Google (Minutes): Major Cloud Deals Double, Compute-Capacity Crunch to Persist into Next Year

The following are the$Alphabet(GOOGL.US) FY25 Q2 Earnings Call Minutes, for earnings interpretation please refer to《Google: AI Wolf Hasn't Arrived, Ad King Steadily Holds the Fort》

I. Review of Core Financial Information

1. Overall Performance:

a. Revenue: Consolidated revenue was $96.4 billion, up 14% year-over-year, with constant currency growth of 13%. Search ads, YouTube ads, subscription platforms and devices, and Google Cloud all achieved double-digit growth.

b. Costs: Total revenue costs were $39 billion, up 10% year-over-year. TAC costs were $14.7 billion, up 10% year-over-year; other revenue costs were $24.3 billion, up 10%, mainly driven by content costs and depreciation expenses.

c. Expenses: Operating expenses totaled $26.1 billion, up 20% year-over-year. This includes legal and other expenses of $1.4 billion; R&D expenses up 16% year-over-year, mainly due to increased compensation and depreciation; sales and marketing expenses up 5% year-over-year.

d. Profit: Operating profit was $31.3 billion, up 14% year-over-year, with an operating margin of 32.4%; net profit was $28.2 billion, up 19% year-over-year; earnings per share were $2.31, up 22% year-over-year.

e. Cash Flow: Free cash flow for the quarter was $5.3 billion, with cumulative free cash flow over the past twelve months totaling $66.7 billion; quarter-end cash and marketable securities balance was $95 billion.

2. Performance by Segment:

a. Google Services

Revenue was $83 billion, up 12% year-over-year, mainly driven by strong growth in search and YouTube ads, partially offset by a decline in network ad revenue.

Search and other related revenue grew 12% year-over-year; YouTube ad revenue grew 13% year-over-year; network ad revenue was $7.4 billion, down 1% year-over-year; subscription platforms and device revenue was $11.2 billion, up 20% year-over-year.

Operating profit was $33.1 billion, up 11% year-over-year, with an operating margin of 40.1%.

b. Google Cloud

Revenue was $13.6 billion, up 32% year-over-year; including high growth in core GCP products and AI products, as well as increased seat count and ARPU for Google Workspace.

Operating profit was $2.8 billion, with the operating margin increasing from 11.3% to 20.7%; backlog was $106 billion, up 38% year-over-year, and up 18% quarter-over-quarter.

c. Other Bets

Revenue was $373 million; operating loss was $1.2 billion.

d. Capital Expenditure (CapEx)

Capital expenditure for the quarter was $22.4 billion, with approximately two-thirds allocated to servers and one-third to data center and network equipment construction.

e. Shareholder Returns

Returned $13.6 billion to shareholders through stock repurchases; paid cash dividends of $2.5 billion.

3. Outlook for Q3 2025 and Future Development

a. Revenue: At current spot exchange rates, Q3 is expected to benefit from exchange rate tailwinds, but exchange rate volatility remains uncertain.

b. Google Services: Ad revenue in the second half will face high base effects from the financial services sector in 2024 and high base effects from ad spending during the U.S. election in the second half.

c. Google Cloud: Product demand remains strong, backlog remains high; capacity deployment progress will determine revenue growth rate, supply-demand tension is expected to continue until 2026.

d. Investment and Capital Expenditure: Full-year 2025 capital expenditure is expected to be approximately $85 billion, higher than the previous forecast of $75 billion; mainly reflecting accelerated investment in servers and data centers. Looking ahead to 2026, CapEx is expected to further increase.

e. Expenses: Depreciation expenses in Q2 increased by $1.3 billion year-over-year, reaching $5 billion, up 35% year-over-year, with Q3 depreciation growth expected to further accelerate; key investment areas are expected to see an increase in headcount quarter-over-quarter; the launch of the new Pixel series in August will incur related expenses.

II. Detailed Content of the Earnings Call

2.1 Executive Statements on Core Information

1. Overall Performance

The second quarter brought comprehensive strong growth momentum to Google. The search business achieved double-digit revenue growth, AI innovation continued to accelerate, and the Gemini series models performed excellently. AI Overviews had over 2 million active users, covering more than 200 countries and regions;

AI Mode launched in the U.S. and India, with over 100 million monthly active users and positive user feedback. YouTube and subscription businesses maintained high growth momentum, with Shorts unit time revenue matching traditional videos, even exceeding in some markets. Google Cloud annualized revenue exceeded $50 billion, with strong demand for AI product portfolios, and record high customer renewals and large contract signings.

2. Artificial Intelligence (AI)

a. Infrastructure: Google has built globally leading AI-optimized data centers and cloud regional networks, covering the industry's most extensive TPU and GPU combinations, meeting various needs from training to inference. Models like Veo 3, Flashlight, and Deep Think support product innovation with leading speed and performance, and research institutions like Safe Superintelligence and Physical Intelligence have dedicated TPUs, verifying their differentiated advantages.

b. Models and Research: The Gemini 2.5 series hybrid inference models have achieved leading results in mainstream benchmarks, with the Flash model continuously upgraded, and Flashlight being faster and lighter. Deep Think reached gold medal level in the International Mathematical Olympiad, attracting tens of millions of developers to build projects. Veo 3 became a hit in video generation, and the Flow tool collaborated with the Gemini App to generate over 70 million videos, with enthusiastic user response after the "photo to video" feature was launched.

c. Products and Platforms: Gemini monthly active users exceeded 450 million, with daily request volume growing over 50% quarter-over-quarter. Platforms like Workspace and Chrome have become important interfaces for AI applications, with Meet supporting 50 million meeting records generated in June alone; Google Vids monthly active users approached one million. The monthly processing of AI tokens exceeded 980 trillion, doubling since the IO conference announcement.

3. Highlights of Each Business Segment:

a. Search

The search business is at a critical stage of AI-driven transformation. AI is reshaping the way users interact with information, making query content more complex and diverse. Overall search queries and commercial query volumes continue to grow year-over-year.

-AI Overviews: This feature is supported by Gemini 2.5, with industry-leading response speed. Global display query volume increased by over 10%, covering more than 200 countries, supporting 40 languages, with over 2 million monthly active users.

-AI Mode: Equipped with stronger inference and multimodal capabilities, particularly suitable for handling longer and more complex problems. Launched in the U.S. and India, with over 100 million monthly active users and positive user feedback.

-Multimodal Search: The combination of Google Lens and Circle to Search is growing rapidly, especially among younger user groups. Circle to Search usage is significantly increasing, particularly in real-time calls in gaming scenarios.

-Deep Search and Personalized Replies: As the next phase of AI Mode, Deep Search will introduce more in-depth research tools to continuously enhance the user search experience.

b. Cloud

Google Cloud maintains strong growth momentum, with strong demand for AI product portfolios. The number of transactions with contract value exceeding $250 million doubled year-over-year, and the $1 billion deals signed in the first half of 2025 have already matched the total for 2024. The number of new GCP customers increased by nearly 28% quarter-over-quarter.

Over 85,000 enterprises (such as LVMH, Salesforce, DBS) use Gemini, driving usage up 35 times year-over-year. In terms of infrastructure, the launch of Anywhere Cache can reduce inference latency by 70%; Rapid Storage can increase storage latency by 5 times, outperforming major cloud service providers. Through PyTorch, JAX optimization, and open-source AI software packages, further support for model training and deployment is provided.

-AI Agent deeply integrated into Cloud products: Wayfair uses the Gemini database to optimize personalized recommendations; Vantel leverages intelligent data agents and BigQuery for quick feedback response; Target uses Gemini security agents to enhance network defense; Capgemini uses development agents to optimize software delivery processes; BBVA saves approximately 3 hours of repetitive work per person per week through Workspace + Gemini, promoting it to global employees.

-Building a thriving AI agent ecosystem: Agent Development Kit downloads exceeded 1 million; Agentspace pre-launch reservations exceeded 1 million, with U.S. customers like Gordon Food Service deploying applications for intelligent decision-making.

c. YouTube

YouTube's various businesses continue to perform well. According to Nielsen data, it has ranked first in streaming viewing time in the U.S. for two consecutive years, with TV becoming the main viewing device. Sports content is watched for over 40 million hours annually, with the NFL set to broadcast the opening game in Brazil in September.

Shorts business is rapidly expanding, with daily views exceeding 200 million, unit time revenue matching traditional videos, even exceeding in some markets. AI-driven new features continue to be launched, including recommendation system optimization, automatic dubbing features, and creator AI toolkits, expanding audience coverage and creation efficiency.

In terms of subscription business, Premium Light service has expanded to 15 new countries, with further market expansion planned.

d. Waymo (Autonomous Driving)

Waymo's autonomous driving business continues to expand, significantly increasing service areas in multiple locations. It has entered operations in Atlanta, doubled the service range in Austin, and expanded by 50% in Los Angeles and the San Francisco Bay Area. Phoenix launched "Teen Accounts" for users aged 14–17, continuously promoting user group expansion. Waymo Driver has completed over 100 million miles of autonomous driving on public roads, conducting tests in more than 10 cities including New York and Philadelphia, with the goal of achieving full passenger service in all cities.

2.1 Performance Situation:

a. Overall Performance:

In Q2 2025, Alphabet achieved consolidated revenue of $96.4 billion, up 14% year-over-year, with constant currency growth of 13%. Core businesses such as search, YouTube ads, subscription platforms and devices, and Google Cloud all achieved double-digit growth, reflecting strong momentum across the company's business scope. Total revenue costs were $39 billion, up 10% year-over-year.

TAC was $14.7 billion, up 10% year-over-year; other revenue costs were $24.3 billion, mainly driven by content acquisition costs. Operating expenses increased by 20% year-over-year, reaching $26.1 billion, mainly driven by the following factors:

Legal-related expenses of $1.4 billion, reflecting the impact of some preliminary settlement agreements;

R&D expenses increased by 16%, due to compensation and depreciation;

Sales and marketing expenses increased by 5%.

Operating profit reached $31.3 billion, up 14% year-over-year, with an operating margin of 32.4%. Net profit increased by 19% year-over-year, to $28.2 billion; earnings per share (EPS) increased by 22% year-over-year, to $2.31. Free cash flow for the quarter was $5.3 billion, with cumulative free cash flow over the past 12 months totaling $66.7 billion. Quarterly free cash flow was affected by seasonal increases in capital expenditures and one-time federal tax payments. Quarter-end cash and marketable securities balance was $95 billion.

b. By Business Segment:

Google Services:

Revenue increased by 12% year-over-year, reaching $82.5 billion, mainly driven by search, YouTube ads, and subscriptions.

Search and other revenue increased by 12% year-over-year, reaching $54.2 billion, with retail and financial services as the main drivers.

YouTube ad revenue increased by 13% year-over-year, reaching $9.8 billion, with direct response ads leading growth.

Network ad revenue decreased by 1% year-over-year, to $7.4 billion. Subscription platforms and device revenue increased by 20% year-over-year, to $11.2 billion, with significant growth in YouTube subscriptions and Google One. Operating profit increased by 11% year-over-year, to $33.1 billion, with a profit margin of 40.1%. The profit margin remained flat compared to the same period last year due to legal expenses.

Google Cloud:

Revenue increased by 32% year-over-year, reaching $13.6 billion. GCP core products and AI services grew much faster than the overall cloud business; Workspace also grew due to increased ARPU per seat and user numbers. Operating profit reached $2.8 billion, with the profit margin increasing from 11.3% to 20.7%, benefiting from revenue growth and expense optimization, but partially offset by higher depreciation costs.

Backlog increased by 18% quarter-over-quarter and 38% year-over-year, reaching $106 billion, reflecting strong demand from new and existing customers.

Other Bets:

Revenue was $373 million, with an operating loss of $1.2 billion. Waymo and other businesses received key resource investments, with significant expansion potential.

c. Capital Expenditure and Shareholder Returns:

Capital expenditure for the second quarter was $22.4 billion, with approximately two-thirds allocated to servers and one-third to data center and network equipment.

Returned $16.1 billion to shareholders, including $13.6 billion for stock repurchases and $2.5 billion for dividend payments.

5. Future Outlook:

a. Revenue and Exchange Rate Impact: The company currently has a good overall growth momentum. The third quarter may benefit from positive spot exchange rate effects, but there are still uncertainties due to volatility.

b. Business Segment Outlook:

Google Services:

The high base in industries such as financial services in 2024 will affect year-over-year data in the second half of 2025; the large number of ads placed during the 2024 U.S. election on YouTube will create year-over-year downward pressure.

Google Cloud:

Products and services continue to be driven by strong demand; despite accelerated deployment, supply-demand tension may continue until 2026.

c. Capital Expenditure (CapEx) Outlook:

Raised 2025 capital expenditure forecast to $85 billion (previously $75 billion) to support Cloud customer demand. The increase mainly comes from additional server investments, delivery time adjustments, and accelerated data center construction. Looking ahead to 2026, CapEx may continue to grow, with specific details to be disclosed in subsequent earnings calls.

d. Changes in Expense Structure:

Significant growth in capital expenditure over the past few years has brought depreciation pressure. Depreciation in the second quarter reached $5 billion, up $1.3 billion year-over-year (+35%); Q3 depreciation growth is expected to further accelerate.

2025 will see an increase in manpower in key investment areas, with total employee numbers expected to rise quarter-over-quarter, and new graduate recruitment will also bring some growth.

Pixel series new products will be launched in August, with related expenses reflected in the third quarter earnings report.

2.2 Q&A

Q: How do you view the impact of changes in consumer behavior on product and platform evolution? How should investors think about this from a traffic and monetization perspective?

A: We are very optimistic about the future. Users are increasingly interested in AI, seamlessly integrating multimodal capabilities (such as Lens and Circle to Search) into interactions, indicating rapid behavioral changes. We focus on building natural experiences and training monetization models on this basis to ensure that AI features (such as AI Overviews and AI Mode) can provide excellent commercialization results without compromising the experience.

Q: How do you view the mixed model of advertising and subscriptions, and what key lessons have you learned as the subscription business continues to expand?

A: YouTube's subscription business (including YouTube TV, YouTube Music, and Premium) is also growing strongly. We adopt a dual strategy—monetizing through ads on one hand and providing more choices through subscriptions on the other—to meet the needs of different users and advertisers.

Q: What is the current situation regarding access to computing resources? Even after adding approximately $10 billion in capital expenditure this year, there is still a tight supply environment. What does this mean for the business?

A: Despite increasing capital expenditure by approximately $10 billion, computing power supply and demand remain tight. Especially in cloud computing, business momentum is strong, and there is a time delay from infrastructure investment to deployment. Therefore, we are simultaneously advancing planning and investment, continuously strengthening the supply side of AI models (including GPUs and TPUs) to support rapidly growing customer demand. In the coming years, as current investments gradually translate into capacity, we are confident in meeting market development needs.

Q: How should future traffic and monetization trends be balanced regarding paid clicks and pricing trends in search growth?

A: We always manage our business with a focus on user outcomes and advertiser ROI, rather than focusing on paid clicks or CPC itself. In the first quarter, paid clicks increased by 4% year-over-year, but multiple factors can affect quarterly changes, such as advertiser spending, product and policy changes, and user engagement. Therefore, we emphasize not overinterpreting a single metric but comprehensively evaluating the balance between monetization efficiency and user value.

Q: What are the main technical obstacles you face in terms of agent-based search and scalable agent capabilities in commercial queries?

A: Our investment in the 2.5 series models, especially the Pro version, focuses on enhancing agent capabilities, which is one of our most valued directions. The current major technical challenges include reliably linking a series of events, ensuring controlled latency and cost, and providing a sustainable and reliable agent experience for users.

Q: Looking ahead to 2026, what internal efficiency sources or friction points are you focusing on?

A: 2026 is expected to be a time when agent experiences achieve more widespread application. In terms of internal applications, we have implemented agent-based coding processes within the company, especially in the software engineering field, receiving positive feedback. Although there are still some friction points initially, these obstacles are gradually being overcome.

Q: How do you view the role of new-generation devices (such as glasses) in expanding AI experiences? Is it possible to enter a stage where mobile phones are no longer the center of user experience?

A: The development of AI is driving the emergence of new hardware, such as glasses and other form factor devices. At the IO conference, we showcased smart glasses developed in collaboration with Warby Parker, with significant improvements in experience compared to previous generations, representing a new wave of innovation driven by AI. Nevertheless, we expect that in the next two to three years, phones will remain the core of user experience.

Q: What are the differences in usage between AI Mode in search and the standalone Gemini app?

A: Regarding AI Mode and the Gemini app, the former is suitable for handling information-oriented queries that require deep AI support, while the latter is more like a proactive, personalized assistant, suitable for complex tasks such as coding and creation. We observe that both interfaces have their specific use cases and user behaviors, so maintaining a dual-interface strategy at this stage helps meet broader user needs.

Q: AI talent mobility is frequent, how do you assess Google's competitiveness in attracting and retaining key AI talent?

A: We have been investing in AI for over a decade, accumulating a world-class talent team with breadth and depth. Top talent values not only compensation but also being at the forefront of technology, working with the best colleagues, and having sufficient computing resources to support their work. We have strong competitiveness in these areas, and our indicators for retaining and attracting talent remain healthy. Although individual personnel changes may attract attention, overall data is stable and positive.

Q: How do the costs of AI manpower and computing resources affect the overall investment structure?

A: In resource allocation, we use a full-stack approach to combine research, manpower, and computing resources, building a strong organizational capability system. These investments increase operating costs but also effectively drive operational efficiency through internal use of AI tools, ensuring disciplined and rewarding resource use. We believe this systematic investment will support us in capturing long-term growth opportunities brought by AI.

Q: As AI Overviews and AI search formats develop, reduced ad impressions may affect click-through rates. How do you view the ability to improve click-through rates in the future?

A: AI Overviews have now expanded to over 100 countries, covering more than 2 billion users, driving increased search satisfaction and user scale. We observe that its monetization rate is roughly equivalent to traditional search, providing a good foundation for next-generation innovative ad formats. Although ad display formats are changing, we will continue to focus on user experience, driving ad product innovation to improve click-through rates and ad effectiveness.

Q: What future expansion might there be in the collaboration between Google Cloud and OpenAI?

A: Regarding collaboration with OpenAI, we are pleased with their choice of Google Cloud as the infrastructure platform. Google Cloud has always been an open platform, long supporting various startups, tech companies, and AI research institutions. We are excited about the current partnership and look forward to expanding the boundaries of collaboration in the future, further providing strong support for the AI developer ecosystem. We will continue to establish stable technical and commercial cooperation foundations in the cloud.

Q: How do you view the advertising environment's trend in the second half of this year?

A: The advertising business continued its strong momentum in the second quarter, with growth mainly coming from verticals such as retail, financial services (especially insurance), and healthcare. However, the third quarter has only just begun, so it is still too early to comment on the trend for the second half of the year.

Q: Google adopts a dual-interface search strategy rather than unified search, do internal indicators show this is the optimal path?

A: Structurally, we adopt a dual-interface strategy to simultaneously meet different user needs for information-oriented and assistant-type tasks. Search mainly serves information-type queries, while the Gemini App provides a more personalized, proactive interaction experience, suitable for complex tasks such as creation and coding. We believe this strategy can better cover the full range of user intentions.

Just as we successfully integrated text, images, and videos into a unified search interface in the past, we also have the ability to integrate the current AI search experience into a more seamless unified framework. We will continue to optimize interface collaboration based on user behavior and feedback, ensuring the final experience is both efficient and natural.

Q: Can you explain the outlook for the cloud business in the second half of the year? Previously mentioned that supply constraints would ease by the end of 2025, but now it is mentioned that constraints will continue until 2026, can you explain this change?

A: We are continuously expanding cloud capacity, including the launch of data centers and servers. Although capacity is increased every quarter, these investments are not "switch-like" immediate effects but require time to gradually translate into available capacity. The accelerated growth in the second quarter reflects the conversion effect of previous investments; and we expect more capacity to be gradually released during the period from the end of 2025 to 2026.

Q: Will agent experiences democratize the internet like search did 20 years ago, or will it lead to traffic being more concentrated among a few winners?

A: Regarding agent experiences, we do believe this will be an important technological advancement, potentially replicating certain features of the early internet, such as expanding access and inspiring more use cases. However, this progress also requires addressing practical issues such as business models, not just a simple technological change. Therefore, it may bring new opportunities but also cause resources to concentrate on more efficient or advanced players. We will strive to promote the development of agent experiences in a healthy ecosystem manner.

Q: How is the performance of Gemini subscriptions within Google One? What measures do you have to accelerate consumer adoption?

A: Gemini subscriptions are making significant progress, especially after the launch of the 2.5 series models, driving rapid migration to AI Pro and Ultra plans. We see very healthy performance in the subscription business this quarter, with Google One becoming an attractive product combination due to the integration of storage and AI services. We expect to continue expanding consumer coverage in the future, further driving paid conversion.

Q: Recent capital expenditure has increased significantly, is the main driver behind this from the cloud? How do you view the return on these investments?

A: Regarding the increase in capital expenditure, it is indeed mainly driven by the development of the cloud business. On the Cloud platform, we see high customer satisfaction, low churn rates, and strong usage stickiness. As the installed base continues to expand, we are confident in improving marginal investment efficiency. These capital expenditures support AI capabilities and cloud services, creating more value for customers, and gradually reflecting in profit margin improvements. Therefore, we believe this is a healthy, long-term investment with good return potential.

<End Here>

Risk Disclosure and Statement of This Article:Dolphin Research Disclaimer and General Disclosure