Google (Minutes): We are the only full-stack AI cloud platform

The following are the FY25 Q3 earnings call minutes for $Alphabet(GOOGL.US) compiled by Dolphin Research. For earnings interpretation, please refer to "Google's Dramatic Turnaround: From AI 'Victim' to AI 'Trendsetter'"

I. Review of Core Financial Data

1.1 Consolidated Performance

1. Revenue: $102.3 billion, up 16% year-over-year (15% growth at constant currency)

2. Total Revenue Cost: $41.4 billion, up 13%.

3. Technology Cost: $14.9 billion, up 8%.

4. Other Revenue Costs: $26.5 billion, up 16%, mainly due to increased YouTube content procurement costs, followed by depreciation and other technical infrastructure operating costs.

5. Total Operating Expenses: $29.7 billion, up 28%.

(1) R&D expenses increased by 22%, mainly driven by AI-related compensation and depreciation expenses.

(2) Sales and marketing expenses remained flat.

(3) Administrative expenses increased significantly, mainly due to a $3.5 billion fine related to the European Commission.

6. Operating Profit: $31.2 billion, up 9%; operating margin 30.5%.

Excluding the impact of the EU fine, operating profit increased by 22%, operating margin 33.9%.

Operating margin benefited from strong revenue growth and continued cost efficiency improvements, but was offset by significant increases in legal and depreciation expenses.

7. Other Income and Expenses: $12.8 billion, mainly from unrealized gains on non-marketable equity securities portfolio.

8. Net Profit: $35 billion, up 33%; earnings per share $2.87, up 35%.

9. Free Cash Flow: $24.5 billion for the quarter, $73.6 billion over the past 12 months.

This quarter's free cash flow benefited from strong operating cash flow and recent tax law changes (involving R&D expense accruals and asset depreciation timing adjustments), but was partially offset by increased capital expenditures.

10. Cash and Marketable Securities Balance: $98.5 billion; Capital Expenditure: $24 billion, with approximately 60% invested in servers and 40% in data centers and network equipment; Shareholder Capital Return: $11.5 billion in stock repurchases, $2.5 billion in dividend payments.

1.2 Segment Performance

1. Google Services: Revenue $87 billion, up 14% year-over-year.

(1) Google Search and Other Advertising Revenue: $56.6 billion, up 15%, driven by all major verticals, with retail and financial services contributing the most.

(2) YouTube Advertising Revenue: $10.3 billion, up 15%, mainly from direct response ads, followed by brand ads.

(3) Network Advertising Revenue: $7.4 billion, down 3% year-over-year.

(4) Subscription, Platform, and Device Revenue: $12.9 billion, up 21%, mainly due to YouTube and Google One subscription services.

(5) Operating Profit: $33.5 billion, up 9%; operating margin 38.5%, impacted by the European Commission fine.

2. Google Cloud: Revenue $15.2 billion, up 34% year-over-year, driven by strong GCP performance.

GCP growth mainly from enterprise AI products, with quarterly revenue reaching billions; Workspace business achieved double-digit growth, benefiting from increased per capita revenue and user seat numbers.

Operating Income: $3.6 billion, up 85%; operating margin 23.7%, up from 17.1% in the same period last year, mainly due to strong revenue performance and continued cost efficiency optimization, but partially offset by increased technical infrastructure usage costs (including depreciation and other operating costs such as energy costs).

Order Backlog: $155 billion, up 46% quarter-over-quarter, up 82% year-over-year, growth mainly driven by strong demand for enterprise AI.

3. Other Businesses: Revenue $344 million; operating loss $1.4 billion.

II. Detailed Information from the Earnings Call

2.1 Key Information from Executive Statements

Strategy and Business Progress

1. Full-Stack AI Strategy: Encompasses world-class AI infrastructure research (including models and tools) and products and platforms that bring AI to global users.AI is driving tangible business results across the company, helping quarterly revenue surpass $100 billion for the first time.

2. AI Infrastructure: Large-scale deployment of advanced chips in data centers, including NVIDIA GPUs and self-developed TPUs; the only supplier in the industry offering both types of chips. TPU capacity continues to expand to meet demand, with Anthropic planning to access up to 1 million TPUs.

3. Models and Toolchains:Gemini 2.5 Pro, Veo 3, and Nano Banana models are globally leading; Veo 3 has generated over 230 million videos, with more than 13 million developers creating based on generative models. In quantum computing, Willow quantum chip runs algorithms 13,000 times faster than the world's top supercomputers, with verifiable results.

4. Revenue Diversification and Differentiation: 13 product lines each with annualized revenue exceeding $1 billion; creating highly differentiated products through proprietary technology to enhance operating margins. Deep product differentiation begins with AI infrastructure, with a decade of AI accelerator development experience bringing the most comprehensive chip product matrix in the industry.

5. Customers and Partners: AI leadership has won clients such as HCA Healthcare, LGAI Research, and Macquarie Bank; nine of the world's top ten AI labs choose Google Cloud. Google Cloud is the only cloud service provider offering self-developed leading generative AI models (including Gemini, IMAGINE, Veo, Chirp, and Lyria), with third-quarter product revenue based on generative AI models growing over 200% year-over-year.

6. YouTube: Ranked first in U.S. household living room streaming viewing time for two consecutive years; NFL event live broadcasts set a record for the highest concurrent viewers on YouTube live (over 19 million fans).Short video per-play revenue in the U.S. market surpasses traditional video ads.AI-powered creator tools accelerate content creation and expand business territory.

7. Waymo:Plans to launch services in London next year and advance Tokyo market layout; expanding to Dallas, Nashville, Denver, and Seattle, obtaining fully autonomous driving operation licenses for San Jose and San Francisco airports. Enterprise Waymo service allows companies to include Waymo in employee commuting options, with usage steadily increasing.

8. Search and Advertising: AI is driving transformative expansion in search usage, with new experiences like AI Overview and AI Mode driving overall search volume growth (including commercial queries), creating more monetization opportunities. AI MAX and search functions are launched globally, serving hundreds of thousands of advertisers, becoming the fastest-growing AI search advertising product.

9. Cloud Business Outlook: Enterprise demand for infrastructure (including TPUs and GPUs), enterprise AI solutions driven by models like Gemini 2.5, core GCP infrastructure, as well as network security and data analysis services, remains strong. Billion-dollar contracts signed in the first nine months of 2025 for cloud business have exceeded the total of the past two years.

2.2 Financial Outlook and Guidance

1. Fourth Quarter Outlook

Revenue: May benefit from favorable exchange rates, but exchange rate fluctuations may affect actual results.

Google Services Advertising Business: Year-over-year growth rate will be negatively impacted, due to a surge in advertising spending during the U.S. election in Q4 2024 (especially on YouTube).

Cloud Business: Demand remains strong, with product order volume increasing by $49 billion quarter-over-quarter.

Supply and Demand Situation: Despite efforts to increase capacity, accelerate server deployment, and data center construction, supply and demand tensions are expected to persist in Q4 2024 and 2026.

Investment and Costs: Capital expenditures for 2025 are expected to be in the range of $91 billion to $93 billion (up from the previous $85 billion), with cash payment timing differences potentially causing data fluctuations.

Capital expenditures in 2026 will increase significantly, with details to be disclosed in the Q4 earnings call.

On the cost side: Depreciation cost pressure continues, with Q3 depreciation expenses at $5.6 billion (up 41%), expected to accelerate slightly in Q4; increased investment in technical infrastructure will impact the income statement through depreciation and data center operating costs (such as energy expenses).

Sales and marketing expenses are expected to be concentrated at the end of the year for product launches and holiday promotion seasons.

Overall momentum: AI products are widely adopted, driven by team innovation and execution, bringing strong growth to search, YouTube ads, subscription platforms, and cloud device businesses; the company is making every effort to increase capacity, but supply and demand tensions are expected to persist in Q4 2024 and 2026.

2.3 Q&A

Q: There is widespread discussion on Wall Street that the monetization rate of autonomous e-commerce may be lower than that of search business. Therefore, I would like to ask: In the transition to a more autonomous world, what factors are you most focused on to ensure a smooth transition for search business and advertising clients?

Regarding Waymo. How far are we from deeply integrating Waymo into the core Gemini platform functions? Can user data (such as travel destinations, hotel stays, airport arrivals, etc.) be integrated into the Waymo system? Can users pre-book Waymo services through their profiles? How far are we from achieving these features? What steps do we need to take?

A: Regarding the first question, it's all in the early stages, but we believe the Agent experience is actually a complement to how people access information. It helps us answer users' questions, helps users complete tasks, and in the process, helps businesses. We are developing multiple Agent experiences in several key verticals (such as travel, commercial shopping, etc.).

We are very focused on creating a seamless user experience, while also focusing on integrating different partner ecosystems in a way that creates value for our partners. By the way, we are also working closely with many partners to improve their own agent experiences through cloud services.

Perhaps we can talk more about shopping, where we are actively using AI to improve the shopping experience. As you know, we provide a more intuitive conversational shopping experience on AI Mode.When building the future of agent-based shopping, we believe it must be a future that benefits both users and merchants. We have also launched new agent-based checkout features that allow shoppers to purchase products from merchant websites using agent AI, etc. We have partnered with PayPal to help merchants build agent-based commercial experiences. We also have new open protocols for agent-to-agent transactions, and so on.

Regarding the second Waymo question, it's a great question. I have actually been thinking about the exact same topic. I plan to meet with the team in a few weeks to discuss and review this work. It's an exciting time. Waymo is clearly scaling up, especially in 2026. I think the possibility you mentioned, of combining Gemini (especially its multimodal experience) with services like YouTube, indeed presents a huge opportunity to significantly improve the in-car experience. This definitely excites us, and you will certainly see new experiences in 2026.

Q: When we focus on AI Overview and AI Mode, we know query volume growth is accelerating, but can you help us understand how click-through rates, conversion rates, and pricing might change in these AI-driven search formats?

Additionally, can you talk about what optimization opportunities you see in the core cost structure? This will provide a buffer for rapid infrastructure expansion and depreciation costs in the future.

A: In the third quarter, search and other income grew across all major verticals, with retail and financial services being the main drivers, and healthcare also contributing to growth.New AI experiences, such as AI Overview and AI Mode, continue to drive overall query volume (including commercial queries) growth, creating more monetization opportunities.

AI Overview has covered over 2 billion users, and we are continuing to expand ads in the overview to more countries on desktop and mobile devices. For AI Overview, even under current advertising benchmarks, its monetization rate is roughly the same. In the long run, richer experiences in AI Overview and AI mode will bring more advertising slot opportunities.

In terms of clicks and pricing, we are committed to creating good results for users and providing attractive ROI for advertisers, rather than simply managing cost-per-click. As you will see in the 10-Q filing, total clicks increased by 7% year-over-year, and cost-per-click increased by 7% year-over-year.

Regarding optimization opportunities in the core cost base, this is an ongoing effort, not a one-time effort. The key is that increased productivity allows us to invest more in business growth and continuously improve the income statement. Specific areas include: controlling the growth rate of headcount, optimizing real estate footprint, and ensuring optimization of construction and overall infrastructure when investing in technical infrastructure. For example, our self-built data centers are optimized to ensure maximum efficiency. Additionally, using AI to enhance Google's productivity is also a focus, for example, nearly half of the code is currently generated by AI. We always ensure that the services or products provided can achieve corresponding economic returns and value, such as Shorts' revenue sharing being lower than streaming, which helps improve gross margins. Although depreciation expenses face pressure with increased capital expenditures, we are making company-wide efforts to operate the business in the most disciplined and efficient manner while maintaining investment in future growth.

Q: When you think about the organization's efforts in custom chips, can you talk about your view on the opportunities each generation of chips brings? This involves both improving operational efficiency within the organization and potential revenue-generating initiatives around these chips outside the organization.

Second question. We have seen YouTube advertising revenue data from the earnings report, can you talk about the expansion of YouTube's subscription business? How do these two parts of the business together build an interesting framework that shows YouTube's monetization model increasingly integrating both advertising and subscription forms?

A: We have seen huge demand for our AI infrastructure products (including solutions based on TPUs and GPUs), which has been one of the key drivers of our growth over the past year. Looking ahead, we expect this strong demand to continue, and we are investing to meet it.

Google Cloud's differentiation advantage lies in our full-stack AI approach: we are the only major cloud service provider building products based on our own models and offering highly differentiated technology. This provides us with opportunities to continuously drive growth in cloud business operating margins, while the infrastructure part will also become a revenue growth driver in the future.

Regarding YouTube, we typically describe its business as a flywheel: it all starts with creators, and we invest heavily to ensure YouTube is the platform of choice for creators. In terms of monetization, YouTube adopts a dual-engine strategy, combining advertising business and growing subscription services. This quarter, both advertising and subscriptions achieved strong growth. YouTube Music and Premium users create more value for creators, music media partners, and YouTube itself than ad-supported users. On average, a YouTube Music and Premium subscriber generates significantly higher gross profit than a purely ad-supported user. Interaction from global users generates revenue through advertising and subscriptions, feeding back to creators, driving more viewing and interaction, forming a flywheel effect. Our focus is to sustain this growth cycle, and we are satisfied with this dual-engine monetization strategy.

Q: With the widespread adoption of Gemini, AI Mode, and AI Overview features among user groups, have these features created notable differences in user behavior and engagement depth? Is this difference reflected across the entire Google ecosystem?

How do you observe the adoption of AI Overview mode? With the increase in commercial query volume and total query volume, how do you view the evolution of the economic model of the search business? How do these changes compare to the incremental cost of providing search results?

A: AI Overview is a natural part of the Google experience, so engagement is very high. AI Mode has a different user group, with some being casual users trying it out, but there is a core group that truly enjoys and is passionate about AI mode. Early adopters resonate strongly with the product and actively seek to use it.

Gemini also has a group of active users seeking the product. Overall, in various use cases, we do see a group of early adopters, followed by more people joining, and users using it more over time, reporting high user satisfaction. The underlying product metrics are quite encouraging.

Regarding AI Overview, even under current advertising benchmarks, whether above, below, or within AI responses, overall, we see the monetization rate roughly the same. This is a good baseline for further innovation. We are excited about this. In AI mode, we are testing and will continue to test and learn before further expansion. Combined with the overall development of commercial queries we mentioned, we believe we are currently in a favorable position. It can be considered that there is a potential opportunity in historically poorly monetized queries, and we can imagine building on this through smart AI integration.

Q: It is clear that when users use AI mode, the query length is significantly longer. Can you talk about how this increase in length might affect your ability to enhance asset returns? And what benefits might you observe from longer query lengths in the early stages? Can you talk about how you internally evaluate the return on investment capital? And what early signs have you observed that give you confidence that current spending is driving long-term return improvements?

A:AI Mode now has over 75 million daily active users in the U.S., with strong week-over-week growth in usage since launch, and query volume doubling this quarter.I mentioned that we are testing ads in AI Mode and will continue testing before further expansion. It's too early to provide any details on the tests.

Regarding ROIC and how we view the overall business and what early signs are encouraging, first, I have to say it's not just early signs, because we have already seen returns in the cloud business, generating billions in revenue from AI this quarter.But overall, we have a rigorous framework and approach to evaluating these long-term investments, aimed at doing two things: ensuring the company builds resilient growth and meeting customer needs in the near to medium term.

So we look at the entire business, evaluating the potential returns of each investment, whether in the cloud business (which is more visible because you see revenue generation and we currently cannot meet customer demand, demand exceeds supply), in the advertising business, you see us investing in transforming search, as Philipp and Sundar mentioned AI Overview and AI mode, investing in recommendation systems in YouTube. So when we make long-term investment decisions, we go through a very rigorous process to evaluate what the returns might be and how long it will take to see returns, giving us high confidence to invest. So it's a very rigorous approach.

Q:Historically, about 20% of Google search queries have been commercial in nature. Can you talk about how new products like AI MAX might increase the proportion of commercial queries from a monetization perspective?

A: AI MAX enhances advertisers' ability to target a broader range of queries. On the other hand, whether query volume will increase due to AI mode is another question, and Sundar has talked about the opportunities he sees in this regard. I think it's important to distinguish between these two points.In the long run, there is an opportunity to expand those queries that are not fully commercialized but may have adjacent commercial relationships into more attractive advertising products, while also creating genuinely interesting user experiences. Additionally, from a more macro perspective, AI Overview and AI mode are significantly improving the search experience, as reflected in all metrics such as user satisfaction, user quality, etc., and its nature is universal, applicable to the universality of human needs. Therefore, we see the breadth of its impact, and over time, this will naturally also apply to commercial categories.

Q: First, with the emergence of new models like Gemini, AI Overview, AI Mode, the addressable market for search and search-like behaviors is expanding. What makes you believe this will also expand the TAM of marketing activities and related overall revenue? Secondly, regarding AI Mode, AI Overview, and traditional Google Search, how do you see their coexistence in the next 12 to 24 months? Is it user choice of mode, or algorithm-driven? Can you talk about your outlook for this process in the future?

A: It is currently a dynamic moment, and we are providing services based on user needs. Search is continuously evolving, and with AI Overview and AI Mode, we can offer diverse experiences to users. Over time, you will see us simplify these experiences, just as we integrated unified search many years ago. Similar evolution will occur in the future, but we will closely monitor to ensure we meet user needs.

Gemini enables us to create a more personalized, proactive, and powerful AI assistant for users. Currently, offering both search and Gemini interfaces allows us to meet a wide range of user needs. We will carefully look for opportunities to continuously optimize the user experience. Regarding the first question, overall, as we have always emphasized, this is a moment of market expansion, and we see increased user engagement. When users engage more deeply, some of the information in these user journeys naturally has commercial attributes.Therefore, I expect that over time, this will also drive the growth of commercial opportunities.

Q: Regarding the innovation speed of frontier models (such as the upcoming Gemini-III), how is it currently? Is it still rapidly innovating or showing signs of slowing down? Meanwhile, you mentioned that many large contracts have been signed in the cloud business over the past 9 months. Has the long-term profitability economics of these deals changed? What should we pay attention to?

A: Regarding the speed of frontier model development, two things are simultaneously true: I am extremely impressed with the speed of execution by the team and the speed at which we are improving these models, but at the same time, every aspect we are trying to improve is indeed becoming increasingly powerful. So I think the pace is accelerating, but sometimes we take time to launch a significantly improved model, which may take a little longer. But I think the fundamental development speed is amazing, and I am excited about the release of Gemini 3.0 later this year. In the cloud business, I want to point out a sign of momentum: the number of contracts over $1 billion signed in the first three quarters of this year exceeds the total of the past two years. So we are undoubtedly seeing strong momentum, and we are executing quickly. Regarding long-term economics, I would say that as a full-stack AI participant, and with us developing highly differentiated products on our own technology, I think this will help us maintain a good development trajectory, as you have seen over the past few years.

<End Here>

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