CoreWeave (CRWV.US) 2025年第三季度业绩电话会
文章语言：
简
繁
EN
Share
Minutes
原文
会议摘要
Core Weave achieved a record-breaking third quarter with $1.4 billion in revenue, a 134% increase year-over-year, and added over $25 billion to its backlog, totaling over $55 billion. The company expanded its active power footprint to 590 MW, grew contracted power to 2.9 GW, and secured major compute contracts with Meta and OpenAI. Core Weave diversified its customer base, launched new products like AI Object Storage, and expanded its global network, solidifying its position as the essential cloud for AI. Despite temporary infrastructure delays, the company projects significant growth in 2026, with CapEx expected to more than double, underscoring its leadership in the rapidly expanding AI cloud market.
会议速览
Core Wea's Q3 2025 Earnings Call Highlights and Q&A SessionThe earnings call for Core Wea's third quarter 2025 was opened with standard legal disclaimers regarding forward-looking statements, followed by a brief introduction to the Q&A session where investors can engage with the CEO and CFO. The operator reminded participants of the call's procedures and the availability of the earnings release and replay on the company's website.
Core Weave Surpasses Expectations with 134% Revenue Growth, Leading AI Cloud PlatformCore Weave achieved 134% year-over-year revenue growth, reaching $1.4 billion, driven by robust customer demand and strategic large-scale compute contracts with major clients. The company's active power footprint expanded to 590 MW, with 2.9 GW of contracted power capacity, positioning it for future growth. Notably, the number of customers exceeding $100 million in revenue over the past year tripled, highlighting the platform's expanding role in powering critical AI workloads for global enterprises.
Core Weave's Strategic Growth: Expanding Data Centers, Enterprise Partnerships, and AI InnovationsCore Weave is transforming operations with AI, securing major enterprise partnerships, expanding its data center footprint globally, and leading in AI workload performance, despite facing temporary delays in construction. The company is scaling its global revenue organization, entering the public sector with Core We Federal, and continues to attract top talent and significant investments in capacity to meet surging demand for its cloud services.
Core Weave's Leadership in AI Cloud: Unmatched Performance, Innovation, and ExpansionCore Weave, a pioneering AI cloud provider, showcases unmatched leadership in AI cloud technology, achieving industry-first recognitions, robust customer satisfaction, and strategic acquisitions. With cutting-edge infrastructure, a diverse suite of services, and a focus on expanding into new markets, Core Weave positions itself as the essential cloud for AI, enabling rapid growth and innovation in the AI sector.
Record Q3 Revenue, Enhanced Customer Diversification, and Growing AI Infrastructure InvestmentsQ3 revenue soared 134% year-over-year to $1.4 billion, propelled by robust demand and effective execution. Notable customer diversification reduced concentration, with over 60% of revenue backlog now attributed to investment-grade clients. Strategic investments in data center and server infrastructure, alongside scaling sales and marketing efforts, underpin continued growth and capture of expanding AI opportunities. Adjusted EBITDA more than doubled to $838 million, marking a significant milestone in operational efficiency and financial performance.
Record Q3 Performance, Strategic Capital Moves, and Future Growth PlansThe dialogue highlights a record-breaking Q3 with a $1.9 billion CapEx reduction due to delivery delays, strategic amendments to financing facilities lowering costs, and a $14 billion debt and equity transaction history. It outlines Q4 and 2025 guidance, including revenue, adjusted operating income, and CapEx expectations, emphasizing the company's robust liquidity, tax benefits, and plans for significant 2026 CapEx growth, reinforcing confidence in long-term business trajectory and market leadership.
Navigating AI Infrastructure Supply Chain ChallengesDiscusses strategies for managing delays in AI infrastructure delivery due to supply chain issues, emphasizing diversification of data center providers and self-build efforts to mitigate risks and ensure customer satisfaction.
Data Center Infrastructure Scaling Amid Industry ChallengesDiscussion highlights successful scaling of infrastructure delivery despite systemic industry delays, emphasizing non-impact on backlog and contract value extraction.
Fungibility of AI Infrastructure: Ensuring Flexibility Across Customers and Use CasesDiscusses the fungibility and flexibility of built infrastructure, emphasizing its adaptability for various customers and applications, including training and inference, ensuring optionality and effective use.
Core Weave's Strategic Growth and Unique Value Proposition in GPU InfrastructureDiscussion highlights Core Weave's strategic diversification, emphasizing reduced impact of individual delays on revenue as the portfolio expands. The dialogue underscores Core Weave's competitive edge in delivering customized GPU solutions, particularly in the face of hyperscalers' dominance, attributing its lead to specialized cloud offerings and capacity expansion.
Revenue Impact of Delayed CapEx and Forecasted Growth in 2026The dialogue discusses the implications of delayed capital expenditures on 2026 revenue, assuring that the majority of delays will be resolved by Q1 of next year, allowing for accelerated growth. It highlights the ability to adjust delivery dates without impacting contract values and forecasts a significant increase in 2026 CapEx compared to 2025, driven by strong customer demand and backlog growth.
Inquiry on Supply Chain Impacts on Customer Acquisition and Nvidia Deal's Influence on BacklogA query explores if supply chain issues affect new customer acquisition or post-signing ramp-ups, and seeks clarification on whether the Nvidia deal is reflected in the backlog metric, highlighting its unique structure and potential market expansion.
Revolutionizing Data Center Capacity Utilization and Revenue GrowthA strategic Nvidia deal enables flexible capacity allocation, enhancing service to emerging companies and AI labs, while maintaining financial discipline. Despite a 40% CapEx reduction due to a single data center delay, the company anticipates robust revenue growth from its expanding portfolio and 2.9 GW contracted power, scheduled to come online over the next 12-24 months.
Exploring Financing Options for Enhanced Data Center Expansion and Client ServiceThe dialogue discusses the company's strategy to consider all cost-effective financing methods, including leasing, to expand its data center footprint and meet client demands, ensuring all necessary resources are secured before committing to deals.
Investor Inquiry Resolved on Communication LineAn investor's question is successfully addressed after confirming improved communication, ensuring clarity and effective dialogue for future inquiries.
Balancing Self-Built and Third-Party Data Centers for Optimal CapEx and Time to MarketDiscusses the strategic blend of self-built and third-party data centers to manage capital expenditure and expedite market entry, emphasizing the necessity of a diversified approach to mitigate risks in complex global supply chains.
Pacing Capacity Sales Amid Insatiable Demand and Diversification StrategyDiscusses strategic approach to selling remaining capacity, emphasizing diversification and targeting investment-grade customers amidst overwhelming global demand for infrastructure driven by tech giants, AI labs, and enterprises.
Addressing Data Center Bottleneck: IP Contributions and Contract LearningsThe dialogue explores the impact of IP contributions on data center builds, emphasizing learnings from delays to mitigate bottlenecks in future contracts.
Building Resilience and Capacity Amid Global Supply Chain ChallengesA discussion highlighted the company's strategic investments in internal capacity and third-party relationships to navigate a persistently constrained global market, emphasizing successful execution in expanding AI cloud services and anticipating ongoing industry challenges.
要点回答
Q：What are the financial results of Core Weave for the third quarter?
A：Core Weave delivered revenue of $1.4 billion, up 134% year over year. The company added over $25 billion in revenue backlog in Q3, bringing the total to over $55 billion, nearly double Q2 and approaching four times year to date.
Q：What are some of the key operational achievements by Core Weave in the third quarter?
A：Core Weave expanded its active power footprint by 120 MW to approximately 590 MW and grew its contracted power capacity over 600 MW to 2.9 GW. The company executed large-scale compute contracts with customers like Meta and OpenAI and grew its relationship with a leading hyperscaler, marking the sixth contract with that customer. Core Weave has also reached $50 billion in RPO, faster than any cloud in history.
Q：How is Core Weave being utilized by AI organizations?
A：Core Weave is being used by sophisticated AI organizations to accelerate breakthroughs in AI innovation. The performance, flexibility, ability, and reliability of Core Weave Cloud have led customers to consistently expand their usage with the company.
Q：What is the nature of the relationship between Core Weave and AI native enterprises?
A：AI native enterprises across sectors are embracing Core Weave to transform operations and unlock new sources of innovation, productivity, and growth. Core Weave serves as a cloud partner for these enterprises, including companies like Jasper in digital marketing and CrowdStrike in cybersecurity.
Q：How is Core Weave gaining traction in the public sector and what is its relation with NASA?
A：Core Weave is making inroads into the public sector, catering to agencies with unique performance and security requirements. The company has launched Core Weave Federal to bring cloud services to US government agencies and the defense industrial base, with NASA already using its services for scientific exploration.
Q：Who is John Jones and what role has he joined Core Weave to fulfill?
A：John Jones has joined Core Weave as the first Chief Revenue Officer. He previously served as the global head of startups and venture capital at AWS. In his new role, he will play an important part in scaling Core Weave's global revenue organization and driving expansion during the next phase of growth.
Q：What is the name of the customer affected by the delays and what agreement adjustments have been made?
A：The customer affected by the delays is agreeable to adjusting the delivery schedule and extending the expiration date of the contract.
Q：What is Core Weave's position in the AI cloud market?
A：Core Weave is the only cloud provider to submit ML perf inference result for GB 300 S, setting the benchmark for real-world AI performance, and has received the highest possible distinction, the Platinum Cluster Max ranking, from Semi-Salic ISIS for two times, highlighting its dominance in the AI cloud market.
Q：What is Core Weave's role in the AI development ecosystem?
A：Core Weave supports AI pioneers and helps put AI to work from tools for developers to solutions for the physical world. It has been expanding its role through acquisitions and integrations to help with the development and monetization of AI.
Q：What are the details of the new storage service launched by Core Weave?
A：Core Weave launched Core Weave AI Object Storage, a fully managed storage service that eliminates friction of moving data with no egress or transaction fees, providing the highest amount of throughput for AI workloads at a cost that is more than 75% lower.
Q：What are the details of the new partnership mentioned in the speech?
A：The new partnership is with CrowdStrike, which is expected to unlock and accelerate partner-driven growth. The partnership with vast data is another example of accelerating the product portfolio and partner go-to-market motions.
Q：What does the company consider as the essential cloud for AI?
A：Core Weave considers itself the essential cloud for AI, combining technical and operational excellence with a rapidly diversifying customer base. It provides the most performing infrastructure, the fastest time to market, and the most advanced capabilities, enabling AI innovators to move faster, scale smarter, and achieve outcomes not possible elsewhere.
Q：What are the highlights of the third quarter results?
A：The third quarter results were highlighted by robust customer demand, strong execution, and a reduction in customer concentration. Q3 revenue was $1.4 billion, up 134% year over year, driven by new and existing customers. The revenue backlog for the quarter ended at $55.6 billion, nearly doubling in the third quarter alone, with a diverse customer base where no single customer represents more than approximately 35% of the revenue backlog. More than 60% of the revenue backlog is tied to investment-grade customers.
Q：What were the interest expenses for Q3, and what factors influenced them?
A：Interest expense for Q3 was $311 million compared to $104 million in Q3 of 2024. Factors influencing the increase included increased debt to support the scaling of infrastructure and better interest rates on debt, partially offset by the benefit from improved interest rates.
Q：How did adjusted net loss and EBITDA for Q3 compare to the prior year?
A：Adjusted net loss for Q3 was $41 million compared to approximately break-even in Q3 of 2024, while adjusted EBITDA for Q3 was $838 million compared to $379 million in Q3 of 2024, marking more than a two-fold year-over-year increase. The adjusted EBITDA margin was 61%.
Q：What was the impact of delays in power shell delivery on Q4 results and future expectations?
A：The delays in power shell delivery from a data center provider impacted Q4 results and future expectations by causing a reduction in CapEx from prior guidance, which will result in a corresponding increase in construction in progress. This will affect financial metrics and the profile of future CapEx. While the delays are temporary, they will still affect Q4 financials and 2025 guidance.
Q：What systemic challenges are impacting the delivery of global infrastructure required for artificial intelligence?
A：Systemic challenges within the supply chains are necessary to deliver the global infrastructure required for artificial intelligence, which are frustrating for both clients and the company. These challenges relate to managing through the complex environment and delivering infrastructure that is currently overwhelming the supply chains.
Q：What is the significance of the contract value adjustment in the context of delays?
A：The fact that the ultimate end customer has shifted the contract back to allow the company to deliver the full contract value despite the delays underscores the value that customers derive from their infrastructure.
Q：What is the main challenge in the AI infrastructure supply chain?
A：The main challenge in the AI infrastructure supply chain is at the powered shell level, which is not related to power availability but to the powered shell itself.
Q：How does the company address potential risks with specific data center providers?
A：The company works with all data center providers to facilitate the delivery of infrastructure. They have had success in delivering infrastructure as they scale, and the specific identity of the data center provider is not as important as addressing the systemic problem affecting the industry.
Q：Is the company's infrastructure flexible and fungible for different customers?
A：Yes, the company's infrastructure is designed to be fungible and flexible, able to be transferred from one client to another, and can be used for both training and inference. They aim to maintain as much optionality and flexibility within their infrastructure build as possible.
Q：What makes the company's software suite a valuable component for their infrastructure?
A：The company's software suite is critical for the infrastructure's flexibility and fungibility. It allows for the effective use of the infrastructure and has consistently been recognized as the best solution for this type of infrastructure.
Q：What unique value proposition does the company have in the context of AI infrastructure?
A：The unique value proposition of the company lies in their ability to quickly and effectively stand up GPU clusters, which is a capability that sets them apart in a market where hyperscalers are awarded hundreds of millions of dollars in contracts. The company's ability to rapidly deploy infrastructure is highlighted as a unique advantage.
Q：What determines the impact of data center delays on the company's financial numbers?
A：The impact of data center delays on the company's financial numbers is determined by the size of the individual data centers relative to the entire portfolio. As the portfolio grows larger, the impact of a data center being a few weeks late becomes less significant.
Q：How does the company plan to deliver GPUs faster and stay ahead of competitors?
A：The company plans to deliver GPUs faster by customizing its cloud for the specific use case, which is required for running Nvidia's ML perf. This enables the company to stay ahead of competitors through product offering and additional capacity, further decommoditizing the compute they deliver.
Q：What are the expected implications of delays on 2026's revenue and growth rate?
A：The expected implications of delays on 2026's revenue and growth rate are minimal as the majority of the delays should be resolved within the first quarter of the next year. The CapEx pushout in Q4 is expected to be addressed in Q1, and the company will ramp capacity through Q1. Revenue associated with the customers is not impacted since delivery dates have been adjusted to maintain the customer's capacity and contract value. The company expects 2026 CapEx to be significantly higher than 2025 due to continued strong customer demand.
Q：Does the Nvidia deal impact the backlog metrics and what does it mean for future revenue?
A：The Nvidia deal does impact the backlog metrics, as the contract allows for capacity to be interrupted and resold to different customers. This flexibility provides profit opportunities for a range of smaller customers. Nvidia's contract is reflected in the revenue backlog but not to a large extent in the RPO, since the capacity is expected to be resold. If not resold, the capacity will remain committed to Nvidia and be recognized as revenue.
Q：What is the company's strategy for the Nvidia deal and how does it align with revenue recognition rules?
A：The company's strategy for the Nvidia deal involves offering services profitably to a range of smaller customers due to the flexibility in the contract to interrupt and resell capacity. Nvidia's contract is included in the revenue backlog but not in the RPO to a large extent, as the accounting rules require excluding the amount expected to be resold to other customers. If not resold, the capacity will be recognized as revenue.
Q：What is the purpose of the contract with Nvidia for computing infrastructure?
A：The contract with Nvidia is designed to provide infrastructure to emerging companies and startups that require computing infrastructure to build their business. It represents a powerful tool for the resiliency and expansion of new companies and is structured as a deal with Nvidia, where they fully underwrite the economics.
Q：Why has the company reduced CapEx by 40% for the year?
A：The CapEx reduction is attributed to a delay with a single data center provider. Most of the delay's impact will be recognized in Q1 and Q4, with a significant increase in construction in progress related to the build-up associated with it.
Q：What factors are considered when choosing a method to finance the company's expansion?
A：The company considers the full suite of potential ways of financing expansion and selects the most cost-effective method to increase scale and serve clients. This may include leasing but also involves creating structures that provide access to capital. The company ensures physical data centers, power, GPUs, and financing are secured for customers.
Q：How does the company plan to balance third-party data center usage with self-builds?
A：The company plans to use a combination of third-party data center providers and self-builds as part of a strategy to de-risk delivery across the portfolio. Self-builds help the company to be closer to physical infrastructure and embedded in the global supply chain for better first-hand information. They consider self-builds an additional piece of the puzzle that enhances their effectiveness in managing a complex supply chain.
Q：What is the strategy for contracting out the remaining power capacity?
A：The strategy for contracting out remaining capacity involves being thoughtful about diversification across the cloud and allocating infrastructure to different applications that will be meaningful contributors to future work. The company aims to ensure that these applications can successfully launch their products. The existing pipeline has diverse customer representation, with no single customer accounting for more than 35% of revenue backlog, and 60% of the backlog is with investment-grade customers.
Q：What control does the company have over the bottleneck issue in the data center build-out?
A：The company has been operating in a systemically supply-constrained market for three years and has developed an organization to help build and deliver additional capacity on the self-build side. By embedding themselves in supply chains, understanding power contracting, and the construction of powered shelves, the company aims to be successful in a challenging environment. They view the current delays as a learning experience for positioning themselves to solve for the bottleneck issue.
