Shares of CoreWeave fell as much as 10% in extended trading on Thursday, as investors focused on margin pressure and aggressive capital spending plans, even after the artificial intelligence-focused cloud infrastructure provider posted stronger-than-expected fourth-quarter revenue.
The company reported revenue of $1.57 billion, ahead of the $1.55 billion forecast by analysts polled by LSEG. Revenue grew 110% year over year, underscoring the intensity of demand for AI compute capacity.
CoreWeave posted a loss per share of 89 cents for the quarter. Adjusted earnings before interest, tax, depreciation, and amortization came in at $898 million, below the $929 million consensus from StreetAccount, suggesting profitability metrics lagged revenue expansion.
Register for Tekedia Mini-MBA edition 19 (Feb 9 – May 2, 2026).
Register for Tekedia AI in Business Masterclass.
Join Tekedia Capital Syndicate and co-invest in great global startups.
Register for Tekedia AI Lab.
The mixed performance came against a backdrop of heightened investor sensitivity toward AI-linked stocks, following recent developments at Anthropic that triggered sharp selling across segments of the sector.
Nvidia supply constraints and massive capital push
CoreWeave’s business model remains tightly intertwined with Nvidia, whose graphics processing units power most advanced AI workloads. CEO Mike Intrator told analysts on a conference call that Nvidia’s chips remain in short supply.
Average prices for Nvidia’s H100 processors in the fourth quarter were within 10% of where they began the year, while older A100 prices increased in 2025, Intrator said. The pricing stability for H100 chips, despite supply constraints, indicates sustained demand from hyperscalers and AI model developers.
CoreWeave is leaning heavily into expansion. The company plans to target at least $30 billion more than its 2025 capital expenditures for 2026. For 2025, total capex stood at $10.31 billion. That scale of projected investment signals an aggressive build-out of data center infrastructure, even as financing conditions remain tight and debt levels elevated.
As of Dec. 31, CoreWeave carried $21.37 billion in debt, a substantial balance for a company that went public just last March. The leverage underscores the capital-intensive nature of AI infrastructure and may partly explain the market’s cautious reaction despite revenue growth.
Power capacity — a critical constraint in AI infrastructure — continues to expand. CoreWeave ended the year with 850 megawatts of active power capacity, above the 827 megawatts analysts had projected. Contracted power reached 3.1 gigawatts, and the company aims to add more than five gigawatts beyond its current contracted footprint by 2030.
The scale of that ambition reflects a long-term bet that demand for AI compute will continue to accelerate, driven by large language models, enterprise AI deployment, and inference workloads.
Backlog surge, competitive positioning, and sector divergence
CoreWeave’s revenue backlog swelled to $66.8 billion from $55.6 billion at the end of the third quarter, providing visibility into future cash flows. The backlog growth suggests customers are locking in long-term compute commitments amid persistent chip scarcity.
The company supplies AI model developers such as Google and OpenAI, placing it at the center of the generative AI ecosystem. During the quarter, CoreWeave also announced a deal with model builder Poolside and launched an object storage service, broadening its product suite beyond pure GPU compute.
The storage offering positions CoreWeave to compete more directly with established hyperscalers, including Amazon Web Services, though the company remains primarily a specialist in high-performance AI cloud infrastructure.
It also increased a credit facility to $2.5 billion from $1.5 billion, bolstering liquidity as it ramps up capital expenditures.
“In 2025, CoreWeave became the fastest cloud platform in history to surpass $5 billion in annual revenue,” Intrator wrote in a blog post, highlighting the speed of the company’s scale-up.
Despite Thursday’s after-hours decline, the stock had risen 36% so far in 2026 as of the regular session close. That performance contrasts sharply with the nearly 22% drop in the iShares Expanded Tech-Software Sector Exchange-Traded Fund over the same period, signaling that investors have differentiated between AI infrastructure providers and broader software names.
Still, the earnings reaction suggests that growth alone is no longer sufficient. Investors are weighing capital intensity, chip supply constraints, debt levels, and margin trajectories more closely as the AI trade matures.
Executives are scheduled to discuss results and provide guidance on a conference call beginning at 5 p.m. ET.