Google will not ship Gemini 3.5 Pro in June. Business Insider reported last week that the model — previewed at Google I/O in May when CEO Sundar Pichai said it would launch “next month” — has slipped to July as the company gathers feedback from early testers and addresses token-consumption issues from its Flash 3.5 generation. That news landed the same week four high-profile AI researchers announced departures to OpenAI and Anthropic, triggering an intraday drop of up to 7.2% in Alphabet shares on June 22 — the stock’s worst session in over a year. For enterprise operators planning Gemini-based workflows in the second half of 2026, this week is a signal to reassess timeline dependencies.

Key takeaways:

• Gemini 3.5 Pro slipped from June to July; July is guidance, not a contractual commitment
• Noam Shazeer (co-author of “Attention Is All You Need”) departed for OpenAI; Nobel laureate John Jumper departed for Anthropic
• Jonas Adler and Alexander Pritzel, key contributors to Gemini model development, also joined Anthropic
• Google is incorporating feedback on token efficiency and long-horizon agentic tasks before the 3.5 Pro release
• Existing Gemini 2.5 Pro and Flash services are unaffected by the delay

What Does the Delay Tell Operators?

The slippage itself is modest — one month. What matters is what it signals. Business Insider reported that Google’s team is gathering feedback from users on its Antigravity platform and LMArena, and is folding in lessons from Gemini 3.5 Flash, including criticism that the model consumed tokens too aggressively. Gemini 3.5 Pro is being designed for long-horizon tasks and agentic workflows — precisely the workloads where unchecked token consumption becomes a budget problem. Choosing to fix this before shipping is defensible. The risk is that it costs Google a procurement window at the moment OpenAI and Anthropic have capable models available on existing enterprise contracts.

Should the Talent Departures Change Your Vendor Assessment?

Noam Shazeer co-authored the 2017 “Attention Is All You Need” paper that introduced the transformer architecture underlying every major large language model today. Google brought him back in 2024 via an effective $2.7 billion acqui-hire of Character.AI specifically to work on Gemini. John Jumper won the 2024 Nobel Prize in Chemistry alongside DeepMind CEO Demis Hassabis for AlphaFold. Both announced departures within days of each other, followed immediately by Adler and Pritzel.

A cluster of departures does not mean Google’s AI capability is finished. The company still has Workspace and Vertex AI distribution that rivals cannot match, dedicated enterprise agent infrastructure, and vast compute. But the question for procurement teams is not “is Google done?” — it is “does this change my roadmap confidence over the next two to three quarters?”

The honest answer is: it introduces uncertainty. Enterprise AI decisions increasingly depend not just on what a model does today, but on confidence that the next six releases will maintain pace. That confidence is harder to hold when the people who built the current generation are building at competitors.

What to Do Now

• Do not treat July as a hard date. If your team has scoped a summer deliverable around Gemini 3.5 Pro capabilities, flag it now and identify a fallback — Claude or GPT-5.x cover the same agentic use cases today.
• Check your Vertex AI or Workspace AI contract window. Current 2.5 Pro and Flash services are not affected. Identify when your next renewal opens — that is the right moment to ask Google for roadmap commitments in writing.
• Evaluate alternatives for coding and long-horizon agent tasks. Run your actual use cases against available models, not just benchmarks. The delay creates a natural window to test.
• Watch whether July holds. If Google ships 3.5 Pro in July with strong agentic performance, this week becomes a footnote. If the date slips again without explanation, that is a materially different signal.

FAQ

Does the Gemini 3.5 Pro delay affect my existing Google Workspace AI or Vertex AI service? No. Gemini 2.5 Pro and Flash continue to run normally. Gemini 3.5 Pro is a future upgrade, not a forced migration. If your Q2 roadmap included milestones tied to 3.5 Pro-specific capabilities, those should move to Q3.

Should I switch from Gemini to Claude or GPT-5 based on the talent news? Not based on this alone. Four researcher departures and a one-month delay are a watch signal, not a migration trigger. Google retains deep Workspace integration, Vertex AI infrastructure, and a large engineering team. The right move is to assess whether your current Gemini workflows are actually underperforming — if not, keep watching rather than initiating a disruptive migration.