Fast growth, significant losses, disappointing guidance, "Nvidia's favored child" Coreweave plummets after its first public financial report

CoreWeave's Q1 revenue surged fourfold year-on-year, but the company warned that AI cloud business revenue expectations are "unstable," and the annual revenue growth rate is expected to slow to 190%.

On Wednesday, AI infrastructure provider CoreWeave released its first earnings report since going public, achieving $980 million in revenue for the first quarter ending in March, an increase of over 400% compared to the same period last year.

Financial data is as follows:

Revenue: First-quarter revenue was $981.6 million, a year-on-year increase of 420% compared to $188.7 million last year; adjusted revenue reached $163 million, a year-on-year increase of 550%.

Profit: Net loss of $314.6 million, compared to a loss of $129.2 million in the same period last year. The adjusted EBITDA margin for the first quarter was 62%. Loss per share for the first quarter was $1.49.

In terms of guidance, the company expects second-quarter revenue to be in the range of $1.06 billion to $1.1 billion, with adjusted revenue between $140 million and $170 million, and capital expenditures between $3 billion and $3.5 billion; the full-year revenue is expected to be in the range of $4.9 billion to $5.1 billion, with capital expenditures between $20 billion and $23 billion.

Full-year revenue growth rate guidance sharply reduced compared to Q1

CoreWeave's revenue growth in Q1 was strong, but investor sentiment towards the company's stock has turned pessimistic. This is because its executives indicated that this year's revenue and capital expenditures may be "unstable."

Even more concerning for the market is that the company expects the full-year revenue growth rate to be 190%, which is less than half of the Q1 growth rate.

CoreWeave CEO Mike Intrator acknowledged on the conference call that "the global trade policy environment continues to fluctuate," bringing uncertainty and causing costs for certain equipment to rise, but at the same time, customer demand is accelerating, and the impact of tariffs is expected to be "relatively small."

CoreWeave's business model involves purchasing NVIDIA AI chips and other data center equipment only after customers sign usage contracts, making the company a barometer for the AI-related spending boom.

After the earnings report was released, CoreWeave's stock initially rose 9% in after-hours trading but quickly fell, ultimately dropping 7%.

Despite the pullback, if the after-hours price holds into Thursday, the company's stock will still be 59% above its IPO price.

Demand remains strong! The company reveals it has signed an agreement with a "hyperscale customer"

The earnings report showed that CoreWeave's remaining performance obligations (contracts signed but not yet generating revenue) decreased from $15.1 billion at the end of 2024 to $14.7 billion.

During the subsequent earnings call, analysts expressed concerns about the trend of the company's reserve revenue declining since the end of last yearCoreWeave stated that company executives explained that the new agreement reached with OpenAI announced in March has not yet been included in the reported reserves, as the company has not yet determined how to account for this transaction in its financial statements. CoreWeave previously indicated that the deal is worth up to $11.9 billion.

During the conference call, Mike also revealed that CoreWeave has signed an agreement with a new hyperscale customer. He also stated that the company will report a $4 billion contract extension with "a large AI company" in documents to be submitted to the Securities and Exchange Commission, which will be reflected in revenue reserves starting from the second quarter.

According to previous media reports, in April this year, CoreWeave was in deep negotiations with Google to lease GPUs to the latter.

Below is a summary of the content from CoreWeave's Q1 earnings call:

Mike Intrator, CEO:

Thank you, Deborah, and welcome everyone. Thank you for joining us for our first conference call as a public company following our IPO in March. We look forward to ongoing productive conversations with all of you. We have had a fantastic start to 2025. The company has continued the momentum of the past few years.

My remarks will focus on key milestones and achievements. We delivered outstanding financial performance in the first quarter, highlighted by record revenue of $982 million and record adjusted operating income of $163 million, representing year-over-year growth of 420% and 550%, respectively. In the first quarter, we completed a strategic transaction with OpenAI, with a contract value of up to $11.9 billion. We also added new enterprise customers and a new hyperscale customer, and signed expansion agreements with several large clients, including a recent $4 billion expansion agreement with a large AI company.

These details will be included in our 10-Q filing. We announced and completed the acquisition of Weights and Biases, one of the industry-leading AI developer platforms with over 1,400 customers. We believe that the combined company will be able to provide additional value to our shared customer base. We continue to set new technical performance benchmarks, including the large-scale deployment of the GB200 Grace Blackwell system on our AI cloud platform, supporting leading AI developers Mistral, IBM, and Cohere. Overall, our financial performance, along with the continued strong trajectory of customer growth and technical achievements, demonstrates the exceptional execution of our team.

CoreWeave's platform enables the world's leading AI companies to train AI models and perform inference at scale. We are just getting started, and the demand for our platform is strong and accelerating. In particular, we are pleased to see a broad increase in demand for inference, as well as the accelerated adoption of AI by our enterprise customers. The growth and adoption of AI continue to be severely constrained by capacity limitationsCoreWeave is expanding as quickly as possible to meet customer demand. While the company is expanding at an unprecedented pace, we recognize the uncertainty caused by the ongoing fluctuations in the global trade policy environment. However, despite this uncertainty, customer demand is accelerating, as reflected in the guidance we are releasing today. Since this is our first earnings call, I would like to briefly outline our mission and describe our strategy for driving long-term shareholder value. CoreWeave is driving AI innovation at scale. Our cloud platform is built specifically for AI workloads, providing infrastructure as a service, and offering highly differentiated cloud software and services on top of it. The general cloud infrastructure of yesterday, which still powers much of the digital world, was not built to support the scale and complexity of artificial intelligence. These clouds were built for hosting websites and running SaaS applications, not for running high-performance training and inference workloads.

The exponential growth and success of our customers' AI products have driven demand for our differentiated services. With a deep understanding of AI customers and workloads, our cloud platform is optimized for AI needs at every layer. As inference becomes more compute-intensive, the ability to run research and production workloads on the same infrastructure provides our customers with the flexibility to optimize their total cost of ownership. Our cloud software and infrastructure services eliminate the complexity of running infrastructure at scale, offering significant differentiation in the market.

We believe CoreWeave does this better than anyone else, and industry experts agree. In March, SemiAnalysis awarded CoreWeave its highest rating of "Platinum" based on its ClusterMAX rating system. We are the only cloud provider to receive this rating. SemiAnalysis highlighted our industry leadership in operating large-scale 10K+ H100 clusters with high reliability.

We take pride in being rated higher than established hyperscale cloud service providers and emerging GPU clouds. Why is all this important? AI is the greatest technological revolution of our lifetime, and we believe the company can create tremendous shareholder value by enhancing its leadership position. Today, we benefit from a network of 33 data centers designed specifically for AI, spanning the United States and Europe, supported by 420 megawatts of active power.

Our total contracted power expands to approximately 1.6 gigawatts, providing us with years of enduring power capacity. CoreWeave operates at the forefront of technology, ensuring exceptional performance and efficiency for our customers every day. The world's most complex AI leaders understand the power of our platform. This enables CoreWeave to grow at an unprecedented pace.

Today, we serve the most important AI companies globally, and with the acquisition of Weights & Biases, we welcome nearly 1,400 more AI labs and enterprise customers. We see tremendous growth opportunities ahead. By 2030, AI is expected to have a cumulative impact of $20 trillion on the global economy. The total addressable market is expected to grow to $400 billion by 2028

To seize this opportunity and serve more customers, CoreWeave is at the forefront of developing innovative financial mechanisms to expand its platform. As of today, we have raised over $21 billion to expand our infrastructure and data center capabilities. The IPO is an important milestone in our financing strategy as it broadens our channels for acquiring low-cost capital, fueling our planned rapid expansion. Additionally, for a company of our scale and growth rate, CoreWeave has already achieved profitability (based on adjusted operating income).

Our business model has strong revenue visibility and attractive, sustainable unit economics. We continue to focus on four key areas: expanding capacity, financing for infrastructure, further differentiating our platform, and enhancing market expansion capabilities. I will elaborate on these areas separately.

First, expanding CoreWeave's capabilities. In the first quarter, we added approximately 300 megawatts of incremental contracted power to our portfolio. To efficiently expand capacity, we will strive to secure the necessary power resources to expand our data center footprint and deploy industry-leading next-generation computing capabilities to meet industry demand. Second, financing for infrastructure.

Our capital expenditures are success-based. Essentially, we only enter into capital expenditure projects when we sign multi-year contract revenues, which are sufficient to cover capital expenditure costs during the contract period. This allows us to responsibly expand the debt structure supporting contract revenues and utilize naturally deleveraging, self-amortizing debt instruments, keeping our leverage ratios relatively low.

Third, differentiating the platform. Rapid adoption and scaling of cutting-edge technologies are at the core of our competitive advantage. After being the first to deliver NVIDIA's H100 and H200 GPUs at scale, we became the first company to universally offer NVIDIA GB200 NVL72 instances and began to enhance Blackwell revenues in the first quarter. Our superior performance, better scalability, and shorter time-to-market provide customers with a competitive edge, helping them seize new AI opportunities. We also recently announced that our MLPerf inference v5.0 results set a new industry benchmark alongside NVIDIA's GB200 Grace Blackwell Superchips.

As we continue to build infrastructure and software differentiation, we launched the next-generation CoreWeave AI object storage, designed for the most demanding training and inference use cases. Pairing CoreWeave AI object storage with CoreWeave's Kubernetes services provides a ready-to-use experience for the most demanding AI customers, production-ready from day one. This quarter, we also announced support for NVIDIA's AI Enterprise Software and NVIDIA's Cloud Functions (NVCF), continuing to enable ecosystem solutions on the platform. Fourth, enhancing CoreWeave's market expansion capabilities

The demand for advanced AI infrastructure is experiencing tremendous growth. In addition to the enterprises and laboratories we serve, CoreWeave has recently expanded its global footprint. This enables the company to broaden its product offerings in new markets while developing existing customer businesses and attracting new clients. In fact, yesterday we announced a partnership with Merlin Edge to open a new data center in Spain, with Mistral AI as our anchor client. As the demand for AI infrastructure continues to grow rapidly, CoreWeave will succeed by executing and solving some of the world's most challenging AI infrastructure problems. Today, our team has grown to approximately 1,400 people, including some of the most accomplished and talented individuals in the AI field. I would also like to thank our investors and analysts for their attention to CoreWeave. I look forward to updating everyone on our progress as we continue to drive innovation and provide a cloud platform built specifically to power the AI revolution.

Now I will hand the meeting over to Nitin, who will discuss our financial performance for the first quarter and the outlook for the remainder of the year.

Nitin Agrawal, Chief Financial Officer:

Thank you, Mike, good afternoon everyone. It is exciting to communicate with investors and analysts for the first time as a public company, and I look forward to working with all of you to scale the company and create shareholder value. We achieved outstanding performance in the first quarter, highlighted by an exceptional 420% year-over-year revenue growth and robust operational profitability measured by adjusted operating income.

Our confidence in the business stems from the strong technological foundation we have built. As Mike emphasized, we are driving AI innovation at scale. Over the past five years, we have made significant investments in building the CoreWeave cloud platform and expanding our infrastructure, consistently demonstrating industry-leading performance. As this is our first earnings call as a public company, I will take a minute to reiterate the differentiated technology and robust business model we have built. We generate revenue by providing world-class AI infrastructure and proprietary management software and application services through the CoreWeave cloud platform.

Most of our revenue comes from long-term commitment contracts, providing us with strong revenue visibility and attractive unit economics. This visibility of growth allows us to take a success-based capital investment approach that aligns with customer contracts. We typically finance capital investments by leveraging self-amortizing structures that naturally deleverage over the contract term. We shared a supplemental slide in our earnings presentation that illustrates our contract mechanism.

Our first-quarter revenue was $982 million, a 420% year-over-year increase. The revenue to be recognized amount (including remaining performance obligations (RPO) and other amounts expected to be recognized as revenue in the future) reached $25.9 billion, a 63% year-over-year increase. In March, we signed a strategic cooperation agreement with OpenAI, with a contract value of up to $11.9 billion.

This new contract collaboration with OpenAI highlights CoreWeave's ability to provide reliable high-performance infrastructure services, supporting leading AI laboratories globally. We also added enterprise clients and a hyperscale cloud service provider, and signed a $4 billion expansion contract with a large AI companyThis $4 billion expansion contract was signed in the second quarter and will be reflected in our revenue to be recognized starting from the second quarter, not included in the first quarter report data.

It should be clarified that this will increase the revenue to be recognized amount of $25.9 billion at the end of the first quarter. While our revenue to be recognized amount is expected to grow rapidly over time, the growth rate will fluctuate at different periods due to our contract business model, the timing and scale of new contract signings, and the nature of revenue recognition.

The operating expenses for the first quarter were $1 billion, including one-time equity incentive expenses recognized after the IPO completion. As customer demand continues to grow significantly, we have increased our investments in data centers and server infrastructure, leading to higher revenue costs and technology infrastructure expenses in the first quarter.

In addition, we continue to invest in sales and marketing to expand and diversify our customer base, as well as general management activities in preparation for going public. The adjusted operating income for the first quarter was $163 million, a year-on-year increase of 550%. The adjusted operating profit margin for the first quarter was 17%, an increase of three percentage points year-on-year.

The net loss for the first quarter was $315 million, while the net loss for the first quarter of 2024 was $129 million. The increase in losses was due to one-time equity incentive expenses after the IPO, higher interest expenses, and increased taxes. The interest expense for the first quarter was $264 million, higher than expected, due to improved vendor payment terms, which reduced the time between vendor payments and asset utilization, thereby reducing the interest costs capitalized in the quarter.

The adjusted net loss for the first quarter was $150 million, while the adjusted net loss for the first quarter of 2024 was $24 million. The adjusted net loss was impacted by higher-than-expected interest expenses due to the previously described factors. The adjusted EBITDA for the first quarter was $606 million, nearly six times that of the first quarter of 2024, with an adjusted EBITDA profit margin of 62%, an increase of six percentage points year-on-year.

Capital expenditures mainly include investments in properties and technology equipment that drive platform expansion, enhance the value of software and technology stacks, and ultimately promote revenue growth. The capital expenditure report does not include construction in progress, as this represents infrastructure that has not yet been put into use. Additionally, the timing of data center capacity coming online and the deployment of the next generation of GPUs may lead to variations in capital expenditures between quarters. Total capital expenditures for the first quarter amounted to $1.9 billion.

We design our capital structure to support rapid expansion. We have a good track record of successfully securing funding for growth financing and reducing capital costs. As the business continues to expand, we expect to continue lowering capital costs. Our balance sheet and liquidity position are robust. As of March 31, we had $2.5 billion in cash, cash equivalents, and restricted cash. Aside from financing payments to OEM manufacturers and self-amortizing debt through contract payments, we have no other debt maturing before 2028.

Earlier this month, we increased our revolving credit facility from $650 million to $1.5 billion, gaining significant additional liquidity. In this quarter, despite the net loss, we recorded an income tax provision due to the impact of non-deductible items and the valuation allowance on the net value of deferred tax assetsDue to similar factors, our tax rates may fluctuate significantly in the future.

While we note the uncertainties and volatility in global trade and the macro environment, we have not observed any impact on customer behavior. In fact, we see customer demand accelerating. We do see certain equipment costs rising, which is felt across the industry. However, these impacts are expected to be relatively small. We will continue to closely monitor the environment and work with suppliers to mitigate the impact.

As demand for products and services remains strong, we are adjusting our plans to increase and accelerate investments in the platform to meet customer needs. Our operations and engineering teams are working tirelessly to deploy more capacity for customers more quickly. Our guidance today reflects not only the increased revenue expectations associated with higher customer demand but also our accelerated and increased platform investments and the impact of these investments on near-term profitability. Our guidance also includes the incorporation of Weights and Biases since the acquisition completed on May 5, which will be included in our financial statements.

Against a backdrop of strong demand, we expect second-quarter revenue to be in the range of $1.06 billion to $1.1 billion. Additionally, as we accelerate and increase investments to meet customer demand, we expect adjusted operating income for the second quarter to be between $140 million and $170 million. We expect second-quarter interest expenses to be in the range of $260 million to $300 million, influenced by similar factors as in the first quarter. We expect second-quarter capital expenditures to be between $3 billion and $3.5 billion, reflecting strategic decisions to accelerate platform investments to meet customer demand. Furthermore, we expect equity incentive expenses for the year to be slightly higher due to IPO-related grants.

For fiscal year 2025, we expect revenue to be in the range of $4.9 billion to $5.1 billion. We expect adjusted operating income to be between $800 million and $830 million, with capital expenditures between $20 billion and $23 billion, due to increased and accelerated investments in the platform to meet customer demand. The fiscal year 2025 guidance reflects the OpenAI contract signed in March, the recent $4 billion expansion contract with a major AI company, and the impact of Weights and Biases.

We are off to a good start in 2025. We continue to execute and rapidly scale our business, as evidenced by our financial performance and the new performance benchmarks set on the platform. The demand for cloud platforms built specifically for AI is stronger than ever, and we see tremendous opportunities. We will continue to invest to meet the needs of our rapidly growing customer base and solidify our industry leadership.

Thank you to our investors and analysts for your support and participation. We look forward to updating you on our progress in the coming quarters.

Morgan Stanley's Kate Wise:

Thank you all for taking my questions, and congratulations on your outstanding performance in the first quarter, which has set a good tone overall. It is indeed difficult to limit to one question. I think many people will have questions about the overall performance this quarter, but perhaps we can start by discussing the revenue performance that exceeded expectationsRevenue exceeded market expectations by 14%, which is a very significant level of outperformance, even for recently listed companies. Could you discuss what mechanisms enabled your company to achieve such a level of revenue outperformance within a single quarter? What different circumstances compared to initial expectations allowed the company to achieve this level of outperformance? Additionally, could you help us understand what is included in the order backlog, particularly regarding the OpenAI contract? What is the reason this contract is not included in the remaining performance obligations (RPO), but still counts towards the revenue backlog? Thank you.

Mike Intrator, CEO:

Thank you, Keith. We are pleased to have the opportunity to answer questions and take this opportunity to introduce the company. Regarding the revenue outperformance, what you are seeing is the coordinated strategic effort of the company in infrastructure investment to be able to build, scale, and provide computing power to customer contracts more quickly. We prioritize delivery speed and quality as the company's top focus. This outperformance is mainly attributed to our ability to drive this operational model in building delivery systems. Nitin, perhaps you can talk about the second part of the question?

Nitin Agrawal, CFO:

Okay. Thank you for your question, Keith. The accounting treatment of the OpenAI transaction has not been finalized, so we have not included it in the RPO. This is a large single contract at a single location. Given the scale and size of this transaction, we have taken a cautious approach to the accounting analysis. I want to clarify that the determination of the accounting treatment will not affect the timing of revenue recognition, the economics of the transaction, or cash flow.

Tash Wangan from Goldman Sachs:

Hello. Thank you very much. Congratulations on your company's performance in the first quarter as a public company. I hope there will be more good news in the future. I am curious if you could give us insight into the growth of the backlog or overall pipeline in the broader enterprise market? I think customer concentration is a concern for most of us. When you look at your sales pipeline, how will your revenue streams diversify more broadly as you progress over the next few quarters, so that investors and analysts can have confidence in the new business volume needed to meet forecasts for the upcoming quarters? However, it must be acknowledged that you have had a very good start, and your performance has exceeded expectations. Therefore, we can feel more confidence in the call. But I wonder if you could talk more about those forward-looking indicators that may continue to drive business growth.

Unnamed Speaker:

Certainly, thank you for your question. Kash, look, there are some basic facts regarding the customer base we serve, right? There are relatively few labs and large consumers (almost all of whom are our customers) that will drive these large-scale contracts. In the early days of the company, we consciously focused on those consumers with the highest demand, who need the largest scale infrastructure, and also require insurable infrastructure, bringing them onto our platform to let them experience working in an environment built specifically for their use cases. That said, we are also very excitedThis is actually one of the things we often discuss internally. It is also the focus of our sales activities, which is to bring more enterprise customers into our infrastructure over time. Therefore, I anticipate that over time we will have an extremely large customer base. This is the nature of the business we are in. Not every participant in the artificial intelligence field has the resources to train foundational models. That said, you will see tremendous growth momentum around enterprise customers. Hopefully, we can discuss the situation with Weights and Biases more specifically. Bringing them in has provided us with an excellent avenue to establish business relationships with essentially the entire enterprise customer base, which are large users of this infrastructure. We are very excited about this.

Mark Murphy from JP Morgan:

Thank you for answering the questions. First, I believe you mentioned new hyperscale customers and new enterprise customers, which is encouraging. Could you elaborate on this and provide some insights into the nature of these customers? What are they like? Are they concentrated in specific industries? What is the nature of these contracts, are they for inference or training? Do their terms align with what you typically see? Any information that can help us better understand the nature of these contracts would be very helpful. Thank you.

Unnamed Speaker:

Certainly. We are very careful with our wording because we have policy restrictions that prevent us from discussing specific customers in detail. Regarding hyperscale customers, there are not many of them, so you can look at our website and speculate on which new customer has been added.

We are very excited. We believe this relationship can bring significant growth as they begin to understand the experience of using the infrastructure and software stack we provide, as well as all the benefits that using this infrastructure will bring them. Regarding customer concentration, since we received a second question on this topic, at the end of the first quarter, no single entity accounted for more than 50% of our backlog. With new contracts coming in, we do not expect this to be the case in the future, but we do believe there will be large consumers.

Regarding enterprise customers, we are seeing many interesting customers from various industries starting to use our infrastructure and solutions as the engine for their companies. We mentioned IBM because we released a news release with them, but in reality, there are broader customer collaborations underway. As I said, this is one of the things we are most excited about internally.

Mark Murphy from JP Morgan:

Understood. This is very helpful. I believe you mentioned the rising costs of certain data center input devices. We still seem to be in a constrained environment. Can you explain which devices have seen price increases? Thank you.

Unnamed Speaker:

Yes. Thank you for your question, Mark. That question is more about the tariffs we have noticed. Certain components have indeed seen slight increases, but we do not expect any significant impact. We are closely monitoring the tariff situation and working with suppliers to mitigate any impact while building a resilient supplier pipelineCurrently, we expect that tariffs will not have a significant impact on our finances, but we are closely monitoring the situation.

Ramo Linshaw from Barclays:

How should we view capital expenditures? The numbers for the first quarter were slightly below market expectations, but seem to align with market expectations for the full year. Since capital expenditures drive revenue, there will obviously be fluctuations. Can you help us understand what happened in the first quarter and how we should view ongoing capital expenditures? Thank you.

Mike Intrator, CEO:

Thank you very much. Regarding capital expenditures, the guidance we released today reflects our investment in platform increases and accelerations to meet customer demand. That is what our numbers reflect. Our capital expenditures follow the investments we make for committed customer contracts and may fluctuate due to data center deployments and infrastructure. As we see strong customer demand, our increased capital expenditure guidance is to meet that strong customer demand.

Tyler Redkey from Citibank:

Thank you very much for taking my question. Regarding interest expenses, the data for the first quarter was high, and the guidance for the second quarter is also high. Can you elaborate on the main drivers? It sounds like it may be due to accelerated deliveries of certain components, but could you explain that in detail? If you see improvements in vendor terms, why wouldn't that be reflected in the capital expenditures for the first and second quarters?

Mike Intrator, CEO:

Thank you very much for your question. Regarding interest expenses, we have seen improvements in our vendor terms, which has reduced the days from payment to asset utilization, thereby lowering our capitalized interest costs for this quarter. We expect this situation to continue throughout the year, which is why you will see some interest expenses rise, as less interest cost is being capitalized. Additionally, as we build our capital expenditure portfolio to meet customer demand, you will see the level of interest expenses associated with that rise. These are the two factors affecting our interest expenses. At the same time, our cost of capital is gradually decreasing. The IPO is an important step in achieving this goal, and we will continue to lower our cost of capital.

Brad Zilnick from Deutsche Bank:

Thank you very much for taking my question, and congratulations on successfully entering the public market and performing well in the first quarter. Nitin, I understand your reasons for not providing guidance on cash flow and its dynamics, but could you give us some insights to help us think about the future pace and range of outcomes? Thank you.

Mike Intrator, CEO:

Yes, I think the best way to consider our cash flow is in terms of our capital expenditures and the contracts we continue to sign. Two factors to keep in mind here are the RPO elements, which are the signing, timing, and scale of contracts, and the payment terms of contracts that may be uneven, while the associated capital expenditures may also be uneven. We will do our best to ensure that it will fluctuate. We will do our utmost to work with you to provide as much visibility as possible in advance, but we do expect these to fluctuate in our businessBrett Bill from Jefferies:

Thank you, Mike. Regarding the software strategy, could you elaborate on what you think will happen in the long term, and what hybrid cloud service providers will do? There are many questions in this area, such as how they might overlap, or perhaps not overlap at all. For Nitin, aside from the hyperscale cloud service providers, when you consider situations beyond pipelines and core businesses, are you starting to see a broader range of other companies willing to get involved, not just the well-known ones we are aware of? Thank you for addressing these questions.

Mike Intrator, CEO:

I'll address the software stack question. When you consider the demand for AI and the types of computing required for AI, to build a software stack that makes this expensive infrastructure as efficient as possible, you have to go back to a blank slate and rethink how to support this infrastructure for customers. The intent of our company from the beginning was to start from the bottom of the stack and work our way up. The acquisition of Weights and Biases allows us to provide software solutions at the top layer of the AI stack. Our strategy is to fill the software stack with purpose-built tools designed to enhance efficiency, improve the quality of experience, and simplify the complexity of operating such infrastructure. As for how other cloud service providers will handle this, I'm not sure. They will formulate their own strategies, but we have already drawn a very direct connection between how we build our software and the performance of the infrastructure we need to operate. We will continue to do so.

Unnamed Speaker:

Regarding the customer pipeline question, Mike has already mentioned it in his prepared remarks. We are pleased to see the acceleration of enterprise AI adoption curves. This is reflected in our customer companies and signed contracts. Additionally, inference is taking up an increasingly large proportion of the infrastructure usage in our AI labs. These two factors are very encouraging for us in terms of the AI adoption curve and the use of dedicated infrastructure.

Gregg Moskowitz from Mizuho:

Thank you very much for answering my question. The guidance for adjusted operating margin for the next quarter and the full year is lower than market expectations. Could you tell us how much of an impact these accelerated investments will have on margins? I would also like to ask if this changes your view on CoreWeave's long-term structural margin leverage, I recall it being around 27% to 28% by 2027. Thank you.

Unnamed Speaker:

Thank you very much for your question. As we discussed earlier, when we build large-scale infrastructure, there will be a period of incurring certain costs before revenue starts to materialize. The scale of the infrastructure we are launching this year is impacting our margins. As we discussed in the conference call, we are accelerating and increasing our investments in the platform, which has affected our short-term margins. To give you a specific idea, we expect that by the end of this year, the deployed power will be more than twice the power we have deployed on our platform to date. This is the scale of infrastructure we are executing this year to meet customer demand

Gregg Moskowitz from Mizuho:

Thank you for answering the questions. The sales and marketing expenses for Coral Reef are relatively small compared to the overall business scale. However, Mike, you mentioned that expanding market development is one of the key priorities. Could you talk about your views on market development when expanding the business and the opportunities for developing a broader enterprise market? As a second small question, please have Nitin answer, we have received some questions regarding the RPO balance. The core metric has declined quarter-over-quarter, but you mentioned that OpenAI contributed $4 billion to the expansion. Given the focus on this metric, could you discuss the factors behind the changes? Thank you.

Mike Intrator, CEO:

When considering the company's sales and marketing strategy, you have already seen us build a sales and marketing team focused on those large customers that need to build and sign on to use large-scale computing. This has always been a primary goal for us in building the company, financing the company, and the focus of the sales and marketing team. But what you see, again using Weights and Biases as an example, illustrates well how we will continue to expand our customer base to many enterprise customers that require very powerful, efficient, and effective computing environments in the coming years. This will be a fundamental change we must make in the coming years.

At CoreWeave, we have always said we are customer-driven. Initially, the infrastructure was needed by these large companies. Now, as enterprise customers become more active, we will be guided by them to a broader customer base, and we will build organizations and continue to grow. As I mentioned, Weights and Biases is an important step in this direction, and we are building organizations to effectively establish relationships, introduce our solutions to them, and ultimately develop them into customers.

Unnamed Speaker:

Regarding the revenue backlog, our revenue backlog includes remaining performance obligations and other amounts we estimate will be recognized as revenue in future periods based on committed customer contracts, which is $25.9 billion this quarter, a year-over-year increase of 63%. As Keith asked earlier, the OpenAI contract has not yet been included in the RPO number because we are still studying its accounting treatment, but this will not affect the timing of revenue recognition or any economic factors related to the transaction.

Ultimately, based on executed contracts, the amount will reach $29 billion, but the $4 billion from the first quarter that has not been executed will show up later. The contract has been completed, and we are in the process of delivery and fulfillment.

Brad Fields from Bank of America:

Thank you very much. I would like to inquire about these large expansion deals. What drives these types of expansion deals? Is it based on the demand for next-generation GPUs? Is it the expansion of large models on CoreWeave? Or is it the addition of new models? I just want to understand what the catalysts for these deals are. What insights do you have on this? What is the cycle like when negotiating these types of deals?

Unnamed spokesperson:

Brad, that's a great question because it provides an opportunity to discuss important trends in the computing space. CoreWeave is uniquely positioned to serve significant consumers of AI infrastructure. We are able to understand how they use the infrastructure and what they do with it from training to inference. This is the most exciting part of the usage that companies like CoreWeave can see.

When customers come to us for scaling (which we call "layering"), they start with one contract and then work in a specially built environment, after which they come back time and again, layering one infrastructure contract on top of another. We see that the initial infrastructure is used to advance model training, and then as new, more advanced infrastructure replaces it, the original infrastructure still in the contract is repurposed for other uses, such as training smaller models or for inference. We have the best front-row observation of this.

When people come to us for infrastructure, they are purchasing cutting-edge infrastructure or previous-generation infrastructure, which enters into long-term contracts and forms the foundation of our business building and financing. They choose us because we have proven capable of providing incredible environments to run and train their models and perform inference. We also have a strong track record of excellent execution, enabling the infrastructure to go live in high-performance configurations.

Mike Sykos from Needham:

Thank you for taking the question. I want to echo the comments in the Q&A or what Mike mentioned in his prepared remarks regarding the broad demand for inference, and whether you see an acceleration in enterprise-level AI adoption curves.

From a metrics perspective, what can you point to? Is there more on-demand revenue this quarter that we can't see? I know we are focused on the RPO number, which decreased by 3% quarter-over-quarter, and the $4 billion that will be realized in the second quarter. We just want to confirm what you are seeing, as you clearly have more visibility into the pipeline and interest from weights and biases, etc. I just want to do a sanity check.

Mike Intrator, CEO:

I want to make sure a few points are clear, and then I will answer your question.

To reiterate, RPO is not decreasing; when you add RPO to the OpenAI contract, it increased significantly from about $15 billion last quarter to $25 billion, a substantial growth of 60%. Additionally, there is $4 billion that will enter backlog orders, including RPO and OpenAI contracts, cumulatively bringing the company $29 billion in revenue obligations. We are very excited about this, as it drives us forward and continues to extend our contractual obligations with customers, providing them with infrastructure and receiving payments.

Regarding inference, among all trends, inference is the most exciting for me. This is because it represents the monetization of AI and the return on investment in AI. When customers shift from continuous training to a combination of training and inference, it indicates that customers using our infrastructure have the opportunity to bring products to market and generate revenue.**

This is exactly what artificial intelligence needs to continue to expand, grow, and succeed. Therefore, as I mentioned, the most exciting trend may be seeing reasoning gain real traction.

Ben Wright from Melius Research:

Thank you for the opportunity to ask a question. I hope you can hear my voice clearly. I would like to further understand the capital expenditure regarding the $3 billion increase over expectations. Does this include factors related to rising costs, or is it due to changes in the GPU mix that you did not anticipate before? I know this is due to increased demand and some backlog orders. Additionally, if I may ask, Nitin, I think it would be helpful to know how you plan to finance the $3 billion capital expenditure that is above expectations? Given the low operating returns in the current quarter, I am unsure if cash flow estimates will decline. So, I would like to understand the sources of funding for this additional capital expenditure. Thank you.

Mike Intrator, CEO:

Hi, Ben. When you see our capital expenditure increase, it indicates that new customers are signing contracts with us that require us to invest more to build more infrastructure. This is a virtuous cycle based on success. Our $3 billion increase in spending is not due to price increases. In terms of the supply chain, we have not seen any factors that necessitate such a significant increase in spending. We have been working to diversify the supply chain and improve its resilience in the current environment. But this does reflect the increase in new customers who are purchasing more infrastructure from us for the remainder of this year. To add, the increase in capital expenditure we are seeing is primarily driven by customer demand growth. This is also reflected in our raised quarterly and annual revenue guidance.

I will make a brief closing statement here. Thank you all for taking the time to be with us today. Thank you for attending our first quarterly earnings call. Personally, this is very exciting for the company and for me. This is a tremendous achievement. Thank you to everyone who has helped build CoreWeave, including employees, investors, customers, and suppliers. This is an exciting time for the company. A lot of work is happening as we continue to innovate and build the infrastructure needed to push the boundaries of artificial intelligence. There are a few points I want everyone to understand about where we are now and our future direction. First, we had an outstanding quarter. We achieved a 420% year-over-year revenue growth. We just signed another $4 billion contract, which will increase our revenue backlog. Demand remains strong, and we are raising our revenue guidance. These are incredible achievements. Second, we continue to drive this growth through success-based financing that matches customer contracts. We combine them, finance them, build them, and then deliver. Third, what you see is a company with deep technical strength. We are building and innovating at all levels to differentiate our platform and drive what may be the most significant technological revolution of our lifetime. Thank you, and we look forward to updating you on our progress in the next quarter