The numbers are staggering, and for a moment, they seem to defy gravity. This week, a trio of companies at the absolute heart of the artificial intelligence revolution—CoreWeave, Nebius, and Supermicro—released quarterly earnings that sent their stocks soaring and painted a clear, unflinching picture of the current moment. We are in the thick of an AI infrastructure arms race, and the sheer scale of investment and demand suggests this build-out is just getting started. The story these reports tell isn’t just about profit margins or earnings beats; it’s about a fundamental reordering of the global technology landscape, where computing power has become the most sought-after commodity on earth.
Nebius, a relative newcomer in the competitive neocloud space, delivered a figure that stops you cold: revenue soared to $582.3 million, a 454% increase from the same period last year. Behind that number are four major AI cloud deals, each averaging over a billion dollars in total contract value. This isn’t just growth; it’s a land grab. The company is racing to construct data centers packed with the high-powered graphics processing units (GPUs) that fuel large language models and complex AI workloads, then renting that raw computational muscle to clients. The cost of playing in this league is astronomical. Nebius reported capital expenditures of $5.7 billion for the quarter, driven overwhelmingly by purchasing those precious GPUs and the supporting hardware. They admitted they could sell out their planned capacity through 2027 but are deliberately holding some back, a move that speaks to both strategic foresight and the intense, immediate pressure from customers clamoring for more.
CoreWeave, another specialized cloud provider built from the ground up for AI, mirrored this narrative of explosive demand tempered by massive spending. Their revenue jumped to $2.5 billion, more than doubling year-over-year. Even more telling is their backlog—the value of signed contracts for future work—which ballooned to $104.2 billion, a 246% increase. They’ve already inked an additional $25 billion in deals queued up for next quarter. The hunger for their service is undeniable. Yet, like its peers, CoreWeave is in a relentless investment cycle, pouring $9.4 billion into capital expenditures last quarter to acquire more chips and build more infrastructure. This spending is reflected in their expanding adjusted net loss, which widened to -$567 million. It’s a classic high-stakes tech playbook: spend aggressively to capture the market, betting that the long-term dominance will justify the near-term financial pain.
Supermicro, the hardware maestro building the specialized servers that house these AI chips, provided the crucial link in the supply chain. They posted adjusted earnings that beat Wall Street expectations, though revenue came in just a whisper below forecasts at $11.1 billion. The real story was their guidance for the next quarter, projecting net sales between $14.5 billion and $15.5 billion—a figure that dramatically exceeded analyst estimates of $11.9 billion. This outlook confirms that the demand for AI-optimized hardware isn’t a passing wave; it’s a sustained tsunami. Companies aren’t just testing the waters; they are building out their AI capabilities at an unprecedented pace, and they need Supermicro’s building blocks to do it.
Taken together, these reports offer a multi-faceted view of the AI gold rush. The cloud providers (Nebius, CoreWeave) are the modern-day equivalents of those who sold shovels and Levi’s jeans, providing the essential, scalable compute that every AI pioneer needs. The hardware makers (Supermicro) are the foundries and toolmakers. Their collective message is unambiguous: the constraint in AI today is not ideas but silicon and the systems that harness it. Businesses across sectors are voting with their wallets, committing billions to secure their share of computational power, fearing that being left behind is a greater risk than overspending.
- Nebius revenue: $582.3 million
- CoreWeave revenue: $2.5 billion
- Supermicro revenue: $11.1 billion
- Nebius capital expenditures: $5.7 billion
- CoreWeave capital expenditures: $9.4 billion
- CoreWeave backlog: $104.2 billion
| Company | Revenue | Capital Expenditures | Backlog |
|---|---|---|---|
| Nebius | $582.3 million | $5.7 billion | N/A |
| CoreWeave | $2.5 billion | $9.4 billion | $104.2 billion |
| Supermicro | $11.1 billion | N/A | N/A |
Of course, this breakneck pace invites skepticism. The eye-watering expenditures and mounting losses at some firms naturally lead to questions about sustainability and bubble-like conditions. The AI trade has been volatile, swinging on every piece of news about chip supply, software breakthroughs, or economic concerns. Investors are perpetually wondering if we’ve reached peak hype or if the runway ahead is still long. For now, the sheer momentum of these financial results suggests the latter. The next major checkpoint arrives soon, when Nvidia, the company whose chips are the very engine of this transformation, reports its earnings. Its performance will be the ultimate bellwether, revealing whether the demand at the foundation of this entire ecosystem—for the GPUs themselves—remains as insatiable as the infrastructure builders imply. Based on what we just saw from those building the world around Nvidia’s technology, all signs point to a resounding yes.
