines, but if you can't convert it to actual, profitable revenue because you're waiting on a contractor to finish a building, then what are we even talking about? It's a house of cards, built on a foundation of "should be taken care of next quarter." The stock dip? That's not a buying opportunity for "long-term investors." That’s the sound of reality knocking. Hard.