What happened with CRWV today?
No read yet
Uses news and trading volume to explain the day’s move and how it relates to the market.
Stop guessing. Get the call, a three-month target price, the best and worst cases, and a source for every claim.
CoreWeave supplies AI labs, technology companies and other enterprises with computing capacity to train AI models and run them to generate responses or predictions. Its CoreWeave Cloud Platform combines GPU-equipped data centers with storage, networking and software. CoreWeave Mission Control helps allocate that infrastructure, manage workloads and monitor performance.
Almost all revenue comes from multi-year contracts under which customers reserve capacity and must pay whether they use it or not. These contracts typically include upfront payments; customers can also buy services on demand and pay for usage. Microsoft accounts for most revenue, leaving CoreWeave heavily dependent on that relationship.
Delivering the contracted capacity requires access to GPUs, powered data-center space and financing. CoreWeave primarily finances infrastructure with debt backed by customer contracts. All GPUs in its infrastructure are NVIDIA GPUs because current customer contracts require them. U.S. export controls and sanctions also govern its platform and can restrict equipment sourcing, international sales and customer deployments.
The risk section is the longest, dullest part of the filing. This keeps only what can really hurt you: where it hits, how hard, when it plays out.
Every company claims an edge. Each one is checked against hard numbers like margins and returns: which hold, and which line gives way first.
CoreWeave’s results also hinge on substantial compute commitments from privately held OpenAI, an important customer missing from this list.
Microsoft bought cloud capacity accounting for approximately 67% of CoreWeave’s 2025 revenue, yet its Azure cloud also competes for customers. Watch Microsoft’s AI capacity purchases: they matter directly to CoreWeave’s revenue, not just to wider cloud demand.
CoreWeave depends on NVIDIA’s chips to deliver the AI cloud capacity customers buy: at the filing date, every deployed GPU used NVIDIA technology, alongside key NVIDIA networking equipment. New GPU availability is the concrete development to watch, because CoreWeave depends on access to current-generation hardware.
Nebius Group competes with CoreWeave for customers paying to train and run AI models, with both selling GPU computing, storage and software as one cloud platform. Nebius’s AI cloud business is the relevant comparison, rather than its other businesses and investments.
CoreWeave relies on Core Scientific for approximately 590 MW of contracted powered data-center space, while Core Scientific relies on CoreWeave for all its Colocation segment revenue. Watch delivery of that space: Core Scientific’s ability to meet CoreWeave’s requirements matters to their relationship and CoreWeave’s capacity expansion.
Meta Platforms is a future revenue anchor for CoreWeave, initially committing to pay up to approximately $14.2 billion for cloud services through December 2031. Meta earns substantially all its revenue from advertisers on its apps; its fulfillment of the cloud commitment is what directly matters to CoreWeave.
CoreWeave’s cloud expansion depends in part on Applied Digital delivering the data-center capacity it leases at Polaris Forge 1, covering 400 MW of critical IT load. Watch the campus’s delivery schedule, which runs from the second half of 2025 through the first half of 2027.
CoreWeave’s capacity buildout is tied to Galaxy Digital’s delivery of leased data-center space at Helios, totaling 526 MW of contracted critical IT load across three phases. Watch the later phases’ delivery, expected to start in 2027, rather than Galaxy’s digital-asset trading results.
IREN competes with CoreWeave for customers buying cloud capacity to train and run AI models, supplying data centers, computing and software together. Watch IREN’s AI cloud capacity coming online: it owns and operates all three layers, unlike platforms more reliant on outside facilities.
Oracle competes with CoreWeave for customers’ AI computing workloads and cloud capacity commitments. Its cloud infrastructure offering is the business to watch, since broader cloud providers can use lower prices or bundled services to pressure CoreWeave’s orders and margins.
Amazon’s AWS competes with CoreWeave for customers paying for cloud infrastructure to run AI workloads. AWS is the relevant business to follow, not Amazon’s stores: broader cloud offerings and bundling can put pressure on CoreWeave’s pricing.
Alphabet’s Google Cloud competes with CoreWeave for customers buying AI cloud infrastructure. Follow Google Cloud rather than Google’s advertising revenue when assessing the competitive pressure on CoreWeave’s customer wins.
Cerebras Systems gives AI model developers an alternative to CoreWeave’s GPU cloud, selling cloud computing and reserved supercomputer capacity for training and running models. Its dedicated-capacity bookings are the relevant offering to watch against CoreWeave’s multi-year capacity commitments.
ChronoScale Holdings Corporation competes with CoreWeave for AI computing customers, with GPU cloud services generating almost all its revenue. Watch the fixed per-GPU-per-hour fees paid by Together AI for dedicated H100 clusters, a concrete comparison for what customers pay for reserved compute.
DigitalOcean Holdings competes with CoreWeave through GPU infrastructure and AI platform services sold to growing technology companies. Its AI offerings—not its entire hosting business—are the useful comparison for demand from customers building AI products.
TeraWulf and CoreWeave both depend on sustained demand for dense AI computing, but TeraWulf leases the facilities to cloud and enterprise customers while CoreWeave sells the computing services. TeraWulf’s long-term hosting leases offer a view of demand for the physical capacity behind AI clouds.
Cipher Digital’s development of 600 MW of high-performance computing facilities for hyperscaler tenants reflects the powered-data-center demand that also supports CoreWeave’s cloud expansion. Watch progress on those facilities, rather than its bitcoin-mining revenue, for the relevant AI infrastructure signal.
Vertiv benefits from the same high-density GPU buildout CoreWeave needs to expand its cloud services, selling power and liquid-cooling equipment to data-center customers. Its data-center power and cooling sales are the relevant signal for infrastructure investment, not evidence of a direct CoreWeave supply contract.
Microsoft Corp earningsMSFT
Microsoft’s report offers a read on cloud and AI capacity demand from a company that is both a major CoreWeave customer and a cloud competitor.
Look at Microsoft’s Azure cloud performance and AI infrastructure investment, not as direct disclosure of its orders from CoreWeave but as signals about the market they share. If cloud demand and capacity investment remain strong, the market for AI cloud services is still expanding; if either slows, customer demand may be weakening while Azure could press harder for the same workloads. The takeaway is a signal about CoreWeave’s customer and competitive environment, not proof of a specific CoreWeave order.
Fed Interest Rate Decision
The Fed’s decision on its benchmark interest rate affects what CoreWeave pays to borrow for data-center and computing infrastructure investment, especially when it raises or refinances debt.
Expected4.0%Last release4.0%
If it comes in above expectations
Higher rates can raise the cost of new or refinanced borrowing and increase interest expense, leaving CoreWeave less cash for infrastructure investment and other needs.
If it comes in below expectations
Lower rates can make new or refinanced borrowing cheaper and ease interest costs, leaving CoreWeave more cash to fund infrastructure investment and other needs.
Core Scientific, Inc. earningsCORZ
Core Scientific’s report can show whether it is delivering the data-center space contracted by CoreWeave, which affects how quickly CoreWeave can expand cloud capacity.
Watch delivery progress for the powered data-center space Core Scientific has contracted to provide to CoreWeave, rather than relying on Core Scientific’s colocation revenue alone. On-schedule delivery would give CoreWeave sites where it can deploy more computing capacity; delays could slow deployment even if customers remain committed to buying that capacity. The conclusion is whether this part of CoreWeave’s physical expansion is getting the space it depends on.
CoreWeave, Inc. earningsThis company
CoreWeave’s report will show whether revenue from committed cloud capacity is keeping pace with the costs of expanding and financing its data centers.
Read revenue from committed contracts alongside cost of revenue and technology and infrastructure expense, which includes depreciation on infrastructure equipment. If revenue grows while those costs take a steadier or smaller share, expansion is converting into better operating economics; if costs grow faster and losses widen, more commitments have not yet translated into better economics. Then check whether interest expense adds to the net loss: that shows whether growth is keeping up with the cost of financing the buildout.
Nebius Group N.V. earningsNBIS
Nebius Group’s report can show whether a close rival offering GPU computing, storage and software is adding AI capacity and pursuing customers.
Look for growth in Nebius Group’s AI cloud business alongside new capacity coming online. If both advance, Nebius has more ability to pursue the same customers for model training and inference, increasing competitive pressure; if deployment is delayed or demand is weak, less competing capacity may be available soon, but that does not prove CoreWeave’s own demand is stronger. The takeaway is whether this direct rival is adding AI cloud capacity customers can choose instead.
Nvidia Corp earningsNVDA
NVIDIA’s report offers clues about GPU demand and supply; CoreWeave relies on those chips to deploy the cloud capacity its customers buy.
Read NVIDIA’s data-center demand and supply disclosures against CoreWeave’s need to deploy GPU capacity. Strong demand alongside expanding supply would show a busy market but could also give CoreWeave more room to build; constrained supply or delivery delays could hold back deployments and customer contracts. The takeaway is whether the GPU supply environment supports CoreWeave’s expansion, not how many chips CoreWeave itself received.
ChronoScale Holdings Corporation earningsCHRN
ChronoScale Holdings Corporation’s report may reveal the fixed hourly fees Together AI pays for dedicated H100 capacity, a narrow pricing signal for CoreWeave’s reserved compute.
If the report discloses the fixed per-GPU-per-hour fee Together AI pays for dedicated H100 clusters, treat it as the price in one specific arrangement, not a market-wide quote. A higher fee could indicate firmer demand and pricing for reserved capacity; a lower one could point to pressure on prices. CoreWeave’s contract terms may differ, so this is only a clue about dedicated GPU pricing—not proof of CoreWeave’s realized prices or margins.
