> ## Content Index > Fetch the complete content index at: https://barbarosozturk.com/llms.txt > Use this file to discover other available public pages before exploring further. # Google Won't Win The AI Assistant Race, And It Might Not Need To - URL: https://barbarosozturk.com/blog/google-wont-win-the-ai-assistant-race-and-it-might-not-need-to/ - Published: 2026-11-13T16:56:59.000Z - Updated: 2026-11-13T16:56:59.000Z - Description: Google's search growth is slowing even as queries hit a five-year high. I don't think it wins the assistant race, but Waymo, robotics and TPUs sold to rival labs could make it the company that profits most from everyone else's AI. - Author: Barbaros - Tags: Artificial Intelligence, Economy & Markets, Google, Robotics Google's third-quarter results told two very different stories. Search, the business that built the company, barely grew from the previous quarter. Google Cloud more than doubled. I think that split shows where Google's future lies, and it isn't in winning the AI assistant race. ![](https://storage.ghost.io/c/64/81/648165b7-979e-43dd-9c43-02101159f467/content/images/2026/11/image-105.png) ### Search Is Slowing, Not Collapsing I'd be careful with the word "drop" here. Search and other revenue was [$63.35 billion in Q3](https://ppc.land/google-search-sales-gain-just-79m-on-q2-as-query-growth-hits-5-year-high/?ref=barbarosozturk.com), still up 12% from a year earlier. But that was only $79 million more than the previous quarter, and growth slowed from 17% in Q2. The strange part is that usage is rising. Sundar Pichai said Google's query growth rate is "the highest it's been in the last five years." More searches are producing less growth, and AI Mode queries [earn less from ads](https://www.neoteo.com/en/alphabets-q3-2026-results-search-and-cloud-grow-free-cash-flow-is-negative?ref=barbarosozturk.com) than classic searches. Pichai said Google is focused on "building the best Search user experience, followed by monetization," without a timeline for ads in AI Mode. As someone who buys Google Ads every week, I read that as an admission that the answer engine doesn't yet pay like the link engine did. I've argued that [chatbots won't break Google's ad pricing power](https://barbarosozturk.com/blog/the-pricing-power-of-google-ads-why-chatbots-will-not-break-the-monopoly/) any time soon. I still believe that, but slower growth from a smaller share of a bigger pie is how a dominant business starts to fade. ![](https://storage.ghost.io/c/64/81/648165b7-979e-43dd-9c43-02101159f467/content/images/2026/11/image-106.png) ### Why Google Won't Win The Assistant Race Gemini is huge by any normal standard. Alphabet said it [reached 1 billion users](https://finance.yahoo.com/markets/stocks/articles/alphabet-googl-q3-2026-earnings-054403755.html?ref=barbarosozturk.com), and Google has Android, Chrome and Gmail to push it. Reach isn't the problem, trust is. On the same earnings call, Philipp Schindler described an agent offering a pricier insurance product with a special discount, and said Google is "confident" it can build ads into agentic experiences. I wrote that [OpenAI and Anthropic will win the personal agent war](https://barbarosozturk.com/blog/why-openai-and-anthropic-will-win-the-personal-agent-war/) for exactly that reason. When an assistant can spend your money, people will pay for one that answers only to them. Google will have a huge free assistant, but I don't think it'll own the most valuable users. The Apple deal is the main counterpoint. Apple's foundation models were built with Google's Gemini models, so Google could power Siri without owning the customer. That's a good business, but it's a supplier role, which is the pattern I see across the rest of this post. ![](https://storage.ghost.io/c/64/81/648165b7-979e-43dd-9c43-02101159f467/content/images/2026/11/image-107.png) ### Waymo Is The Physical AI Lead Waymo is the strongest proof that Google can monetize AI in the physical world. It was doing about [400,000 paid rides a week](https://www.claimsjournal.com/news/national/2026/02/12/335668.htm?ref=barbarosozturk.com) in February and [500,000 by late March](https://insideevs.com/news/791245/waymo-rides-per-week-2026-doubled/?ref=barbarosozturk.com), across 10 markets. Its target is more than 1 million weekly rides by the end of 2026. Waymo raised $16 billion at a $126 billion valuation earlier this year, with Alphabet putting in about $13 billion. Robotaxis are slow to scale, because every car is a physical asset that has to be built, mapped and insured. That's also why the lead is hard to copy. I've argued that [Waymo could become an ad platform](https://barbarosozturk.com/blog/why-waymo-is-googles-next-advertising-monopoly-the-ad-supported-commute/) as rides get cheaper. That fits this thesis better than ads in an assistant, because a screen in a car doesn't control what you buy. ![](https://storage.ghost.io/c/64/81/648165b7-979e-43dd-9c43-02101159f467/content/images/2026/11/image-108.png) ### Robots Need A Brain Supplier Robotics is earlier, but Google is positioning DeepMind as the brain other companies build on. At CES in January, Boston Dynamics and Google DeepMind [announced a partnership](https://www.therobotreport.com/boston-dynamics-google-reunite-on-next-gen-atlas-humanoid/?ref=barbarosozturk.com) to use Gemini Robotics models in the new Atlas humanoid. All 2026 Atlas units are committed to Hyundai's training centre and to DeepMind itself. Hyundai aims to manufacture up to 30,000 humanoids a year by 2028. No financial terms were disclosed, so I can't say how Google gets paid yet. Still, the model is familiar. Android gave Google a place in every phone it didn't build, and Gemini Robotics could do the same for robots. I've written about [Google X's moonshot culture](https://barbarosozturk.com/blog/decoding-the-google-x-moonshot-factory/), and this is one of the few moonshots that now has a clear customer. ![](https://storage.ghost.io/c/64/81/648165b7-979e-43dd-9c43-02101159f467/content/images/2026/11/image-109.png) ### TPUs Are The Quiet Giant The biggest shift is in chips. For years, Google only rented out TPU capacity in its own data centers. In April, Alphabet said it would [sell TPUs to select customers](https://finance.yahoo.com/markets/stocks/article/google-to-sell-tpu-chips-to-select-customers-in-latest-shot-at-nvidia-214900221.html?ref=barbarosozturk.com) to install in their own facilities, which Pichai said would "expand our addressable market opportunity." The customers include its rivals. Anthropic signed a multiple-gigawatt TPU agreement in April, with chips coming online from 2027. Meta [signed a multi-year deal](https://winbuzzer.com/2026/03/03/meta-signs-multibillion-dollar-deal-rent-google-tpus-xcxwbn/?ref=barbarosozturk.com) worth billions to rent TPUs and is reportedly in talks to buy them outright. The results are already visible. Google Cloud revenue rose [103% to $30.7 billion](https://www.neoteo.com/en/alphabets-q3-2026-results-search-and-cloud-grow-free-cash-flow-is-negative?ref=barbarosozturk.com) in Q3, with a backlog of $561 billion. That's the logic of selling shovels in a gold rush. Google doesn't need to win the assistant race if every contender is paying it for compute. ![](https://storage.ghost.io/c/64/81/648165b7-979e-43dd-9c43-02101159f467/content/images/2026/11/image-110.png) ### The Risks This strategy is expensive. Alphabet expects $195 billion to $205 billion in capital spending this year and said it will "increase significantly" in 2027. Free cash flow was negative $9.1 billion in Q3, partly because it built up TPU inventory before delivery. Nvidia remains the default choice for most AI developers, and it argues its chips are more flexible. Selling to rivals also creates a strange position. Google is helping fund Anthropic and Meta's AI, the same companies competing with Gemini. And Waymo still has to prove it can scale profitably to dozens of cities, with Tesla and others chasing it. ![](https://storage.ghost.io/c/64/81/648165b7-979e-43dd-9c43-02101159f467/content/images/2026/11/image-111.png) ### Conclusion: Google As The Supplier I expect Google's search growth to keep slowing as more questions get answered without a click. I also expect it to lose the most valuable assistant users to companies whose only customer is the user. But Google is building three businesses that don't depend on winning that race: robotaxis, robot brains and AI chips. Each one lets it profit from AI no matter who owns the chat window. **My Take:** Google is turning from the front door of the internet into the infrastructure behind everyone else's AI, and that could prove to be the better business. ## Sources Pages I opened for the facts in this post: - [PPC Land: Google search sales gain just $79M in Q3](https://ppc.land/google-search-sales-gain-just-79m-on-q2-as-query-growth-hits-5-year-high/?ref=barbarosozturk.com) - [Neoteo: Alphabet's Q3 2026 results](https://www.neoteo.com/en/alphabets-q3-2026-results-search-and-cloud-grow-free-cash-flow-is-negative?ref=barbarosozturk.com) - [Alphabet Q3 2026 earnings call transcript](https://finance.yahoo.com/markets/stocks/articles/alphabet-googl-q3-2026-earnings-054403755.html?ref=barbarosozturk.com) - [Claims Journal: Waymo's path to 1 million weekly trips](https://www.claimsjournal.com/news/national/2026/02/12/335668.htm?ref=barbarosozturk.com) - [InsideEVs: Waymo doubles weekly ridership](https://insideevs.com/news/791245/waymo-rides-per-week-2026-doubled/?ref=barbarosozturk.com) - [The Robot Report: Boston Dynamics and Google reunite on Atlas](https://www.therobotreport.com/boston-dynamics-google-reunite-on-next-gen-atlas-humanoid/?ref=barbarosozturk.com) - [Yahoo Finance: Google to sell TPU chips to select customers](https://finance.yahoo.com/markets/stocks/article/google-to-sell-tpu-chips-to-select-customers-in-latest-shot-at-nvidia-214900221.html?ref=barbarosozturk.com) - [WinBuzzer: Meta's TPU deal with Google](https://winbuzzer.com/2026/03/03/meta-signs-multibillion-dollar-deal-rent-google-tpus-xcxwbn/?ref=barbarosozturk.com)