Google's Gemini 3.5 Pro delay is only a few weeks on the calendar, but it lands at the worst possible moment: just as rivals are hiring away the people Google most wants to keep.
Google can live with a July launch. What it can't shrug off so easily is the timing. Business Insider reported this week that Gemini 3.5 Pro, the frontier model Google previewed at I/O in May, has slipped from a planned June release into July while the company gathers feedback from early testers and keeps tuning the model for longer, more agentic tasks.
That sounds sensible enough. No serious enterprise buyer wants a rushed frontier model, and nobody benefits from a benchmark-friendly system that breaks the moment you put it inside a real workflow. But you don't judge a delay in isolation. You judge it against everything else happening around it.
Right now, that picture is rough for Google.
Business Insider also reported that Noam Shazeer, a vice president of engineering working on Gemini and one of the authors of the 2017 paper "Attention Is All You Need," is leaving Google for OpenAI. That paper introduced the transformer architecture that sits under most modern large language models. John Jumper, the Google DeepMind scientist whose AlphaFold work helped win the 2024 Nobel Prize in Chemistry, has also said he's leaving for Anthropic. The Wall Street Journal reported that Alphabet shares fell 5% on June 22, wiping out $225 billion in market value, after investors digested those departures.
Losing one celebrated researcher is a story for the org chart. Losing Shazeer and Jumper in the same week is a story for the market.
You don't have to pretend Google is suddenly weak to see the problem. This is still Google DeepMind, with enormous compute, distribution through Search and Workspace, and a cloud business that already sits inside the companies every AI lab wants to sell to. Gemini 3.5 Pro could ship in July and be excellent. Google has done that before.
The issue is confidence. Enterprise buyers don't only buy the model in front of them. They buy the pace of the roadmap behind it, the quality of the next six releases, and the belief that the best people in the lab still want to be there.
The delay is manageable, the talent signal is not #
A few weeks of delay, by itself, is ordinary. Business Insider said Google is using feedback from selected users on its Antigravity platform and LMArena, and that the company is folding in lessons from Gemini 3.5 Flash, including concerns around token use. That's exactly the kind of work you want done before a general release, especially if the model is meant to handle long-horizon coding and agent tasks rather than quick chatbot answers.
But here's the thing: OpenAI and Anthropic don't need Google to fail. They only need Google to look slower at the moments when buyers are choosing defaults.
If you're a CIO deciding what model stack to standardize on for the second half of 2026, July is not an abstract date. Procurement calendars are real. Security reviews are real. Developers start building internal habits around whatever tool is available when the budget clears. A model that arrives a month late can still win on quality, but it has to work harder because the conversation has already started without it. That is where the departures hurt. Shazeer is not just another senior name. Google paid $2.7 billion in 2024 to license Character.AI technology and bring him and parts of that team back, as Axios and other outlets reported at the time. When someone like that leaves again less than two years later, people notice. They should.
Jumper's exit cuts differently. AlphaFold was one of Google's cleanest arguments that AI leadership was not only about chatbots, coding assistants, and enterprise subscriptions. It showed that DeepMind could still produce scientific work with public weight. Anthropic hiring him gives it more than talent. It gives it credibility in a field where credibility is now part of the product.
Frankly, the market reaction was not irrational. It may have been blunt, because markets usually are, but the concern underneath it is straightforward. The best frontier labs are competing for chips, customers, distribution, and researchers at the same time. Google has advantages in the first three. It cannot afford to look careless with the fourth.
What you should watch in July #
The next test is not whether Google publishes a polished launch post. It will. The test is whether Gemini 3.5 Pro arrives with enough public evidence to change buyer behavior: real benchmarks, clear pricing, a usable model card, and developer feedback that says the agentic improvements hold up outside Google's own demos.
If Google ships that, the June delay becomes a footnote. If it ships a vague release with limited details, rivals will use the gap. Anthropic has built serious momentum with Claude among developers, especially for coding. OpenAI still has the strongest consumer and platform pull. Google doesn't get to win this round by reminding everyone it invented half the field. You have to keep proving it.
That is the real story behind Gemini 3.5 Pro's July delay. The calendar slipped a few weeks, but the pressure is coming from somewhere deeper: whether Google can keep frontier AI moving fast enough while the people who helped define the frontier decide their next work belongs somewhere else.
Also read: GitHub's record June shows usage-based pricing is the unlock enterprise AI has been waiting for • Elizabeth Warren wants Congress to treat AI algorithms the same way it treats any other antitrust threat • SignalFire data shows engineering hiring is more resilient than almost anyone expected