CoreWeave’s Q1 was a bit of a mixed bag, with the company beating revenue estimates by 5.5% yet guiding Q2 revenue roughly (6.5%) below consensus. Despite the soft guide, management maintained its full year guide at $12 to $13 billion, suggesting a strong ramp into the back half of the year, with revenue forecast to accelerate more than 80 points from Q2’s guided 108% to 190% by Q4.
Margins contracted across the board in Q1, with adjusted operating margin coming down 16 points YoY to just 1%. Similar to revenue, management remained optimistic on the margin front, projecting 2H operating income growth of >11X versus 1H, and a return to low double-digit adjusted operating margins by Q4.
Looking through 2026 and 2027, CoreWeave is targeting a rather aggressive ramp in active power, noting that it expects to have a “substantial majority” of its 3.5GW contracted power pipeline become active by the end of next year. For context, CoreWeave currently has >1GW active, adding roughly 150MW in the quarter. Bringing new capacity online is imperative for CoreWeave, as despite its backlog rising 49% QoQ to almost $100 billion, current backlog conversion looks to be largely priced in.
The challenge remains capex and the strain it is placing on CoreWeave’s balance sheet. Part of the reason that CoreWeave saw weak price action after the report is they raised fiscal year capex; “For the full year, we now expect CapEx of $31 billion to $35 billion. The increase on the low end from our previous guidance is related to increases in component pricing.”
To simplify this, CoreWeave needs a euphoric market environment for the stock to do well (and likely one where interest rates are going down), as for every $1 made about $2 is spent in capex. The concern is when does the bleeding stop, or will CoreWeave remain on a never-ending build cycle given the rapid iterations around GPUs and system components.
The second concern is the supply environment is becoming trickier with the increasing complexity of AI systems. CoreWeave’s management described this best when it was stated: “Like the truth of the matter is the limiting factor isn't just power, it's labor, it's memory, it's storage, it's our ability to bring up infrastructure.” Each of those introduces an element of risk, more so when you are upside down on capex.
CoreWeave is No Longer Just a GPU Provider
As we pointed out a year ago in our analysis CoreWeave: AI Infrastructure Built for the Next Decade; Upside Down Business Model:
“CoreWeave brands itself as the world’s first “AI hyperscaler” as they offer both infrastructure and a software platform for developing large language models and deploying them. Being dubbed an AI infrastructure player means CoreWeave must offer a compelling value proposition to attract business from arguably the largest competitors in the world – AWS, Microsoft Azure and Google Cloud. In its S-1 fling, the company points out it was built for AI workloads as opposed to the legacy cloud infrastructure-as-a-service providers that were primarily optimized for the cloud software era and e-commerce era. CoreWeave also asserts that outdated cloud infrastructure leads to lower utilization rates when you factor in usage. its S-1 fling, the company points out it was built for AI workloads as opposed to the legacy cloud infrastructure-as-a-service providers that were primarily optimized for the cloud software era and e-commerce era. CoreWeave also asserts that outdated cloud infrastructure leads to lower utilization rates when you factor in usage.
The company also offers proprietary software to help achieve higher total system performance and more favorable uptime relative to competitors.”
This quarter, CoreWeave highlighted more progress in its full-stack software approach, offering customers a higher degree of flexibility in consumption. CoreWeave introduced Flex Reservation and spot pricing this quarter, which aim to help customers budget for and manage unpredictable bursts of demand, with both of these new offerings “immediately oversubscribed.” CoreWeave also recently launched CoreWeave Interconnect in collaboration with Google Cloud, following its SUNK Anywhere and LOTA Cross-Cloud offerings.
These three services aim to minimize friction in managing multi-cloud environments, helping organizations run AI workloads anywhere in the cloud. CoreWeave says the three “are already proving to be highly effective at capturing increased wallet share.”
Outside of GPUs, CoreWeave is also seeing strong momentum in CPUs, networking and storage, with both CPUs and networking similarly expected to exceed $100 million ARR and storage noted to be multiplying quickly. Overall, these still remain just a tiny fraction of CoreWeave’s ARR.
Hyperscalers Becoming CoreWeave’s Main Customers
CoreWeave’s main value proposition lies within its ability to offer high GPU utilization rates, early access to next-gen systems such as Vera Rubin, and circumventing the hypervisor layer with bare metal servers. However, infrastructure is likely to weigh on margins (and currently is), which is why CoreWeave is aiming to pivot towards a SaaS-augmented model with Omni.
CoreWeave says Omni enables them to “deploy and operate our full cloud stack in customers' own data centers with their GPUs,” with early interest said to be strong across cloud, enterprise and sovereign customers. While this opens the door for CoreWeave to build higher-margin, recurring revenue streams at third-party data centers on third-party-owned GPUs, layering in on top of its core GPU rental model, the main challenge with this approach is that its main customers are hyperscalers, who do not need this offering.
For example, Meta and OpenAI are CoreWeave’s two largest customers, contributing ~65% of revenue in Q1, with contractual obligations worth roughly $57.6 billion in total or well over half of its backlog. CoreWeave also has a unique relationship with Google, selling capacity to Google Cloud which will then resell that to OpenAI.
Although CoreWeave believes its customer base is beginning to diversify, such as with financial services approaching $10 billion, and ten customers committing to >$1 billion in spending, the overall nature of its business remains highly concentrated in the hyperscalers. Thus, the company’s target customer, one that needs both compute and high-margin software add-ons, is not appearing in its current customer base. Realistically, this presents a major challenge of how CoreWeave will be able to monetize a SaaS offering considering two-thirds of its business lies within customers who likely have no need for it.
Targeting Aggressive Ramp in Power; Carries Execution Risk
CoreWeave is targeting an aggressive ramp in active power over the next 18 months, as it aims to convert a “substantial majority” of its 3.5GW contracted power pipeline over to being active and revenue-generating. This is part of CoreWeave’s broader goal of reaching >8GW of active power by 2030, or an 8X increase over the next four years. The main question that this begs is how CoreWeave can sustain the level of capex needed for such an aggressive ramp.
As of Q1, CoreWeave surpassed 1GW of active power, adding more than 150MW during the quarter, while simultaneously increasing its contracted power to 3.5GW, up more than 400M during the quarter.
Breaking this down, and assuming that management’s commentary applies only to the 3.5GW figure and not subsequent intra-quarter increases in the contracted power pipeline, we can roughly estimate CoreWeave’s trajectory and potential capex costs associated with this buildout.
Under that framework, and assuming “substantial majority” equates to ~80% of the 2.5GW of contracted but not yet active power, this would estimate CoreWeave bringing on 2GW of new active power by year-end 2027, taking total active power to 3GW. This would be roughly 3X its current capacity.
Management stated in Q1 that they “remain firmly on track to reach or exceed our target of more than 1.7GW by the end of 2026,” so the above assumption would represent a sharper acceleration of capacity additions in 2027, requiring around 1.3GW of new additions if the 1.7GW active power target is met. However, some analysts are expecting a much more rapid ramp in power, with Oppenheimer penciling in potential new additions of 1GW by Q3 2026.
CoreWeave also shed a bit more light over its power strategy: “We plan to continue to expand our contracted power footprint through leases while also accelerating our development of self-build sites, which will provide us with greater operational control and long-term financial upside. We expect our first self-build site to come online later this year.”
CoreWeave’s Leasing Strategy Reduces Capex Shock, But Adds Counterparty Risk
There are a couple of puts and takes to this approach, notably that the lease-based approach can offer a faster time to power with lower capex, while also opening the door up to execution risk from its counterparts. For example, it was reported in December that Core Scientific’s facility in Texas faced a two-month delay due to weather, with this spilling over to CoreWeave as a result (per management on Q4’s timing miss: “the delays in powered-shell delivery associated with the data center provider will have an impact on our fourth quarter results.”)
However, the benefit of this lease-heavy approach is that it offers CoreWeave a higher degree of financial flexibility when it comes to managing its balance sheet, which is in rather rough shape with less than $2.3 billion in cash to nearly $25 billion in debt.
Connecting CoreWeave’s active power target to its projected capex for 2026 illustrates why it is opting to go the lease route versus prioritizing primarily greenfield development.
This is evident when viewing this via the lens of its deal with Core Scientific. Under that deal, CoreWeave is paying $1.5 million per MW in capex during development; this would project CoreWeave’s capex obligations to be roughly $525 million for the remaining ~350MW of capacity Core Scientific has yet to deliver.
Compare this to a greenfield 350MW build with construction costs of ~$11 million per MW. Such a build would cost CoreWeave $3.85 billion, meaning the lease approach saves it roughly 86% on capex. This also allows CoreWeave to funnel a majority of its capex dollars into high-end GPUs – at current estimates for GB300 GPUs to cost $33 million per GW per Morgan Stanley, CoreWeave could outfit a 350MW facility with GB300 racks for $11.55 billion, versus around $15.4 billion from the ground-up, or savings of ~25%. For CoreWeave, every dollar of savings matters in this buildout.
In total, bringing on 2GW of power by the end of next year, such as what we outlined above, could cost ~$66 billion via primarily leases (where capex goes almost directly to GPUs), or $88 billion including construction for GB300 capacity. With Rubin estimated to cost $41 million per MW, per MS, costs could reach $82 to $104 billion for 2GW.
Interest Payments Surging to $2B Annualized
The reason that every dollar matters for CoreWeave is tied to its debt, as the company is forking out nearly $2 billion annualized as of Q1 on interest payments, with this figure only set to rise further in Q2. This is one of the core problems with utilizing debt to fund its buildout – interest payments are rising faster than revenue, and reductions in the weighted cost of debt, stated as an 80 bp reduction year-to-date, are not yet enough to slow this train down.
For Q1, CoreWeave’s net interest payments reached $536 million, or more than $2 billion annualized, rising more than 38% QoQ. For Q2, CoreWeave guided for interest payments to be $650 to $730 million, or up nearly 29% QoQ, and moving up to almost $2.8 billion annualized with no real signs of slowing. On a YoY basis, interest payments will be up more than 2.5X next quarter.
When looking at CoreWeave’s projected capex needs for 2026 – not even including 2027, which is likely to be much higher considering the active power goals management has laid out that – alongside its thin balance sheet with debt 11X more than its $2.27 billion in cash, it’s clear that CoreWeave is not close to breaking this cycle.
With nearly 90% of its debt carrying >6% to 15% interest rates and not maturing until 2029 at the earliest, this cycle has years to continue.
Backlog Reaches a Record $99.4 Billion, Fully Priced In?
CoreWeave reported record revenue backlog additions in Q1 as it began signing initial Vera Rubin deals, while contracting out capacity expected to come online in 2027. Notably, CoreWeave is largely sold out of 2026 GPU capacity, providing a high degree of visibility into this year’s revenue and ARR guidance, with potential for upside if capacity can come online faster.
Backlog reached $99.4 billion in Q1, up 284% YoY and nearly 49% QoQ, accelerating from 20% QoQ in Q4 and driven by substantial growth from both existing and new customers. Management added that the backlog is all tied to contracts either currently online, or expected to come online in 2026 or 2027.
For additional color on backlog, CoreWeave expects 36% of this backlog to be recognized as revenue over the next 24 months, representing roughly $35.8 billion.
Looking ahead to Q2, backlog should move substantially higher as CoreWeave closed a deal with Meta worth $21 billion through 2032, a $6 billion deal with Jane Street (who is also backing CoreWeave with a $1 billion equity investment), and a multi-year deal with Anthropic in April.
Turning briefly to ARR, CoreWeave inched its 2026 ARR target higher, now projecting to exit the year with annualized revenue of $18 to $19 billion, raised from $17 to $19 billion previously. Management opted to maintain its 2027 exiting ARR guide at $30 billion, adding that 75% of that ARR is already contracted and booked.
The challenge here is that CoreWeave’s backlog conversion and ARR targets roughly line up with consensus estimates, suggesting that growth is currently priced in. Taking Q1’s $2.1 billion in revenue with the entirety of the expected backlog conversion (although a portion will be recognized in 2028) equals out to $37.9 billion, whereas current consensus estimates point to 2026 and 2027 combined revenue of $37.6 billion.
For ARR, CoreWeave’s exit ARR target of $18.5 billion for 2026 suggests December 2026 revenue of $1.54 billion, which, based on ramp dynamics, would likely project Q4 revenue to be around $4.4 billion. This is below current estimates for $4.56 billion in revenue in Q4. The same applies for 2027’s exit ARR of $30 billion, which projects December 2027 revenue of $2.5 billion, or Q4 2027 revenue around the $7.3 billion range. This again is below current estimates for $7.57 billion.
The takeaway here is that CoreWeave has to build and expand capacity at a quicker rate, placing much more emphasis on capex and funding needs, given current revenue targets over the next seven quarters already look to be largely priced in. Management stated that they have “already secured sufficient power capacity to deliver on our 2027 [ARR] target and expect to continue to add new capacity and customer commitments,” yet that power still has to be brought online and become revenue-generating.
Financials
Revenue Misses Estimates yet Growth to Inflect into 2H
While CoreWeave reported a solid 5.5% beat to revenue estimates in Q1, its Q2 guidance fell short of the mark, with the neocloud guiding for revenue nearly (6.5%) below consensus at midpoint.
Q1 revenue increased 111.7% YoY to $2.08 billion, maintaining a similar YoY growth pace as the prior quarter. On a sequential basis, Q1 revenue increased 32.2% QoQ, a sharp acceleration from Q4’s 15.2% QoQ print. Strong pricing trends aided growth in the quarter, as CoreWeave noted that “average pricing for the A100s, H100s and H200s and L40s all increased QoQ,” while near-term capacity remains largely sold out.
For Q2, CoreWeave guided for revenue to be $2.45 billion to $2.6 billion, roughly (6.5%) below consensus estimates for $2.7 billion at midpoint. This would project a slight deceleration in YoY growth to 108.2%, while QoQ growth would also moderate back to 21.5% QoQ.
Management was straightforward in saying that they do expect revenue to inflect as they cross from Q2 into Q3, setting the stage for a stronger 2H. Given the full-year guide of $12 to $13 billion, CoreWeave is looking at around $7.9 billion in 2H revenue given the implied 1H revenue of $4.6 billion. Current estimates have 2H plotted out (with a bit of upside at $8 billion) at $3.47 billion in Q3 for 154.2% YoY and 37.2% QoQ growth, and $4.56 billion in Q4 for 189.9% YoY and 31.4% QoQ growth.
This acceleration into 2H is tied to a few factors, with the first simply being CoreWeave’s active power ramp translating to a larger installed base for it to generate revenue from. It is also supported by strong pricing dynamics with management noting that prices were rising across the board for Ampere, Hopper and Blackwell GPUs, as well as a potentially larger mix of higher-priced Blackwell Ultra instances entering the fray. Pricing strength is extending into 2027 as CoreWeave begins to contract out next year’s capacity, providing more legs for revenue growth to remain strong.
Margin Questions Remain, though Expectations of Improvement in 2H
Q1 showed a rather bleak picture for margins, with contraction appearing across the board from gross to net margins. However, management expects Q1 to be the trough for margins for the year, with the largest inflection expected to occur crossing from Q2 to Q3, similar to revenue. Management also reaffirmed that they remained on track for sequential margin expansion through the rest of the year.
Q1 gross margin was 66%, down seven points YoY and two points QoQ. Management chalked this up to timing of capacity and scale. For the timing point, management explained that during data center fit-out for leased capacity, it incurs lease, power and depreciation expenses for one to two months before revenue generation begins. For scale, management explained “So if you think of it as we're running 50 megawatts, and we add 300 megawatts in a quarter, the impact on gross margin is going to be enormous. On the other hand, when you're running 2,000 megawatts and you add 50 megawatts, it's not going to have as material an impact on your gross margin.”
Q1 GAAP operating margin was (7%), down four points YoY and one point QoQ. Adjusted operating margin was 1%, in line with management’s guide and contracting 16 points YoY and five points QoQ. For Q2, adjusted operating margin was guided to be roughly 2.4% at midpoint, a slight sequential expansion yet still down more than 13 points YoY – more on this and 2H dynamics below. The difference between the two margin figures boils down to SBC.
Q1 GAAP net margin was (36%), down four points YoY and seven points QoQ. Adjusted net margin was (28%), down 13 points YoY and 10 points QoQ.
Double-Clicking on Margin Dynamics
Q1’s call featured some discussion about CoreWeave’s margin trajectory given the Q1 contraction, signs of inflection in Q2 and management’s expectation to return to low double-digit adjusted operating margin by Q4. This would represent more than 10 points of expansion relative to Q1.
To put this in dollar form, CoreWeave delivered $21 million in adjusted operating income in Q1, guided for $60 million at midpoint in Q2, and maintained its FY26 guide for $1 billion at midpoint. This suggests 2H adjusted operating income of $919 million at midpoint, or >11X what CoreWeave delivered in 1H. Rightfully so, analysts questioned about this sharp inflection and why management has conviction in achieving this, with CFO Nitin Agarwal stating that it comes down to executing its plan of ramping active power, and seeing adjusted operating income begin to outpace revenue growth in 2H.
On the point of operating margins, CoreWeave downplayed the impact of rising component costs, such as memory. While component price inflation was a factor in the slight FY26 capex raise from $32.5 billion to $33 billion, management emphasized that deal pricing means any such costs are effectively passed on through to customers:
“We build our contracts to incorporate the cost of all of the components that are necessary to deliver infrastructure. And so by and large, we are insulated from the price inflation on some of the components because we include that in our pricing that we ultimately bring to clients in order to target the margins that Nitin spoke to, right, is up in the mid-20s is how we think about it on a unit basis. …
And so we've done a really good job of understanding what the components and electricity costs are going to be, we have it structured so that it is effectively passed through when we enter into the contract once again, to ensure that we're able to hit our targeted margins on a unit basis.”
EPS and Adjusted EBITDA
Considering the margin contraction this quarter, CoreWeave reported larger losses than expected this quarter. FY26 and FY27 revenue revisions over the past six months illustrate how much farther CoreWeave has moved from profitability, emphasizing that this path remains challenging.
Q1 GAAP EPS was ($1.40), coming in below the ($1.20) estimate. Q1 adjusted EPS was ($1.12), missing the ($0.91) estimate by more than 22%, and widening from ($0.61) a year ago.
Q2 is expected to see minimal improvement, with GAAP EPS projected to be ($1.24) and adjusted EPS projected to be ($1.03). Despite the expected progress on the margin front, CoreWeave is estimated to remain unprofitable this year, with Q4 adjusted EPS currently projected to be ($0.34).
To illustrate how much further CoreWeave has moved from profitability, its FY26 adjusted EPS estimate now sits at ($3.29), down from ($0.23) in November. For FY27, adjusted EPS estimates sit at ($0.63), down from $2.26 in November, with profitability not projected until Q4 FY27. This is likely driven by a handful of factors – gross margin compression from timing of bringing capacity online, possible impacts to gross margin from higher power prices, increased D&A hitting the opex line via technology and infrastructure expenses as CoreWeave expands its GPU fleet, component price inflation, and ballooning interest pyaments weighing on widening operating margins. It’s also important to note that specifically for component pricing, it’s likely that we are currently just seeing a baseline: “Having said that, in the last 6, 9 months, there has been an acute shortage of certain components that have moved up.”
Turning to adjusted EBITDA, CoreWeave reported $1.16 billion in Q1 for a 56% margin, down six points YoY and one point QoQ.
Cash Flows, Balance Sheet in Rough Shape due to Elevated Capex Needs
As expected, CoreWeave’s cash flows and balance sheet are in rather rough shape, though the company highlighted that it has no debt maturities until 2029 aside from self-amortizing contract-backed debt and OEM vendor financing. This gives some leeway for revenue to scale and cash flows to (hopefully) begin improving to help service debt, considering interest payments on said debt are now running at more than $2 billion annualized and expected to reach above $2.5 billion annualized next quarter.
Q1 operating cash flow was $2.98 billion for a 144% margin, driven primarily by depreciation and changes in accounts receivable. This improved from a 99% margin in Q4 and a 6% margin a year ago.
Q1 free cash flow was ($4.71 billion) for a (227%) margin, widening from a (159%) margin in Q4 and (137%) a year ago. Capex was $7.7 billion in the quarter, while Q2 capex was guided to be similar at $7 to $9 billion.
Cash totaled $2.27 billion while debt reached $24.86 billion. Debt will move higher in Q2 as CoreWeave closed a $3.1 billion loan in mid-May as well as $4.5 billion of senior notes in April. Management commented in Q1’s call that they have reduced the cost of debt by 80 bp year to date, though interest expenses are still rising. Overall, CoreWeave has raised >$20 billion year-to-date, including the “first ever investment-grade Delayed Draw Term Loan backed by HPC infrastructure, achieving an A- equivalent rating from Moody's, Fitch, and DBRS.”
Conclusion
There are two ways to view CoreWeave, and both deserve a discussion, as the company is at the intersection of the quality AI trade and AI hype. At one point, we had CoreWeave on a Top 15 list, yet as more circular investments unfolded and capex ballooned relative to revenue, our process, which is to seek strong fundamentals, caused us to drop the stock from the list. That circularity is intensifying, as even its non-hyperscaler customer Jane Street is providing a $1 billion loan on a $6 billion sale.
Key supplier Nvidia also bought another $2 billion in CoreWeave shares in Q1, which creates a triangle where Nvidia’s is the supplier, investor and customer. Right now, the economics are such that 2026 capex is $30-$35B versus $12.5B in revenue at the midpoint, or about $2.60 in capex for every $1 of new revenue. Another risk to consider is CoreWeave’s ARR targets, as 2026 and 2027 targets both suggest potential Q4 revenue this year and next a bit below current consensus estimates, placing the emphasis on accelerating its buildout, and thus capex, to fuel higher growth.
Overall, macro may matter more to CoreWeave than any of the stock-specific information above. As a heavily debt-funded, long-duration cash-burn story, it will do better in a lower-rate environment. For this stock, if we do enter, it will be 90% technicals.
Please note: The I/O Fund conducts research and draws conclusions for the company’s portfolio. We then share that information with our readers and offer real-time trade notifications. This is not a guarantee of a stock’s performance and it is not financial advice. Please consult your personal financial advisor before buying any stock in the companies mentioned in this analysis. Beth Kindig and the I/O Fund do not own shares in CRWV at the time of writing and may own stocks pictured in the charts.
Damien Robbins, Equity Analyst at I/O Fund contributed to this analysis.
Recommended Reading: