Table of Contents
Other reviews from this season include:
a. Key Points
- Strong growth across Cloud and Search.
- Raised annual CapEx guidance.
- Will sell its custom AI chips to customers outside of Google Cloud.
b. Demand
- Beat revenue estimate by 2.6%.
- Constant currency (CC) revenue growth was 19% Y/Y.
- Beat Beyond revenue estimate by 8.8%.
- Missed YouTube revenue estimate by 1%.
- Beat search revenue estimate by 2.2%.
c. Profits
- Beat EBIT estimate by 9.7%.
- EBIT rose by 30% Y/Y compared to 20% growth last year.
- R&D rose by 25% Y/Y compared to 14% growth last year.
- Traffic acquisition costs (TAC) rose by 11% Y/Y compared to 3% growth last year.
- Missed FCF estimate by 13%.
- CapEx was $35.7B for the quarter vs. $17.2B Y/Y.
- Beat $2.67 EPS estimate by $0.09 ex-equity gains.
d. Balance Sheet
- $126.8B in cash, equivalents.
- $107B in non-marketable securities.
- $46.5B in debt.
- Added $31.1B in senior unsecured notes during the first quarter.
- Diluted share count fell by 0.4%.
- Dividends rose by 5% Y/Y.
e. Guidance & Valuation
Alphabet now expects $185B in 2026 CapEx vs. $180B previously. And notably, it expects 2027 capital expenditures to meaningfully grow vs. this very lofty 2026 number. That is excellent news for AI infrastructure bulls, as it's a great data point for hinting at this super cycle rocking well into 2027. That was notable to me. Furthermore, just like for Amazon, this spend is tied to direct and concretely visible revenue opportunities. As we'll explore later in this piece, the two-year backlog for Google Cloud is nothing short of humongous, and hefty capital expenditures are needed to service it. Alphabet is adamant that this spend will come not only with this large revenue, but also with a return on invested capital (ROIC) that will make shareholders happy.
f. Call & Release
Search:
Breaking News: The Search King is still the Search King. 19% revenue growth is largely a byproduct of the fantastic work Google has done to lace AI throughout the native search and new AI experiences. This innovation is resonating deeply, enhancing user frequency and driving incremental impressions to monetize. AI Mode is killing it and AI Overviews are killing it while upgrading both with Gemini 3 has already cut average output costs by 30%+. Live Search is killing it. Everything is killing it.
In a perpetual effort to always be improving, Alphabet continues to cut output latency despite adding all the new options for consumers to enjoy. Since 2021, that metric has improved by 35%, providing another tailwind to overall engagement.
And while things look great for this already gigantic part of the business, you could argue AI search monetization remains early. They’re still really only ramping ad load within AI mode and haven’t even begun to branch out beyond subscriptions in the Gemini app. That happening is inevitable and provides another leg of future expansion to look forward to. So far, products like Direct Offers (merchant-funded promotions within AI mode outputs) are garnering positive feedback while Alphabet is now working on another retailer-centric product to plug their catalogs into Gemini product recommendations. Speaking of retailers, Alphabet Universal Commerce Protocol (UCP; open and standardized rail for agentic work) continues to add key partners, with Meta, Amazon, Salesforce, Stripe and more joining this quarter. Ulta Beauty, Sephora and Macy’s are adding new shopping experiences through UCP, while Target and Wayfair signed large cloud deals because of tools like these that amplify utility.
There are also long runways across several other parts of the AI search-related monetization journey. Ad quality will keep improving as the mega-cap introduces better models as Alphabet equips buyers with tools like AI Max. This more deeply understands user intent and pairs that interest with relevant advertisers. It optimizes existing campaigns like those run through Performance Max (PMAX). So far, it’s helping Hilton 15x clicks per dollar and Etsy notch 10% higher search volume. Additionally, agentic transformation will likely enable new shopping experiences to create more actionable purchasing opportunities within search use cases. This year will be a year of building product market fit, ensuring an excellent experience and not yet prioritizing monetization. After that, a boatload of commerce use cases within agentic search should begin to ramp their financial contribution more meaningfully.
- Google sees an opportunity to raise the historically 20% of impressions that include ads.
More on Google Cloud:
This segment is obviously booming and could have posted even higher growth without capacity constraints. New customer additions doubled Y/Y while existing customers are spending 45% more than expected, showing the actual value behind these products. $100M and $1B cloud deals doubled Y/Y and operating income tripled thanks to fantastic fixed cost leverage stemming from crazy good demand. This is despite 44% overall Y/Y growth in infrastructure-related depreciation.
The backlog doubled sequentially (not a typo) to reach $462B, with $231B expected to be recognized within the next two years. If we split it in half for the sake of simplicity, that means ~$115B in already contracted annual business over the next two years for a company that just crossed an $80B annualized run rate. Again… that’s just already contracted business. There will surely be more deals signed. Where is this massive demand coming from? A combination of what it views as a superior full stack AI offering and immensely strong sector tailwinds. Just like for Amazon, it must be noted that Anthropic and OpenAI are responsible for a lot of this fantastic backlog growth, but the figure is still obviously impressive. It just also comes with exposure to this current super-cycle.
Enterprise AI Opportunity:
The enterprise ramp is progressing nicely. While the base is still small, that’s quickly changing. Active users rose by 40% Q/Q, revenue rose by 800% Y/Y and partner-sourced seats rose 9x Y/Y. All in all, this momentum meant Enterprise AI Solutions became the “primary growth driver for cloud for the first time.” That’s important. A big source of growth for these businesses right now is coming from purchasing (or designing and having TSM manufacture) compute and renting it out to customers to run their cloud workloads. That is naturally tied to rapid CapEx growth and historically amazing demand for compute workloads continuing to rapidly grow. AI services like Gemini, on the other hand, should be more structural in nature like enterprise software has been for the last few decades. It is not nearly as tied to the need to catch up or infrastructure footprint requirements or availability shortages that drive pricing power.
Want to read 40 other reviews with this level of detail on popular public companies every single season (& so much more)? You can with a subscription that costs less than $1 per day. Sign up below.
As outlined in our coverage of Google’s Cloud next event, the company launched its Gemini Enterprise Agent Platform and combined different data products under BigQuery and a new Agentic Data Cloud. It also unveiled a slew of new cybersecurity tools with the Wiz deal now closed. If you’d like to read more about those launches, our coverage can be found in section 4 of this article.
AI Infrastructure:
The rumors have been true. Alphabet is gearing up to provide its TPUs to other clouds in the coming months. The revenue contribution from this will be small in 2026 and ramp meaningfully in 2027. Recognition of revenue from this segment will be lumpy, and based on delivery timing. That will likely make it hard to model, and could lead to random upside vs. consensus here and there for the cloud segment. In terms of how this impacted the massive jump in backlog, agreements are included in the figure, but the “majority” of the growth was related to Google Cloud agreements.
This news is exciting. For years, one of the key limitations for TPUs and custom processors in general was thought to be that they only work in a specific cloud for a specific company. That is no longer true. Alphabet’s chips, purpose-built for specialized machine learning workloads, appear to be in high demand across other cloud vendors. That puts a dent in the “GPUs are better than custom processors because they’re more general and malleable” argument. Still, Alphabet knows its customers want choice, and also knows that Nvidia GPUs are world-class products that are great for certain use cases within its cloud. There’s a reason why Nvidia garners a sky-high EBIT margin. It’s elite technology and in-demand from many of Alphabet’s customers. That won’t change.
- As discussed at Cloud Next, the company’s new 8th-gen training TPUs offer 3x processing power and 2x performance vs. Ironwood (7th-gen). Its inference TPUs boost performance there by 80% as well. This is especially noticeable considering Ironwood was already considered superior for certain AI workload efficiency compared to GPUs. This could build on that lead.
YouTube:
YouTube has now led Nielsen U.S. streaming market share rankings for 3 straight years. To maintain and build on this lead, they’re adding Gemini models to YouTube content discovery, content creation tools and advertising campaigns. While these models are expensive to build, their value can be applied to countless products in valuable ways. And with Alphabet enjoying distribution of 7 products to billions of people, they have more ways to leverage these investments than most.
While ad revenue for YouTube continues to grow more slowly than other segments, they’re not concerned by this. It’s related to shifting focus towards premium subscriptions and cannibalizing some of the growth here as a result.
- Demand from small advertisers was highlighted as a strength this quarter. Good to hear for the overall economy.
- Supergoop enjoyed a 55% brand lift with YouTube’s new popular podcast menu.
- Crossed 10M channels publishing YouTube shorts per day.
AI Models & Token Revenue.
Gemini models are processing 16B tokens per minute compared to 10B Q/Q. They now have 330 customers processing 1T+ tokens on a trailing 12-month basis and 35 at 10T+ For context, Gemini 3 Flash charges $0.50 per million input tokens and $3 per million output tokens. If we assume an average cost of $1 per million, this is about $1M in spend per 1T+ token customer or $680M in trailing 12-month token revenue from all 365 customers in these two cohorts combined. That’s less than 1% of annualized Google Cloud revenue, showing how small this piece of the business still remains. At the same time, the ramp is going extremely well and showing clear signs of soon becoming much more material.
- Nano Banana 2 crossed 1B images generated in just over half the time it took Nano Banana 1 to get there. Never thought I’d be typing “nano banana” in an earnings review.
- Gemma 4 (its open model) is up to 50M downloads a few weeks into launch.
Other Bets Reshuffle:
Alphabet continues to shed assets under this part of the business to free up more capital to invest into Waymo. It deconsolidated both Verily from the balance sheet following an external capital raise and GFiber after it combined with Astound Broadband.
- Waymo is in 11 U.S. cities and crossed 500,000 rides vs. 250,000 a little less than a year ago.
- Wing (drone delivery) added Walmart and DoorDash as retail partners. It also announced its intent to start operating in the Bay Area.
g. Take
Fantastic quarter. The search business is rocking, and obviously proving that AI applications are both incremental and easily monetizable. The slower YouTube advertising growth is by design, as Google favors premium subscriptions over that lower-margin part of the business. The cloud business is performing at a historically impressive clip, and hasn't even begun to enjoy the incremental ramp that external TPU sales to hungry customers will provide. There is so much to be excited by, which is why spending is so elevated. And regardless of that, EBIT margin continues to expand, while leadership remains highly confident in long-term free cash flow generation, being more than compelling enough to justify these investments. Signs are beginning to emerge that they're right, and I think those signs will become more abundant over the coming quarters. It’s easy to be confident in the streaming, search, full stack AI, autonomous vehicle and web browsing leader. It’s easy to be confident in Alphabet.
I remain a bullish shareholder.
Want to read 40 other reviews with this level of detail on popular public companies every single season (& so much more)? You can with a subscription that costs less than $1 per day. Sign up below.