According to senior reporter Vivi Lin’s post on X on June 20, Google Gemini co-head Noam Shazeer left to join OpenAI; DeepMind vice president John Jumper also left to join Anthropic. Character.AI co-founder Daniel De Freitas has also entered this round of talent reshuffling.
Three Departures From Google Touch the Most Important Mainline of Modern AI
According to Reuters:
Noam Shazeer (leaves Google to join OpenAI): Co-author of《Attention Is All You Need》, one of the founders of the Transformer architecture
John Jumper (leaves Google DeepMind to join Anthropic): Co-creator of AlphaFold, the 2024 Nobel Prize in Chemistry winner; AlphaFold has already predicted more than 200 million protein structure predictions
Daniel De Freitas (Character.AI co-founder, an important figure in conversational AI): In 2024, Google brought him and parts of its Character.AI team back in a deal worth about $2.7 billion (Axios report). Now it is once again linked to the question of “can Google retain talent?”
Analyst @soul_ai points out that these three people represent three of the most important technology mainlines in modern AI: Transformer, conversational AI, and AlphaFold (AI for Science).
Google Unveils 13 “Billion-Level” Products, Gemini Has 900 Million Monthly Active Users
Google’s published business metrics provide a framework for assessing its competitive strength:
Google I/O 2026: Google has 13 products with ten thousand-level monthly active users in the “billion-level” range, with 5 exceeding 3 billion users; Gemini app monthly active users exceed 900 million
Alphabet Q1 2026: Google Cloud operating revenue of $6.6 billion, operating profit margin of 32.9%
@soul_ai’s analytical framework is: Google is not just a “model company.” It is an end-to-end AI company—owning both infrastructure (TPUs, data centers), multiple models (Gemini, Gemma, AlphaFold), products (Search, YouTube, Android, Gmail), and the distribution power of billions of existing users.
Google Invests Up to $40 Billion in Anthropic; Anthropic Promises $200 Billion to Google Cloud Over Five Years
According to Reuters and related Google Cloud reports, Google’s strategic position includes a counterintuitive dimension: Alphabet has promised to invest up to $40 billion in Anthropic (including $10 billion in cash; the valuation is reported as $350 billion; another $30 billion is tied to performance). Anthropic, according to reports, has promised to spend $200 billion on Google Cloud over five years, planning to use up to 1 million Google TPUs. OpenAI has also already reached a compute-power collaboration with Google Cloud.
@soul_ai’s conclusion is: Google is becoming a supplier of the underlying infrastructure for competitors, “selling shovels” in the AI gold rush.
Frequently Asked Questions
Does the departure of Noam Shazeer and John Jumper represent a fundamental decline in Google AI research capability?
Based on analyst @soul_ai’s framework (personal analysis, not an official assessment), this is a classic Silicon Valley talent war ahead of an IPO, not a signal of a decline in Google AI research capability. OpenAI and Anthropic attract these talents—by itself, that shows Google is one of the deepest AI talent pools in the world. At the same time, achievements like AlphaFold come from teams rather than individuals, and DeepMind’s research culture remains even after individual talent leaves.
What does Alphabet’s $4 billion investment in Anthropic mean?
According to Reuters and Google Cloud reports, Alphabet has promised to invest up to $40 billion in Anthropic. Anthropic also promises to spend $200 billion on Google Cloud over five years and plans to use up to 1 million Google TPUs. This means one of Google’s main AI competitors—and also an important potential infrastructure customer for Google Cloud.
Which Google businesses are most directly affected by these three departures?
Based on the three people’s role backgrounds: Noam Shazeer is involved in Gemini’s co-head responsibilities; his departure directly affects Gemini development. John Jumper is involved in Google DeepMind’s AI for Science direction; his departure directly impacts the continuation of AlphaFold’s subsequent research. Daniel De Freitas’s conversational AI research involves consumer-grade product directions related to character AI. Google has not yet publicly responded on succession arrangements for the related roles.
