AnswersPublished28 sources
Google DeepMind’s AI Talent Advantage Has Collapsed Across Europe
Google DeepMind’s share of research and advanced engineering hires across Europe, the Middle East and Africa fell from 49% in 2022–23 to 18.6% in 2025–26. The shift reflects more than pay: DeepMind is under pressure to commercialize Gemini, while researchers may perceive less autonomy than the lab once offered.
Published byEdited with GPT-5.6 LunaImages generated with GPT Image 1.5
Research answer
Key takeaways
- Google DeepMind’s share of research and advanced engineering hires across Europe, the Middle East and Africa fell from 49% in 2022–23 to 18.6% in 2025–26.
- The shift reflects more than pay: DeepMind is under pressure to commercialize Gemini, while researchers may perceive less autonomy than the lab once offered.
- DeepMind retains major advantages in compute, infrastructure, scientific AI and research depth—but its former position as Europe’s default destination for elite AI talent is no longer secure.
AI Prompt
openai.comCreate a landscape editorial hero image for this Studio Global article: How has Google DeepMind’s position as Europe’s dominant employer of elite AI researchers changed from 2022–23 to 2025–26, including the fall. Article summary: Google DeepMind has shifted from Europe’s clear first-choice elite-AI employer to one major competitor among several. It remains a formidable research organization with exceptional infrastructure, but its regional hiring. Topic tags: general, general web, user generated, news. Style: premium digital editorial illustration, source-backed research mood, clean composition, high detail, modern web publication hero. Use reference image context only for broad subject, composition, and topical grounding; do not copy the exact image. Avoid: logos, brand marks, copyrighted characters, real person likenesses, fake screenshots, UI text, readable text, watermarks, charts w
Google DeepMind has not become an unimportant AI lab. It remains one of the industry’s largest and best-resourced research organizations. But its role in the talent market has changed dramatically: the lab has gone from Europe’s clear first-choice employer for elite AI researchers to one powerful competitor among several.
The clearest evidence is its shrinking share of new research and advanced-engineering hires across Europe, the Middle East and Africa. Zeki Data estimates that DeepMind’s share fell from 49% in 2022–23 to 18.6% in 2025–26—the sharpest regional decline Zeki recorded for a major AI lab. 2
The numbers show a sharp loss of hiring momentum
The change is not simply that DeepMind is losing more employees than it hires. The lab is reportedly still adding research and advanced-engineering staff overall. The problem is that its advantage has narrowed considerably.
According to Zeki data reported by Fortune, DeepMind’s ratio of arrivals to departures fell from roughly 12:1 in the second quarter of 2023 to about 2:1 in the third quarter of 2026. 2 A ratio of 2:1 still indicates a net inflow, but it no longer describes an organization with overwhelming pull in the market.
The contrast with rapidly expanding rivals is important. One analysis reported hiring growth of 97% for OpenAI and 152% for Anthropic, compared with 27% for DeepMind. 1 Those figures describe growth rates rather than overall talent quality, but they help explain why DeepMind’s share of regional hiring has fallen even while the lab remains large.
The available figures should also be read as measures of visible hiring and movement, not a complete census of AI expertise. They indicate a weakening recruiting position, not the disappearance of DeepMind’s research base.
Where departing DeepMind researchers are going
The most visible warning sign is the seniority and importance of some departures. In June 2026, Gemini co-lead Noam Shazeer left Google for OpenAI, while AlphaFold researcher John Jumper announced a move to Anthropic. 33
35
40
Gemini contributors Jonas Adler and Alexander Pritzel were also reported to be leaving for Anthropic. 35 Later, Jeff Dean, Sanjay Ghemawat, Oriol Vinyals and Quoc Le were reported to have left Google to start a new company.
17
18
These moves matter for more than their effect on headcount. Senior researchers carry institutional knowledge, technical judgment and recruiting credibility. Their exits can affect the development of foundation models and infrastructure directly, while also influencing how prospective employees view the organization.
The destination pattern is equally significant. Zeki data reported by DiarioBitcoin found that, among DeepMind employees who left during the previous year, 25% went to Anthropic, 21% to Meta and 14% to OpenAI. 3 Separately, SignalFire-based reporting in 2025 found that DeepMind employees were moving to Anthropic at a far higher rate than the reverse, with the imbalance estimated at almost 11:1.
8
No single dataset captures every move, and the figures use different time periods and methodologies. Together, however, they point to a consistent pattern: Anthropic, OpenAI and Meta have become credible destinations for people who once might have viewed DeepMind as the natural choice.
Compensation is part of the explanation—but not all of it
Frontier AI labs are competing in a labor market where cash, equity and access to scarce computing resources can materially influence career decisions. Reporting has described aggressive compensation packages from rivals, especially as OpenAI and Anthropic expand their European operations. 4
11
But compensation alone does not explain why DeepMind’s position has weakened. The lab’s original appeal was a combination of prestige, technical resources and freedom to pursue long-horizon research. As Gemini has become central to Google’s competitive and commercial strategy, DeepMind has faced greater pressure to deliver products and models on corporate timelines.
That creates a potential trade-off. A research organization can gain focus and execution speed by concentrating resources on Gemini, agents and related products. At the same time, researchers who value open-ended investigation may see less room for exploratory work.
Reports have also pointed to tighter publication constraints and a reduced sense of autonomy, although the public evidence does not quantify how much either factor has contributed to individual departures. 7 It would therefore be too strong to attribute the exodus to one policy or one executive decision. The more defensible conclusion is that compensation, commercial pressure, organizational change and the growing appeal of rival labs are reinforcing one another.
The AlphaFold restructuring sent a powerful signal
DeepMind’s shift in priorities is especially visible in the treatment of AlphaFold. Reporting in July 2026 said the team behind the Nobel Prize-winning protein-structure system had been dismantled or reorganized as the lab moved resources toward large language models, Gemini and broader scientific automation. 49
50
DeepMind said researchers had moved into areas including Gemini projects, enzyme design, nuclear fusion and genomics. 49 That means the change should not be described simply as abandoning scientific AI. It is better understood as a reallocation of research around a broader Gemini-centered strategy.
Even so, the organizational signal is consequential. AlphaFold represented the kind of long-horizon, domain-specific research that helped make DeepMind distinctive. Reassigning much of that work can reassure investors and product teams that the lab is focused on near-term strategic priorities. For researchers, it may raise a different question: how much independence remains for projects whose commercial payoff is uncertain or distant?
Hassabis’s new role increases Alphabet alignment
In August 2026, Demis Hassabis gave up day-to-day responsibility as DeepMind’s CEO and became chair of Google DeepMind and chief scientist of Alphabet. Koray Kavukcuoglu, previously the lab’s chief technology officer, assumed operational control and reports directly to Alphabet CEO Sundar Pichai. 17
18
22
The change has an obvious strategic logic. Hassabis can work across Alphabet on long-term AI and scientific priorities, while Kavukcuoglu can focus on Gemini development, frontier research, and the Gemini app and developer teams. 22
53
It also changes DeepMind’s institutional identity. The reporting structure places more authority directly with Alphabet’s central leadership, rather than preserving the impression of a relatively independent London-based research lab. Reuters reported that Kavukcuoglu would have final say on major DeepMind decisions. 18
That could improve coordination and speed. It could also make the lab look more like a product and corporate AI division than the unusually autonomous research institution that originally attracted elite scientists. Whether that perception becomes a recruiting problem will depend on what researchers are actually allowed to work on and publish—not only on the titles of its leaders.
DeepMind still has important strengths
The decline in hiring share should not be confused with a collapse in technical capability. Metix’s 2026 talent map counted 2,837 people in Google DeepMind’s visible technical pool, behind Meta’s 4,092 and ahead of Anthropic’s 1,251 in that dataset. 6
DeepMind also retains advantages that rivals cannot easily reproduce:
- Large-scale computing and infrastructure through Alphabet’s resources and custom hardware ecosystem.
- Scientific AI expertise, including the legacy of AlphaFold and work connected to biology and scientific discovery.
- Research depth in areas such as reinforcement learning, multimodal systems, agents and world models.
- Distribution through Google’s consumer, cloud and developer products.
- Capital and organizational reach sufficient to fund both frontier-model development and longer-term research.
DeepMind’s own research portfolio continues to include systems such as SIMA 2, Gemini Robotics and Genie 3, reflecting ongoing work in agents, robotics and interactive world models. 56
These assets mean that DeepMind remains a formidable competitor. The central issue is whether its resources can be converted into products and research results quickly enough—and whether the organization can retain the people needed to make that conversion happen.
Where DeepMind appears to be losing ground
The lab’s most exposed areas are those where research leadership, product iteration and developer adoption reinforce one another. Rivals have built strong identities around consumer assistants, coding tools and frontier general-purpose models. OpenAI and Anthropic in particular have become attractive destinations for researchers who want to work on highly visible products with a direct connection to the current model race. 53
Contemporary benchmark comparisons have also contributed to the perception that Google’s leading shipped models are behind some competitors, although rankings vary by model, task and date. Fortune cited assessments placing Google’s best shipped model behind systems from Anthropic, OpenAI, xAI, Meta and a Chinese lab. 6 That is a snapshot rather than a permanent ranking, but perception matters in a talent market: researchers often want to join the organization they believe is setting the pace.
Meta presents a different kind of challenge. It can combine substantial compensation with a huge distribution ecosystem, while Anthropic and OpenAI offer focused frontier-model missions and strong recruiting narratives. Google’s infrastructure advantage is real, but infrastructure is not self-executing. It matters only if the organization can keep the scientists and engineers who determine what to train, how to evaluate it and which products to ship.
The real change is the loss of a monopoly on prestige
The strongest conclusion is not that DeepMind has become weak. It is that its old combination of prestige, autonomy and technical resources no longer gives it an uncontested advantage in Europe.
From 2022–23 to 2025–26, DeepMind’s share of regional elite-AI hiring fell from 49% to 18.6%, while its arrivals-to-departures ratio fell from about 12:1 to 2:1. 2 High-profile departures have made the shift visible, and the lab’s tighter connection to Alphabet has intensified questions about its research identity.
DeepMind can still recover. Its path depends on executing on Gemini and related products without eliminating the scientific freedom that made the organization unusually attractive in the first place. The next phase of the AI talent war will test not just who can pay the most, but which lab can offer researchers the strongest combination of autonomy, resources, influence and evidence that its work will matter.
Studio Global AI
Continue your research
This page includes a source-backed answer you can continue inside Studio Global.
People also ask
What is the short answer to "Google DeepMind’s AI Talent Advantage Has Collapsed Across Europe"?
Google DeepMind’s share of research and advanced engineering hires across Europe, the Middle East and Africa fell from 49% in 2022–23 to 18.6% in 2025–26.
What are the key points to validate first?
Google DeepMind’s share of research and advanced engineering hires across Europe, the Middle East and Africa fell from 49% in 2022–23 to 18.6% in 2025–26. The shift reflects more than pay: DeepMind is under pressure to commercialize Gemini, while researchers may perceive less autonomy than the lab once offered.
What should I do next in practice?
DeepMind retains major advantages in compute, infrastructure, scientific AI and research depth—but its former position as Europe’s default destination for elite AI talent is no longer secure.