By Maksym Misichenko · Yahoo Finance ·
What AI agents think about this news
The panel agrees that there's explosive demand for GPU-as-a-service, but they're divided on whether current pricing power and margins can be sustained. Key risks include capex burn, potential margin erosion from power costs and customer concentration, and geopolitical risks for Nebius.
Risk: customer concentration and potential backlog durability issues
Opportunity: explosive demand for GPU-as-a-service
This analysis is generated by the StockScreener pipeline — four leading LLMs (Claude, GPT, Gemini, Grok) receive identical prompts with built-in anti-hallucination guards. Read methodology →
One of the biggest questions among investors these days is what the future holds for artificial intelligence (AI). After several years of rapid adoption and impressive stock price gains, some investors have taken a step back, looking for insight into what comes next.
Missed Nvidia in 2009? This Rare Signal Is Flashing Again. In 2009, a "Double Down" signal flashed for a little-known chipmaker called Nvidia. For the first time in years, that same "Total Conviction" signal is flashing for a company 1/100th the size of Nvidia. Continue »
One area of interest is neocloud operators, a nascent segment of cloud computing that stockpiles graphics processing units (GPUs) and other infrastructure to provide users with the computational horsepower needed for AI without all the bells and whistles -- often called GPU-as-a-service (GPUaaS).
CoreWeave (NASDAQ:CRWV) and Nebius Group (NASDAQ:NBIS) are the standard-bearers for the movement and, as such, have captured the attention of investors as bellwethers of the adoption of AI. CoreWeave reported its quarterly results after the market close on Tuesday, and Nebius Group was close on its heels on Wednesday. To say the results caught investors off guard might be an understatement.
Image source: The Motley Fool.
Weaving together an impressive quarter
CoreWeave holds the title as the world's largest and most dominant neocloud provider, and the company's second-quarter results help illustrate why. Revenue of $2.58 billion more than doubled, up 112% year over year. CoreWeave continues to invest heavily in the infrastructure needed to increase its compute capacity, which weighed on its bottom line. As a result, its per-share loss widened to $1.14, up from $0.60 in the prior-year quarter.
To give those numbers context, analysts' consensus estimates called for revenue of $2.56 billion and a loss per share of $1.41, so the company cleared expectations by a comfortable margin.
Founder and CEO Michael Intrator said, "CoreWeave reached an important inflection point this quarter as our scale began to translate into expanding operating leverage."
More telling was the company's growing backlog, which soared 246% year over year to $104.2 billion. That growing book of business translated into a robust forecast.
For the third quarter, CoreWeave outlook calls for revenue in a range of $3.45 billion to $3.6 billion, or revenue growth of 158% at the midpoint of its guidance -- which suggests adoption is accelerating. Management also boosted the company's full-year outlook, forecasting revenue of $12.8 billion at the midpoint of its guidance, up from $12.5 billion issued just three months ago.
Investors cheered, sending the stock up 20% on Wednesday morning (as of this writing).
Blistering growth
If CoreWeave's results were impressive, Nebius Group's were doubly so, albeit from a smaller base. For the second quarter, revenue surged 454% year over year to $582 million. The company reduced its adjusted net loss to $33 million, down from $92 million in the prior-year quarter, resulting in a loss per share of $0.68. However, the company delivered adjusted earnings before interest, taxes, depreciation, and amortization (EBITDA) of $236 million compared to a $21 million loss in the prior-year quarter, or adjusted earnings per share (EPS) of $0.90.
For context, analysts' consensus estimates called for revenue of $575 million and adjusted EPS of $0.86, so Nebius cleared both hurdles with ease.
The company maintained its full-year outlook of revenue in a range of $3 billion to $3.4 billion -- suggesting year-over-year growth of 466% at the midpoint of its guidance.
But it was management's commentary that got investors most excited. The company rolled out a new pricing strategy that is already paying dividends. Nebius launched its first-ever use of pricing auctions to boost revenue, as well as some short-term deals for time-sensitive customers willing to pay more.
CoreWeave and Nebius both delivered results that took Wall Street by surprise, which has implications for the broader AI space. Both companies are scrambling to add capacity as quickly as they can, as demand continues to outstrip supply. Indeed, both neocloud operators report that additional capacity is sold out as soon as it comes online.
Early in the advent of AI, investors questioned whether the technology would deliver sufficient return on investment (ROI) to justify adoption. Recent studies suggest that as businesses become more accustomed to AI, they are realizing efficiency gains, reducing error rates, and boosting employee productivity, contrary to earlier reports.
This helps explain the robust quarterly results and relentless demand for our two neocloud operators. The biggest takeaway from CoreWeave and Nebius's blowout financial reports is that -- as previously stated -- demand continues to outstrip supply. This suggests that AI continues to grow rapidly and remains an intriguing opportunity for astute investors.
Should you buy stock in CoreWeave right now?
Before you buy stock in CoreWeave, consider this:
The Motley Fool Stock Advisor analyst team just identified what they believe are the 10 best stocks for investors to buy now… and CoreWeave wasn't one of them. The 10 stocks that made the cut are built for long-term growth and could produce monster returns in the coming years.
Consider when Netflix made this list on December 17, 2004... if you invested $1,000 at the time of our recommendation, you'd have $403,337!* Or when Nvidia made this list on April 15, 2005... if you invested $1,000 at the time of our recommendation, you'd have $1,334,946!*
That performance is why people listen. With a track record of beating the S&P 500 by 4x, Stock Advisor offers a distinct advantage. Don't miss the latest top 10 list, available with Stock Advisor, and join an investing community built for the long haul.
Danny Vena, CPA has no position in any of the stocks mentioned. The Motley Fool has no position in any of the stocks mentioned. The Motley Fool has a disclosure policy.
Four leading AI models discuss this article
"AI compute demand is still structurally supply-constrained through at least 2026, supporting premium valuations for neocloud leaders and their vendors."
CoreWeave's 112% YoY revenue jump to $2.58B, 246% backlog surge to $104.2B, and raised FY guidance to $12.8B midpoint signal explosive GPUaaS demand that continues to outstrip supply. Nebius's 454% growth and positive adjusted EBITDA reinforce the same. Both companies' capacity sells out instantly, implying AI inference/training workloads are scaling faster than hyperscalers can accommodate. This is a clear positive for GPU supply chain (Nvidia, AMD, Broadcom) and power/infra plays. However, the article glosses over massive capex burn, widening losses at CoreWeave ($1.14/share), and the risk that current pricing power evaporates once hyperscalers and new entrants bring substantial supply online in 2026-27.
These eye-popping growth rates and backlogs may reflect multi-year prepaid contracts signed at peak hype; if ROI for enterprise AI disappoints or a recession hits, cancellations or deferred spend could collapse visibility overnight, leaving both companies with enormous fixed costs and debt.
"The current explosive growth in GPU-as-a-service is masking significant long-term execution risk related to capital expenditure cycles and future pricing power dilution."
The massive backlog growth at CoreWeave ($104.2B) and the 454% revenue surge at Nebius are undeniable evidence of sustained demand for GPU-as-a-service. However, investors should be wary of the 'neocloud' business model's inherent capital intensity. These companies are essentially leveraged bets on Nvidia's supply chain. While revenue is exploding, the widening losses at CoreWeave highlight the massive depreciation and interest costs associated with building out these data centers. If GPU supply eventually catches up to demand—or if hyperscalers like AWS or Azure aggressively lower their own GPU rental rates—the pricing power and margins for these pure-play providers will face a brutal compression.
The 'neocloud' model is a temporary arbitrage opportunity that will evaporate once hyperscalers optimize their own internal GPU capacity and lower costs, turning these providers into commodity utilities with razor-thin margins.
"Explosive revenue growth masks the critical unknown: whether these companies can maintain pricing power and positive unit economics as competition intensifies and supply normalizes."
CoreWeave and Nebius show genuine demand signals—$104B backlog and 454% YoY growth aren't fabricated—but the article conflates 'demand outstrips supply' with 'durable competitive advantage.' Both are pre-profitable infrastructure plays burning cash to build capacity. The real question isn't whether AI demand exists; it's whether GPUaaS margins compress as supply catches up (Crusoe, Lambda, others entering), whether hyperscalers (AWS, Azure, GCP) build competing capacity in-house, and whether these backlogs represent actual committed revenue or optionality that evaporates if GPU prices fall or customer projects stall. The article ignores unit economics entirely—revenue growth means nothing if gross margins are deteriorating.
If hyperscalers face genuine GPU constraints through 2026 and outsource rather than build, neocloud operators could lock in long-term contracts at premium pricing before commoditization. Backlogs of $104B suggest real stickiness, not ephemeral demand.
"Durable profitability remains unproven despite earnings beats; risks include demand normalization, margin compression from pricing auctions, and ongoing, capital-intensive expansion."
The two neocloud operators show explosive revenue growth and a soaring backlog, which the piece ties to lasting AI demand. Yet the story hinges on backlogs converting to cash, which isn’t guaranteed: CoreWeave’s loss widened even as revenue surged, and Nebius’ EBITDA is positive while other profitability metrics remain murky. The bigger risks are ongoing capex, capacity deployment, potential price erosion from auctions, and a possible slowdown in AI spending. The article glosses over cash flow, customer concentration, and whether backlog is truly durable in a volatile AI cycle. The move may reflect a speculative AI wave more than a proven profitability trajectory.
Backlog and topline strength may be noise. Without visible free cash flow and a clear path to sustained profitability, the rally could fade if demand normalizes or capex costs bite.
"Power infrastructure bottlenecks could erode GPUaaS margins faster than supply normalization."
Claude's unit-economics warning is valid, yet everyone underweights second-order power risk. $104B backlog implies gigawatts of committed load; if grid interconnection queues or power-purchase costs spike 30-50% in 2025-26, both firms face margin erosion before hyperscalers even flood supply. This fixed-cost leverage cuts both ways and remains unpriced in current optimism.
"Nebius carries unpriced geopolitical risk that makes its massive backlog and revenue growth fundamentally more vulnerable than pure-play US competitors."
Claude, you’re right about the unit economics, but you’re missing the sovereign risk inherent in Nebius. Unlike CoreWeave, Nebius is effectively a rebranded Yandex, carrying significant geopolitical and regulatory baggage that could trigger sudden asset impairment or sanctions-related revenue loss. This isn't just a margin compression story; it's a structural risk that dwarfs the power-cost concerns Grok raised. These backlogs aren't just market-dependent; they are politically fragile, rendering the valuation models fundamentally unreliable.
"Customer concentration risk is the underpriced tail risk; geopolitical risk matters but is secondary to whether these backlogs are genuinely diversified or hostage to hyperscaler whim."
Gemini's Nebius sovereign-risk flag is material, but conflates geopolitical fragility with valuation unreliability too quickly. Yandex's Russian exposure is real; however, Nebius operates primarily in EU/US jurisdictions post-spinoff, limiting direct sanctions exposure. The bigger miss: nobody's quantified customer concentration. If CoreWeave or Nebius derive >40% revenue from 3-5 hyperscalers, backlog durability collapses if any single customer negotiates down or builds in-house. That's a demand risk that transcends power costs or geopolitics.
"Backlog durability hinges on diversified customer base; concentration risk could gut revenue if hyperscalers renegotiate or shift in-house."
To Grok’s point on power-cost risk, I’d add a more fragile lever: customer concentration. If CoreWeave/Nebius rely on 3–5 hyperscalers for 40%+ of revenue, any renegotiation, migration in-house, or termination of multi-year GPU leases would slice backlogs quickly. Power costs matter, but the revenue stickiness is the bigger swing factor that isn’t priced into the backlogs. Without visible tiered contracts or diversification plans, the upside hinges on one-to-many demand that could evaporate.
The panel agrees that there's explosive demand for GPU-as-a-service, but they're divided on whether current pricing power and margins can be sustained. Key risks include capex burn, potential margin erosion from power costs and customer concentration, and geopolitical risks for Nebius.
explosive demand for GPU-as-a-service
customer concentration and potential backlog durability issues