---
title: "Anthropic siphons off Google's top talent! After another breach in the AI talent defense line, Google's ambitions in AI face market skepticism"
type: "News"
locale: "en"
url: "https://longbridge.com/en/news/290758655.md"
description: "Two top AI researchers at Google, Jonas Adler and Alexander Pritzel, plan to move to competitor Anthropic. This move intensifies the recent talent exodus at Google, which includes Nobel Prize winner John Jumper, who previously joined Anthropic, and Noam Shazeer, who moved to OpenAI. The departure of these key personnel may weaken Google's leading position in the AI field, raising market concerns about Google's ability to execute its AI strategy and its competitive strength"
datetime: "2026-06-25T01:27:02.000Z"
locales:
  - [zh-CN](https://longbridge.com/zh-CN/news/290758655.md)
  - [en](https://longbridge.com/en/news/290758655.md)
  - [zh-HK](https://longbridge.com/zh-HK/news/290758655.md)
---

# Anthropic siphons off Google's top talent! After another breach in the AI talent defense line, Google's ambitions in AI face market skepticism

According to reports from media citing informed sources, two top artificial intelligence researchers from the American tech giant Google (GOOGL.US) plan to leave the company to join competitor Anthropic PBC. This adds another case to a series of high-profile departures from Google recently, which could weaken the search giant's leading position in the AI application field and also suggests that Google's AI blueprint may face significant resistance at this stage.

Informed sources stated that Jonas Adler and Alexander Pritzel are both seen internally at Google as key contributors to the company's exclusive AI super application ecosystem—Gemini AI large model ecosystem. They will be moving to the developer of Claude. According to media reports, these sources requested anonymity as the relevant information has not been made public. Adler was primarily responsible for and deeply involved in the company's AI programming-related developer work, while Pritzel participated in the entire training process of the artificial intelligence system.

Google is undoubtedly an early pioneer in the field of artificial intelligence, but for most of the current AI boom, it has been playing catch-up to the two leading AI application forces, OpenAI and Anthropic, until the end of last year when it found its leadership rhythm with more powerful AI large models and chips. However, in recent days, the company has already lost two well-known employees: Nobel Prize winner John Jumper has moved to Anthropic, and star researcher Noam Shazeer has joined OpenAI. Their departures have shocked long-term investors in Google and raised further doubts about the company's actual competitive ability in the fierce race to build superior AI large models and AI intelligent agent products.

As of Wednesday's market close, Alphabet, Google's parent company, saw a slight decline in stock price, dropping as much as 1.2% during the trading session.

This round of changes among senior AI developers does indicate that Google's AI blueprint is facing some stage resistance, but it does not equate to a complete failure of Google's AI strategy, nor does it mean a slowdown in the global AI monetization process. The recent frequent movement of these top AI professionals shows that the AI commercialization process is beginning to shift from the first stage of "model release and user growth" to the second stage of "who can turn the model into high-frequency workflows, corporate budgets, cloud revenue, and vertical industry productivity." The talent siphoning by Anthropic highlights that the capital market and institutional investors are rewarding AI-native companies that are more focused, carry less historical baggage, and can more easily embed AI large models/AI intelligent agents like Claude into corporate workflows.

**Google's AI talent defense line faces another impact: Anthropic siphons Gemini brain**

These latest departures underscore that Google is facing pressure from two soon-to-be-public startups; these companies offer a significant opportunity for employees of large tech firms, even those with substantial income, to join before an IPO and gain rare and substantial returns. In at least one case, the departure of a Google employee seems to have occurred after a shift in priorities regarding how to allocate valuable computing resources, a matter that has prompted other employees to leave the company entirely According to two informed sources, shortly before Shazeer announced plans to join OpenAI, AI computing infrastructure resources originally allocated to him for an important project were reallocated to a key team at Google DeepMind in London. The sources stated that this move was intended to strengthen collaboration between teams and streamline Google's budget process for pre-training; pre-training is the initial phase of AI development, during which models learn from vast datasets.

Adler, Pritzel, Jumper, and Shazeer did not respond to media or insider requests for comments. Anthropic also declined to comment. A Google spokesperson stated that the company remains confident in its position in the AI talent market and referenced remarks made earlier this week by Google DeepMind CEO Demis Hassabis.

"There is a significant amount of talent movement among all leading AI labs, and we have indeed won a considerable share of top talent. We have the largest and most diverse pool of research talent among all labs to date," Hassabis said at an event in Cannes. "Right now, this is an extremely competitive market, the most competitive market in the history of the tech industry."

Shazeer's career trajectory reflects the fierce talent war defining the landscape of AI applications and monetization. After co-authoring a groundbreaking paper that helped catalyze the AI boom, he left Google in 2021 to found the chatbot startup Character.AI, and then returned to Google in 2024 through an unusual licensing deal that valued his company at $2.5 billion.

Upon returning to Google, Shazeer co-led the development process of the company's flagship AI large model—namely, the Gemini AI model-related super application ecosystem. According to two informed sources, he had also been researching a new AI infrastructure before his departure. One of the sources stated that the infrastructure is still based on Transformer technology; this technology was proposed by Shazeer and his colleagues in 2017 and has since become one of the core methodologies recognized by AI developers.

According to current and former Google employees, Shazeer is both admired and controversial within Google. Two informed sources stated that his comments on transgender identity and the Gaza conflict sparked controversy among some employees.

Meanwhile, Jumper has become a new face of Google's most ambitious AI efforts after winning a Nobel Prize for milestone research in using AI to predict protein folding. Adler and Pritzel will join Anthropic alongside Jumper, with whom they collaborated on cutting-edge research.

In recent months, key members of Jumper's protein folding research team have left Google DeepMind. According to one informed source, some of them have moved to Isomorphic Labs; this is a company spun off from Alphabet that focuses on AI drug design Anthropic competes with Google while also collaborating with it, actively poaching talent from the tech giant. According to a 2025 industry analysis by venture capital firm SignalFire, the likelihood of engineers from Google's AI business studio DeepMind leaving to join Anthropic is nearly 11 times that of the reverse flow.

Like Google, the developer of the Claude AI application ecosystem is also exploring AI applications in life sciences and healthcare to expand the use of its technology. Anthropic recently completed a new round of financing with a valuation of $96.5 billion, surpassing OpenAI, and is considering going public as early as this fall.

In the UK, where DeepMind's leadership is located, AI researchers are often bound by long-term non-compete agreements, which are enforceable under UK law. According to an informed source, Jumper may not start working at Anthropic until next year.

Another researcher, Arthur Conmy, wrote on X on Wednesday that he will join Anthropic to work on AI safety-related developer tasks. According to his LinkedIn profile, during his time at DeepMind, Conmy was a senior research engineer involved in the Gemini 2.5 model and AI programming work.

**Google's AI blueprint faces obstacles, but the global AI commercialization process is not slowing down**

Jonas Adler and Alexander Pritzel are seen as key contributors to the Gemini model, involved in AI programming and model training, respectively; coupled with John Jumper's transition to Anthropic and Noam Shazeer's move to OpenAI, it indicates that Google is experiencing significant "talent outflow" pressure in the areas of scarce cutting-edge model talent, AI programming, life sciences AI, and foundational AI training research architecture.

Especially with Anthropic's upcoming IPO and its valuation skyrocketing to $96.5 billion, it offers top AI researchers a more attractive choice between "the stability of large company cash flow and the flexibility of startup equity," which could weaken Google's speed, team continuity, and internal morale in certain cutting-edge directions.

However, from a commercialization perspective, this seems more like the growing pains of the AI race entering a four-dimensional competition of "talent, computing power, productization, and organizational coordination," rather than a slowdown in global AI monetization. Alphabet's official figures for the first quarter show that Google Search and other advertising revenues grew by 19%, Google Cloud revenue grew by 63%, surpassing $20 billion for the first time, and backlog orders nearly doubled to over $460 billion. The number of paid monthly active users for Gemini Enterprise grew by 40% quarter-over-quarter; these data indicate that Google's AI commercialization is still advancing in search, cloud, enterprise AI, and subscription sectors. The real issue is not "AI cannot be monetized," but rather that Google must perform more complex resource scheduling between protecting its search advertising cash cow, advancing Gemini model capabilities, allocating scarce computing power, retaining top talent, and accelerating enterprise implementation The commercialization of AI has not slowed down; instead, it has transitioned from the first stage of "model release and user growth" to the second stage of "who can turn models into high-frequency workflows, enterprise budgets, cloud revenue, and vertical industry productivity." The talent siphoning at Anthropic highlights that the capital market is rewarding AI-native companies that are more focused, have lighter historical burdens, and can more easily embed AI large models/agents like Claude into enterprise workflows. Google, on the other hand, has stronger distribution channels, cloud infrastructure, TPU chips, and search cash flow, but also bears more complex organizational inertia and pressure to rebalance its advertising model.

In other words, Google's AI ambitions have not been completely disproven; rather, they have entered a more challenging "execution realization phase": if the GeminiAI application ecosystem, AI search, AI programming, enterprise agent-based workflow AI agents, and cloud computing power can continue to translate into revenue and profit, talent loss will only be a valuation fluctuation item; if there is a continuous outflow of core researchers, lagging model iterations, and erosion of search monetization, the market will reprice it as a "defensive giant in the AI era."

### Related Stocks

- [GOOG.US](https://longbridge.com/en/quote/GOOG.US.md)
- [GOOGL.US](https://longbridge.com/en/quote/GOOGL.US.md)
- [GGLS.US](https://longbridge.com/en/quote/GGLS.US.md)
- [ANTH.NA](https://longbridge.com/en/quote/ANTH.NA.md)
- [GOOW.US](https://longbridge.com/en/quote/GOOW.US.md)
- [GGLL.US](https://longbridge.com/en/quote/GGLL.US.md)
- [GOOY.US](https://longbridge.com/en/quote/GOOY.US.md)
- [GOU.US](https://longbridge.com/en/quote/GOU.US.md)
- [GOOX.US](https://longbridge.com/en/quote/GOOX.US.md)
- [GOOP.US](https://longbridge.com/en/quote/GOOP.US.md)
- [MAGX.US](https://longbridge.com/en/quote/MAGX.US.md)
- [OpenAI.NA](https://longbridge.com/en/quote/OpenAI.NA.md)

## Related News & Research

- [Jefferies Says Google's AI Talent Losses Are Just 'Noise,' Here's Why](https://longbridge.com/en/news/290600282.md)
- [Meta Faces Growing Pressure From Trump Administration To Submit Advanced AI Models For Federal Testing: Report](https://longbridge.com/en/news/290638185.md)
- [Google Calendar finally has more color options for events](https://longbridge.com/en/news/290216781.md)
- [Sunbelt Securities Inc. Trims Stake in Alphabet Inc. $GOOGL](https://longbridge.com/en/news/290322605.md)
- [Anthropic launches Claude Tag, turning Slack into an AI-staffed workspace](https://longbridge.com/en/news/290730289.md)