Every few weeks, someone new declares that Google has fallen behind in AI. A recent opinion piece by Pedro Dias, a search industry veteran with 20 years in the game, takes a more interesting position than either extreme: Google will be fine, but it will not lead in AI, because it cannot afford to. The search business still prints money, and no company voluntarily breaks the machine that pays for everything.
This is one analyst’s opinion, not a confirmed forecast. But the evidence behind it deserves a serious look, because if he is right, it changes how marketers should think about where search traffic is heading.
The Argument in Plain Terms
Start with the inventory, because it is genuinely impressive. Google crawls the web, runs Chrome and Android, and owns YouTube. Its cloud order backlog stood at 514 billion dollars at the end of Q2. The Gemini app crossed a billion monthly users in August, and Google says its AI Mode in Search has passed a billion users too. Comscore data shows Gemini’s share of US AI assistant prompts climbing from 17 percent in January to 30 percent in June, while ChatGPT slid from 70 percent to 50 percent.
Dias accepts all of that. His point is that an inventory tells you what a company can do, not what it will choose to do. When he looks at where Google puts its scarcest resources, he sees a company acting like Microsoft: rich, everywhere, and slow to move against the business that pays for everything. Microsoft had the cash and the distribution too, yet Bing still holds about 4.5 percent of worldwide search after moving first with OpenAI technology in 2023. The pattern is a company that is very good at protecting a lead and very slow at disrupting itself.
What the Evidence Actually Shows
The timeline of Google’s flagship model is the centerpiece. In February, Gemini 3.1 Pro topped the Artificial Analysis Intelligence Index, beating Claude Opus 4.6 by four points while costing less than half as much to run. At I/O in May, Google promised Gemini 3.5 Pro for June. It never arrived. In mid-July, Bloomberg reported that the model’s coding ability was short of where Google wanted it. Since then, Google has shipped Gemini 3.7 Flash, 3.8 Flash, speech, video, music, and robotics models, but the API changelog, last updated September 22, still shows no 3.5 Pro.
Then there is the compute question. In AI, compute is the scarce resource. In May, The Information reported that Anthropic had committed to spend 200 billion dollars over five years on Google’s cloud and chips. A Reuters exclusive in August reported that tight capacity, plus internal disputes between the teams running different parts of Gemini, left areas like coding short of resources. A TPU rented to Anthropic is revenue this quarter. A TPU spent on another training run is a bet.
The enterprise numbers tell a similar story. Menlo Ventures’ 2025 enterprise report put Anthropic at 40 percent of enterprise LLM API spend versus Google’s 21 percent, and 54 percent versus 11 percent in coding. Menlo backs Anthropic, so discount those figures, but the money teams spend on getting real work done is still flowing mostly elsewhere.
And then there are the people. Noam Shazeer, co-lead of Gemini, left for OpenAI in June. Jeff Dean left after 27 years to co-found Discovery Loop. Demis Hassabis stepped back to chair DeepMind. Fortune reports that DeepMind’s hires per departure fell from roughly 12 in 2023 to roughly two now, while Anthropic managed 22 in 2025. Researchers who want to build what comes next go where what comes next is the whole business.
Why This Matters for Your Search Traffic
When Google put generative AI into Search, it put it inside the results page: AI Overviews on top, AI Mode beside it, and ads moving into both. Search and other revenue grew 17 percent in Q2 to 63.3 billion dollars, and Pichai credits the AI features with driving query growth. A defense that pays that well is the most comfortable position a company can be in. Comfort, in Dias’s reading, is the problem.
What he believes actually threatens Search is the assistant that does the job and never shows a results page at all. It writes the code, compares the contracts, books the trip, and nobody clicks an ad along the way. That is the category where Google is furthest behind, and Dias does not think that is a coincidence. Building it properly means building the thing that makes Search optional, and every incentive inside Google points the other way.
To be fair, Google can sprint when it is frightened. ChatGPT triggered a code red inside Google in December 2022, and the recovery was real. But Google sprints when there is a fire. Without one, it moves at Microsoft speed.
What Marketers Should Do About It
First, treat AI Overviews and AI Mode as the permanent shape of Google’s results page, not an experiment that will be rolled back. Google has wrapped its most valuable surface in AI and is monetizing it with ads. Optimize for being cited in those answers, because the citation is becoming the click.
Second, diversify your traffic sources. If the assistant that skips the results page is the real long term threat, and Google is structurally slow to build it, then the surfaces that grow fastest over the next few years may not be Google’s at all. Build presence on the platforms that already drive AI citations: community discussions, podcasts, video, and reference content.
Third, track the two signals Dias says would change his mind: a Gemini product that does a job end to end with no results page and no ads, pushed as hard as AI Mode; or Google openly letting such a product take queries from Search and celebrating it on an earnings call. If either happens, you will want to be early.
Key Takeaways
- One veteran’s read: Google has every resource to lead in AI, but its incentives point toward protecting Search revenue, not disrupting it.
- The evidence cited includes the delayed Gemini 3.5 Pro, compute going to cloud customers instead of training runs, weak enterprise API spend share, and senior researcher departures.
- Google’s AI strategy keeps users on the results page with AI Overviews and AI Mode, which grew Search revenue 17 percent in Q2.
- The real long term threat to search traffic is the assistant that completes tasks without ever showing a results page, and Google is least motivated to build it.
FAQ
Is Google actually losing the AI race?
No. The Gemini app passed a billion monthly users, and its share of AI assistant prompts nearly doubled in the first half of the year. The argument is narrower: Google is not leading in assistants that work without a search results page, which is where the future may be.
Why would Google hold back its best AI?
The claim is about incentives, not ability. Every disruptive AI assistant would take queries away from Search, a 63.3 billion dollar business last quarter. No leadership team volunteers to shrink its most profitable division.
What does this mean for SEO?
AI Overviews and AI Mode are the permanent shape of the results page. The practical move is to optimize for being cited inside AI answers, since the citation is increasingly your only visibility.
Should I stop investing in Google-focused marketing?
No. Search still works and still grows. Just stop assuming Google will remain the only discovery surface that matters, and build presence on the platforms that feed AI citations today.
What would prove this argument wrong?
A Google product that completes tasks end to end with no results page and no ads, marketed as aggressively as AI Mode; or Google celebrating that product taking queries from Search on an earnings call.