Google delays next Gemini launch amid coding stumbles; NotebookLM becomes Gemini Notebook

Google's next Gemini release has slipped after falling short of internal benchmarks, according to reports that describe coding-capability stumbles, clashing internal teams and frustrated engineers. Chief AI Architect Koray Kavukcuoglu is now unifying Google's fragmented internal AI coding tooling — an implicit admission that organizational friction, not just model quality, is part of the problem.

MarketWatch reported Alphabet shares fell on the news, with the framing that Google is 'struggling to keep up' as rivals ship agentic coding models — pointed timing given xAI's Grok 4.5 coding launch and Kimi K3's code-arena dominance the same week. The delay lands Google in an awkward narrative position after positioning Gemini as its comeback vehicle.

Alongside the delay, Google pushed forward on product branding and integration: NotebookLM was rebranded to Gemini Notebook with Google Drive sync, and Gemini 3.5 plus 'Antigravity' upgrades are coming to the AI Pro tier. Sundar Pichai also touted Intel using Gemini Enterprise to speed semiconductor development, and Logan Kilpatrick rolled out managed-agent cost controls, a free tier and scheduled triggers in the Gemini API — signs Google is shipping incremental agent tooling even as the flagship slips.

The competitive context is stark: the entire week's story is momentum shifting toward open-weight Chinese models and fast-iterating challengers, and a high-profile Google delay feeds the 'is Google falling behind?' narrative that DeepMind spent 2025 trying to bury. What to watch: whether Kavukcuoglu's tooling unification actually accelerates the next Gemini, and how much of the delay reflects a genuine capability gap versus internal coordination problems. Google DeepMind's separate Weather Lab update was a rare on-schedule bright spot.