NotebookLM Is Now Gemini Notebook. The Rename Is Cosmetic; the Cloud Computer That Writes and Runs Code Is Not.

TL;DR

On July 16, Google announced that NotebookLM is now Gemini Notebook, folding its most-liked AI product (over 30 million people and more than 600,000 organizations, per Google's own numbers) into the Gemini brand. The rename is the headline, but the substance is underneath: every notebook is getting a secure cloud computer that can natively write and execute code for data analysis grounded in your sources. That is live now for AI Ultra subscribers and Workspace customers on the top AI tiers, and rolling out to all Pro users on the web over the coming weeks. Notebooks already sync with the Gemini app, and Google says they are headed into AI Mode in Search next. A citation machine just became a compute surface.

Three Names in Three Years

The product debuted as Project Tailwind at I/O 2023, shipped as NotebookLM, became the sleeper hit of Google's AI portfolio, and is now Gemini Notebook, per 9to5Google. The logo picks up the blue and purple Gemini gradient, and Google says it "remains a standalone product focused on being your premier research tool" while doing more across the wider ecosystem. Google's naming department remains its highest-velocity engineering team.

The scale numbers explain why Google wants this thing wearing the flagship badge: 30 million plus individual users and 600,000 plus organizations is real distribution for a research tool that started life as an I/O side project.

The Feature Hiding Under the Rebrand

Buried in paragraph four of the announcement is the part that changes what the product is. Google is rolling out an update that gives every notebook "a secure cloud computer," letting Gemini Notebook "write and execute code natively, helping you conduct complex data analysis grounded in your sources," with new output formats to match.

Until now, NotebookLM's whole trick was disciplined reading: it answered from your uploaded sources with citations, and that grounding was the reason people trusted it. The ceiling was that it could only talk about your documents. Now it can compute over them. It is the difference between a librarian who summarizes your files and one who takes them into a back office with a spreadsheet and comes back with the pivot table. The summarizer paraphrases your quarterly CSVs; the analyst runs the regression on them.

Per 9to5Google, the underlying model upgrade (Gemini 3.5 plus Google's Antigravity agent stack) started reaching AI Ultra users last month. What is new this week is the productization and the rollout schedule:

Google has not published specs, quotas, or supported runtimes for the sandbox, so treat "cloud computer" as a capability description rather than a documented environment. What is documented is the direction: code execution grounded in a private source corpus, inside a consumer product with 30 million users.

One Corpus, Three Surfaces

The second half of the announcement is distribution. You can already open and create notebooks directly inside the Gemini app, with full cross-app syncing between the app and the standalone product. And Google says notebooks will "soon" land in AI Mode in Search, meaning the research corpus you built in one tool becomes context for your searches.

Read that flow as a data-gravity play. Every PDF, transcript, and CSV you feed a notebook makes the next Google surface smarter about you, and makes leaving the ecosystem more expensive. SiliconANGLE frames the rebrand as an ecosystem and accessibility move, which is polite for the same observation.

It also mirrors what the competition did last week: OpenAI folded its standalone Codex app and Atlas browser into ChatGPT, and now Google is pulling its most successful standalone AI product into the Gemini umbrella. The era of the charming one-off AI app inside a big lab appears to be closing; everything gets absorbed into the flagship assistant.

Why Builders Should Care

First, sandboxed compute per user is now table stakes. ChatGPT has shipped sandboxed code execution since 2023, Claude runs code against your uploads, and now the pattern lands in a mainstream consumer research app. If your product summarizes documents but cannot compute over them, the bar just moved in front of 30 million users.

Second, grounding plus execution is a genuinely useful combination, and it is the one to copy. NotebookLM's original insight was that answers pinned to user-supplied sources beat open-ended chat for trust. Adding a code sandbox to that loop, where the code itself works on the grounded corpus, is a stronger primitive than either half alone. Self-hosters have the pieces today: an open-weight model, a document store, and a local sandbox runner get you the same loop without shipping your research corpus to Mountain View.

Third, watch the Search integration. Notebooks inside AI Mode means private corpora start shaping search answers. That is a new grounding channel for anyone thinking about how their content, or their users' content, surfaces in an AI-mediated Google.

Caveats

• The cloud computer is gated to paid tiers. The announcement lists AI Ultra and top Workspace tiers now, all Pro users on the web in "the coming weeks," and says nothing about the free tier.
• "Coming weeks" and "soon" are Google-soft dates. The AI Mode in Search integration has no timeline at all.
• No specs, quotas, supported languages, or security documentation for the sandbox have been published yet.
• Nothing about the underlying models changed in this announcement; the Gemini 3.5 upgrade reached the product earlier, per 9to5Google.
• The 30 million user and 600,000 organization figures are Google's own, without an independent audit or a definition of "user."

Key Takeaways

• Google renamed NotebookLM to Gemini Notebook on July 16, with a new Gemini-gradient logo; it stays a standalone research product.
• The real feature: a secure cloud computer per notebook that writes and executes code natively for data analysis grounded in your sources.
• Rollout is live for AI Ultra and Workspace AI Ultra/Expanded customers, hits all Pro users on the web over the coming weeks, and is unannounced for the free tier.
• Notebooks already sync with the Gemini app, and Google says they are coming to AI Mode in Search, turning your research corpus into cross-surface context.
• For builders: grounded RAG plus sandboxed code execution is now the baseline pattern in consumer AI research tools, and it is replicable with open models and a local sandbox.

Sources: Google (official announcement), 9to5Google, SiliconANGLE