This telephone conference focused on Core Wave Company's financial performance for the second quarter of 2025, emphasizing the importance of future prospects and reminding participants to pay attention to potential risks. The meeting provided an interpretation of the financial report, including a comparison of GAAP and non-GAAP financial indicators, and promised to provide a replay of the meeting through the website.

In the second quarter, the company's revenue increased by 207% year-on-year to reach $1.2 billion, with adjusted operating income reaching $200 million for the first time, achieving a double milestone. The demand for AI cloud services is strong, and the company is actively expanding its capacity. By the end of the second quarter, it has nearly 470 megawatts of active power and has signed total power contracts of 2.2 gigawatts. The customer base has expanded from large enterprises to AI startups in various sectors such as finance, healthcare, and industry. The demand for cloud services continues to grow, making the company the preferred platform in the industry.

The dialogue detailed the latest developments in AI cloud services and storage products, including the large-scale deployment of NVIDIA technology, innovative archival storage products, third-party storage system integration, full-stack observability features, and the launch of AI model inference services. These initiatives are aimed at enhancing the performance and reliability of AI workloads, optimizing storage costs, supporting large-scale production environments, meeting customer needs through flexible capacity products, and accelerating the development of the AI ecosystem.

The conversation discussed deepening participation in the capital market, completing the issuance of high-yield bonds and GPU financing, as well as vertical integration with Core Scientific to reduce capital costs, improve operational efficiency, and strengthen customer service capabilities. This is aimed at leveraging existing resources and new investments to expand data center scale, enhance AI cloud platform strength, and is expected to bring significant cost savings and growth opportunities in the coming years.

The company announced that its second quarter revenue reached $1.2 billion, a year-on-year increase of 207%, primarily driven by customer demand. Revenue backlog reached $30.1 billion, an 86% increase year-on-year. Capital expenditures were $2.9 billion, used for rapid expansion of data centers and server infrastructure. Despite a net loss of $291 million, adjusted EBITDA reached $750 million, a threefold increase year-on-year, with an adjusted EBITDA profit margin remaining at 62%. The company raised $6.4 billion through the capital markets to optimize its capital structure to support rapid growth.

Since the beginning of 2024, the company has successfully raised over $25 billion through a series of innovative financing initiatives to support the infrastructure development of top AI labs and companies globally. This includes the issuance of unsecured high-yield bonds for the first time, additional bond offerings, and the completion of a third delayed draw term loan facility totaling $11.9 billion in financing for the OpenAI contract. These achievements not only demonstrate the company's ability to lower capital costs but also highlight its enhanced depth and breadth in the capital markets, meeting the goals set during the IPO. Additionally, despite reporting a net loss in the second quarter, the company still needs to record income tax expenses due to the effects of non-recurring items and deferred tax asset valuations, which could result in fluctuations in future tax rates due to similar factors.

The company estimates revenue of 12.6 billion to 13 billion in the third quarter of 2025, adjusted operating profit of 16 billion to 19 billion, and capital expenditures of 29 billion to 34 billion, driven by demand-driven capital expenditure growth. Full-year revenue forecast raised to 51.5 billion to 53.5 billion, with capital expenditures of 200 billion to 230 billion. The company will continue to invest to meet customer demand and strengthen its market leadership position.

In the conversation, it was mentioned that due to a potential hearing impairment, it was decided to temporarily skip the current issue and instead discuss the next agenda item. The unresolved issue will be returned to later to ensure the smooth progress of the meeting.

Discussed the importance of renewing contracts with large-scale clients, emphasizing business expansion through hardware upgrades. At the same time, proposed measures such as acquisitions, vertical integration, and optimized deployment processes to increase capital return and reduce costs. These strategies aim to accelerate business growth, enhance customer value, and ensure the company continues to achieve outstanding performance.

Discussed the flexibility of AI infrastructure in handling training and inference tasks, emphasizing the significant growth of inference workloads and its importance for the commercialization of AI. At the same time, it points out the bottleneck of power facilities in the current supply chain, as well as supply constraints for key components such as GPUs and medium-voltage transformers. These factors together constitute the main challenges facing the industry.

Discussed the global interest of various governments in building AI data centers, particularly the construction of modern AI data centers. Service providers need to consider the acceptance of American technology when dealing with different countries. Mentioned progress in cooperation with Canada and Europe, and clarified the timing of contracts signed with large cloud service providers recently, with one contract already reflected in the second-quarter revenue data and the other contract to be reflected in the third-quarter data.

The conversation discussed the details of the GPU computing service contract, including the scale of core GPU services and bias extension, but specific data has not been disclosed yet. The mentioned service contract is part of the company's product portfolio and a comprehensive update on revenue and order status will be provided through the third quarter financial report.

The conversation discussed the structural supply constraints in the data center market, emphasized the importance of relationships with major consumers, and pointed out that different workloads have varying sensitivity to latency. For applications like inference chains, computational power is more important than latency, and deploying closer to population centers can provide low-latency options. As AI advances, the demand for workloads that are not sensitive to latency will increase, and these can be placed in more remote areas.

Discussed the economics of inference and training in AI computing services, pointing out that both are similar in economics under long-term contracts, but the initial release of new models may cause short-term demand spikes. Introduced upcoming flexible capacity products, including spot pricing models, expected to provide different generational GPUs, and explored the differences with traditional reserved or pay-as-you-go models.

The conversation discussed the challenges that businesses face in high-demand markets, which include rapidly depleting their constructed computing resources, mainly due to an increase in demand for model services from existing or new customers. Businesses are striving to expand their capacity to provide more computing resources with the goal of attracting new users, supporting emerging companies, and discovering new computing needs, thus building services that are adapted to market demands. However, achieving this goal in a demand-driven market environment is quite challenging.

Discussed the significant growth in customer demand for computing power, especially the demand from large customers for high-performance computing, and how the company can meet this demand by redeploying old GPU clusters, including strategies such as using computational resources for inference tasks and extending contract periods.

Discussed the timing arrangement of capital spending and income growth in the fourth quarter, explaining that due to systematic planning for power supply and infrastructure construction, income and capital spending will show a lagging trend. At the same time, emphasized the short-term profit pressure caused by the early occurrence of costs such as data center leasing in the early stages of large-scale expansion, as well as the preparatory work carried out to ensure the smooth start of customer contracts.

The conversation revolves around the demand for different segments of the AI market, pointing out that in addition to large-scale enterprises, the demand from AI labs and companies themselves is also growing, especially companies in emerging fields such as Moon Valley, whose products are bringing new growth points to the market. Customers in the financial sector not only represent enterprise-level demand but also provide new growth points unrelated to existing sources of revenue.

The dialogue discussed the huge demand for computing resources from major banks and AI companies, and how this has promoted the application of AI technology in different economic sectors. At the same time, the importance of establishing partnerships between enterprises and AI suppliers was mentioned, as well as the software and hardware support provided by these suppliers to meet customer needs. Additionally, the dialogue emphasized the progress of the capital market, especially the participation of the debt market, in reducing borrowing costs and its economic impact on building and expanding infrastructure.

The conversation centered around the sales team's interaction with customers, mentioning sales progress and customer acquisition strategies in the context of increased operational investment. It also discussed cloud service resource allocation issues, especially the availability of on-demand and Spot instances, emphasizing the importance of improving resource allocation efficiency to attract new customers.

The dialogue focuses on the outstanding performance of the two companies in the field of AI infrastructure integration, including the development of three new products: the Weight Deviation Integrated Product created in collaboration with the Core Weight Team, the Weight Deviation Inference Product, and the Weave Product, these products have improved customer efficiency in using AI infrastructure. In addition, the importance of on-demand and spot computing models in expanding into new markets is emphasized, as well as the company's confidence in becoming a milestone in the industry by 2025, demonstrating successful continuous growth and strategic implementation.