Talent Intelligence Report · Talent Map

Frontier AI Labs Talent Landscape & Flow Map

Built on Metix AI's global talent database of 860M+ profiles, a full-population profile of 13 frontier AI labs across the US, UK and France: who feeds and loses talent to whom, each lab's staff composition and retention, a named talent-flow network, and an org reconstruction of the engineer and researcher tiers.

Report date 2026-06-11 Produced by Metix AI Coverage 13 labs · 19,057 current-employee profiles across the US, UK and France
Executive Summary

01 Meta's research track and DeepMind together are 58% of the technical pool

The figures below follow Metix AI database methodology (data as of roughly the first half of 2026); the population = talent currently employed at the 13 frontier AI labs and based in the US, UK or France.

13
Frontier AI Labs
OpenAI to Mistral, across the US, UK and France
19,057
current-employee profiles
Across the US, UK and France
11,914
technical talent pool
Research + Engineering + MTS + Safety
41.5%
Research-track share
Research Scientists + Research Engineers
36.3%
Technical-pool PhD rate
Whole technical pool

① Meta's research track and DeepMind together are 58% of the technical pool

Meta (AI research track) at 4,092 and Google DeepMind at 2,837 together account for 58% of the visible 11,914-person technical pool. OpenAI (2,628) and Anthropic (1,251) follow. Profile coverage differs: DeepMind is counted as a standalone entity, so researchers sitting under the Google parent are undercounted; SSI's team is tiny and secretive.

② Anthropic is the net winner; xAI is contracting

On SignalFire's methodology, Anthropic leads on two-year retention at 80%, with the OpenAI→Anthropic net flow running roughly 8:1; of xAI's 12 co-founders, all 11 besides Musk have departed (including 4 co-founders who left in quick succession in early 2026). Tenure data confirms it: median tenure for current technical talent is just 8 months at xAI and 9 months at Anthropic, both in a phase of rapid expansion and turnover.

③ Two poles on the research track: flat at OpenAI, explicit at Meta/DeepMind

OpenAI and Anthropic make heavy use of the flat "Member of Technical Staff" title (the visible research-track share is in single digits), while Meta's research track and DeepMind explicitly label Research Scientists (73.3% / 50.2%). The implication: evaluating OpenAI/Anthropic talent has to come back to the body of work, not the title.

④ Mobility rebounds in H2 2026

Meta MSL's high-priced 2025 hiring cohort has a median tenure of 23 months, with 40.0% past the 36-month mark; layer on mid-level mobility at xAI and OpenAI's expiring retention grants, anda 1703-person "high-mobility" list makes the second half of 2026 the high-water mark for mobility in this talent market.

About this report.Covers the distribution, flow network and representative profiles of frontier AI lab technical talent. The full long list and contact details are available through the Metix AI platform.
The Talent War 2024-2026

02The Battlefield: a two-year talent-war timeline

Every event below has been verified item by item against public sources (source list in the research backup); we keep only the facts that bear on talent-flow judgments. All dollar figures are as reported; most have not been confirmed by the companies.

① Meta's multi-billion talent grab (summer 2025) reset the price anchor into nine figures

In 2025-06 Altman publicly accused Meta of offering "$100M signing bonuses"; within two weeks 8 people jumped from OpenAI to Meta (including the Zurich trio); Meta hired foundation-model lead Ruoming Pang from Apple (a reported $200M+ multi-year package); of the 44-person MSL roster, roughly 75% hold a PhD. Meta officially denied only the "single signing-bonus" wording.

② The defensive toolkit takes shape: equity recalibration + mass retention grants

OpenAI Chief Research Officer Mark Chen's internal note — "someone has broken into our home and stolen something" — was followed by recalibrating comp; as reported in 2025-08, the company handed roughly 1,000 research/engineering staff a retention bonus of about $1.5M each, vesting over two years. Failed acquisition turning into hiring campaigns became Meta's standard play (SSI and Thinking Machines were both probed and then hired; the Tulloch case was reported at up to $1.5B/6 years, a figure Meta denied).

③ Anthropic is the net winner; xAI is the net outflow

On SignalFire's methodology: the OpenAI→Anthropic net flow runs about 8:1 and DeepMind→Anthropic about 11:1; two-year retention is Anthropic 80% > DeepMind 78% > OpenAI 67% > Meta 64%. In 2026-05 Karpathy joined Anthropic's pretraining team (company-confirmed). On the xAI side: of the 12 co-founders, all 11 besides Musk have departed (as of 2026-03), including 4 co-founders who left in quick succession in early 2026.

④ H1 2026: reverse hiring and a widening front

In 2026-02 OpenAI hired back two researchers from Meta within a single week (Ruoming Pang and Pengchuan Zhang); the talent war widened from researchers to commercialization executives (Salesforce/Snowflake/Datadog leaders, Palantir FDEs) and quant talent (a named Jane Street case moving to Anthropic); and Chinese tech giants' reverse hiring was confirmed (Yonghui Wu joining ByteDance Seed, Shunyu Yao becoming Tencent's chief AI scientist).

Timing-window read.Against this report's tenure and flow data (Sections 3-4): ① the 1-year wobble window for Meta MSL's high-priced 2025 hiring wave lands in H2 2026; ② mid-level follow-on mobility after xAI's co-founder exodus is already underway; ③ OpenAI's retention grants vest over 2 years through 2027-08, and the 6 months before expiry are the leading edge of the window; ④ the $100,000 H-1B surcharge was struck down by a federal court on 2026-06-08, temporarily lowering cross-border hiring costs.
Talent Panorama

03Talent Landscape: the assets of 13 labs

Population = 11,914 technical staff (Research Scientists / Research Engineers / MTS / Engineering / Safety & Alignment) across the US, UK and France.

3.1 Technical talent-pool size

Meta (AI research track)
4,092
Google DeepMind
2,837
OpenAI
2,628
Anthropic
1,251
xAI
710
Mistral AI
263
Thinking Machines
64
Reflection AI
21
H Company
19
World Labs
12
Physical Intelligence
9
Kyutai
8
Metix AI database methodology, not company headcount. Meta is the identifiable AI-research-track subset (conservative methodology). n = 11,914.

Read: Meta (AI research track) and Google DeepMind together account for 58% of the visible technical pool. Note that profile coverage differs across labs: DeepMind is counted as a standalone entity, so researchers sitting under the Google parent are undercounted; SSI's team is tiny and secretive (publicly reported at about 20 people), so visible profiles are naturally sparse.

3.2 Size × research density: two lab archetypes

50100200500100020005000020406080Overall mean 41.5%OpenAIAnthropicGoogle DeepMindMeta (AI research track)xAIMistral AIThinking MachinesReflection AIWorld LabsPhysical IntelligenceH CompanyKyutaiTechnical talent-pool size (people, log axis)Research-track share %
Research track = Research Scientists + Research Engineers (MTS not counted, so this is a lower bound). Green line = overall mean 41.5%.

Read: research density is bimodal. Meta's research track (73.3%), DeepMind (50.2%), Mistral, Kyutai and H explicitly label the research track and clear the 50% mark; OpenAI (9.6%), Anthropic (6.4%) and xAI use the flat MTS title, so their research share is understated.

3.3 Geographic distribution: the SF Bay Area dominates

Bay Area
5,545
Rest of US
1,889
Seattle
1,083
New York
887
London
627
Paris
244
Rest of UK
115
Rest of France
47
Grouped by each profile's home city. A further 1477 profiles list no city. By country: United States 10,506 · United Kingdom 1,053 · France 355

3.4 Function × lab matrix: who is stockpiling which kind of talent

Meta (AI track)DeepMindOpenAIAnthropicxAIMistral AIThinking MachinesReflection AIH CompanyResearch Scientist27079182106835113110Research Engineer29150743121232MTS (unspecified)23170296242035919Engineering10841388607164231121327Safety & Alignment82166452331
Cell = the lab's technical-staff headcount in that function (color normalized across the whole matrix). MTS is the research/engineering-agnostic title specific to OpenAI/Anthropic/SSI.

Read: OpenAI and Anthropic have the thickest MTS column (a flat title that doesn't split research from engineering), while Meta's research track and DeepMind have the thickest Research Scientist column; Safety & Alignment is Anthropic's relative signature column. For buyers: if you want "clearly designated Research Scientists," hire from Meta/DeepMind; if you want "full-stack MTS," hire from OpenAI/Anthropic.

3.5 The intake pipes: where they come from

Source (previous employer)Current employerStartups/other · 5054Google · 2052Academia/research · 1657Meta/FAIR · 507Amazon/AWS · 405Stripe/Scale alumni · 401Microsoft · 326Apple · 240Chinese tech giants · 184NVIDIA · 88Other sources (combined) · 83Quant funds · 77OpenAI · 67Tesla · 30DeepMind · 29Twitter/X · 17Meta (AI research track) · 3831Google DeepMind · 2694OpenAI · 2464Anthropic · 1209xAI · 671Mistral AI · 253Thinking Machines · 53Reflection AI · 19H Company · 15World Labs · 8
Technical talent's current lab (Top) and most recent external role (skipping early stints at the same lab), n = 11242. Band width = headcount.

Read: three main intake pipes stand out. ① Internal transfers within the big techs are the thickest pipe (Google→DeepMind is the largest, since it's the same system to begin with); ② direct hiring from academia/research (especially Meta's research track and DeepMind) is the primary entry point for research talent, confirming that top PhDs are spoken for before they graduate; ③ the large "startups/other" volume reflects the high-frequency flow among frontier labs and the broader AI startup ecosystem. Quant funds (Jane Street/Citadel alumni) have emerged as a new pipe, with named cases already appearing at Anthropic/OpenAI.

3.6 Hiring waves: the temporal shape of the talent war

0100200300400500600700800900100011001200130014001500160017001800190020002100220023002400250026002700280029003000310032003300340035003600370038003900400041004200430044004500460047004800201670201711320182032019263202033920214872022853202389520242792202547632026685Meta (AI research track)Google DeepMindOpenAIAnthropicxAIMistral AIOther labs
How to read it: this counts the "start year of current employees' current role," so earlier cohorts are diluted by attrition (survivorship), and more recent years are closer to true hiring intensity. 2026 only includes hires up to the data cutoff (roughly the first few months).

Read: 4763 of the current technical staff joined in 2025, 5.3x the 2023 trough (895), and the explosive hiring of 2024-2025 is the temporal shape of this talent war. Mind survivorship: earlier cohorts are already diluted by attrition, so the curve understates historical hiring and gets closer to true intensity the more recent the year.

3.7 Tenure structure: where the mobility windows are

612182401020304050OpenAIAnthropicGoogle DeepMindThinking MachinesMeta (AI research track)xAIMistral AIMedian tenure of current technical staff (months)Share with tenure ≥ 36 months %
Bubble area = sample size. Only labs with a technical-pool sample of ≥ 30 are included. Tenure = duration from the start of the current role to the data cutoff.

Read: the upper-right "veterans zone" = Meta's research track (median tenure 23 months, 40.0% past 36 months, the sediment of FAIR old-timers); the lower-left "new recruits zone" = xAI (8 months), Anthropic (9 months) and Mistral (8 months), all expanding fast, with onboarding-honeymoon staff relatively stable before entering their first round of mobility at 12-24 months. OpenAI (13 months) and DeepMind (15 months) sit in the middle.

Talent Flow

04Flow & Talent Movement: who feeds and loses talent to whom

Net flow since 2024 (sources include the four dissolution/diffusion origins Character.AI/Inflection/Stability/Adept). Destination structure and ratios are the more robust read.

4.1 Lab mutual-flow matrix (net flow between companies)

Meta (AI research track)DeepMindOpenAIAnthropicxAIMistral AIThinking MachinesReflection AIOpenAI leavers403556132203Anthropic leavers29733DeepMind leavers115935430111011xAI leavers821021Mistral leavers123Character.AI alumni6231152163Inflection alumni22121Stability alumni5322Adept alumni256111
Rows = source (that company's leavers), columns = current lab. Cell = headcount. Off-diagonal flow (e.g., OpenAI leavers now at Anthropic) is the talent-flow network.

Read: the talent-flow network is this report's most directly actionable asset for VCs and recruiters. The off-diagonal cells are the proven flow channels (lowest psychological resistance for candidates): the more people who leave one lab and land at another, the more "open" that hiring path is. The diffusion paths of the three teams dissolved by big-tech acqui-hires — Character.AI, Inflection and Adept — are especially worth watching.

4.2 Retention-rate comparison (database methodology vs SignalFire)

Anthropic
87.5%
Google DeepMind
63.3%
OpenAI
73.6%
Mistral AI
92.1%
xAI
69.6%
Metix AI visible retention = current employees / (current employees + visible leavers), all-history methodology, affected by the search cap and so directional only. SignalFire's 2025 report, 2-year retention methodology: Anthropic 80% > DeepMind 78% > OpenAI 67% > Meta 64%.

Read: the two methodologies corroborate each other directionally. Metix AI visible retention (current / current + visible leavers) runs high in absolute terms because of the sample cap, but the ranking across labs matches SignalFire's 2-year cohort methodology (Anthropic 80% > DeepMind 78% > OpenAI 67% > Meta 64%): Anthropic holds onto people, Meta doesn't.

4.3 Inflow/outflow balance (since 2024, sample)

← Visible outflow (since 2024)→ Visible inflow (current employees who joined since 2024)11144123Net +3009OpenAI3202344Net +2024Anthropic12332878Net +1645Google DeepMind7331443Net +710xAI40473Net +433Mistral AI
Inflow = joined 2024-2026 and still employed (technical pool); outflow = departed since 2024 (conservative).

Read: inflow and outflow are both samples, so net values are directional. Labs in expansion (Anthropic/OpenAI) show inflow well above visible outflow; against the backdrop of a co-founder exodus, xAI's outflow direction is clear. True outflow exceeds the conservative count, but the ranking is trustworthy.

4.4 Where leavers go: the founding rate is a VC's signal light

OpenAI9%10%7%52%4%18%n=1966Google DeepMind14%10%8%42%10%15%n=2394Anthropic6%12%52%22%n=393xAI7%5%53%28%n=749Mistral16%9%7%60%9%n=45Character.AI33%6%9%42%6%n=175Inflection6%74%13%n=1458Stability AI78%14%n=2060Adept4%76%14%n=1150Other frontier labsFounded a companyInternet big techsStartups/otherQuant fundsChinese big techs / foundation-model labsAcademia/researchDestination not yet public
Each row = the current-destination composition of that source's leavers. "Founded a company" is identified by current titles containing founder/stealth, so it's a lower bound.

Founding-rate ranking (share of leavers now founders)

H Company
18.2%
Anthropic
12.0%
Google DeepMind
10.4%
OpenAI
9.5%
Mistral
8.9%
xAI
6.9%
Character.AI
6.3%
Inflection
6.2%
Adept
4.1%
Stability AI
3.2%
VC view: the sources with the highest founding rate are the first concentration of spinout deal flow.

Read (VC view): the sources with the highest founding rate are the first concentration of spinout deal flow. Leavers from dissolved teams like Character.AI/Inflection/Adept found companies and join new labs at notably higher rates; DeepMind/OpenAI leavers flow more to other frontier labs and internet big techs. "Founded a company" is identified by current titles containing founder/stealth, a lower bound, so the true number of founders is higher.

Org Reconstruction

05Org Reconstruction: rebuilding down to the mid-level

Paid org charts like The Information's only cover the executive layer. This section uses full-population profiles to push the reconstruction down to the team-lead tier and IC depth. Levels are classified from public job titles — not the official org structure, just an overview of team-tier structure; OpenAI/Anthropic's many MTS titles carry no level information. Names are masked by default in the public version.

OpenAI · technical-track org reconstruction

2628 total (database methodology)

Below the dual research leadership (CRO Mark Chen + Chief Scientist Jakub Pachocki), the public org chart stops at the VP layer. After the 2026 departures of Tworek/Weil/Peebles and others, the mid-level is the key to understanding OpenAI.

Leadership (founders/executives/director level) · 48
B●● W●●
Sr.Director Of AI And Data Platform
R●● H●●
Head Of Hardware
K●● W●●
Director Sales Engineering
G●● C●●
Head Of App
K●● G●●
Director - Corporate Security Protective Intellig…
L●● C●●
Senior Director Of AI Systems Engineering
B●● S●●
Head Of Industrial Security
I●● L●●
Head Of AI Infrastructure
Team-lead tier (Manager / Lead) · 131
E●● K●● · Software Engineer ManagerJ●● L●● · ManagerR●● X · Engineering ManagerD●● T●● · AI Platform Product Manag…S●● C●● R●● J●● J●● W●● · AI Platform Product Manag…L●● M●● · AI Platform Product Manag…L●● M●● · AI Platform Product Manag…J●● K●● · Team Lead, Emerging RiskM●● B●● · Crisis ManagerDenny Stiegler · S●● P●● · Corporate Security ManagerW●● G●● · AI Platform Product Manag…
IC depth
50 Staff/Principal59 Senior2340 Other ICs

Anthropic · technical-track org reconstruction

1251 total (database methodology)

Doubled in a year to about 5,200 people (publicly reported). Pretraining-team head Nick Joseph; in 2026-05 Karpathy joined that team (public source).

Leadership (founders/executives/director level) · 15
B●● S●●
Director Of AI Systems
V●● C●●
Director Of AI Engineering
A●● E
Director Of Engineering
S●● R●● H●●
Head Of AI Engineering
T●● Y●●
Vice President
H●● Y●●
Vice President
M●● L●●
Head Of ML, Trust Safety At Anthropic
A●● S●●
Distinguished Scientist
Team-lead tier (Manager / Lead) · 49
M●● L●● · Environment, Health And S…D●● L●● · Member Of Technical Staff…R●● T●● · Engineering Manager, Prod…X●● Z●● · Member Of Technical Staff…M●● M●● · Member Of The Technical S…T●● N●● · Member Of Technical Staff…J●● G●● · Hardware Operations LeadS●● C●● · Interpretability Team Man…D●● V●● · Strategic Sourcing LeadS●● H●● · Technical Program Manager…Y●● W●● · Research Manager, Interpr…J●● J●● · IT Engineering Project Ma…
IC depth
3 Staff/Principal11 Senior1173 Other ICs

Google DeepMind · technical-track org reconstruction

2837 total (database methodology)

Dual hubs in London + the Bay Area (CTO Kavukcuoglu relocated to Mountain View in 2025, doubling as Google's chief AI architect). Team-lead attrition is its main risk (Microsoft AI hired 20+).

Leadership (founders/executives/director level) · 72
D●● S●●
Distinguished Scientist
C●● B●●
Director Principal Scientist
M●● G●●
Distinguished Engineer
M●● M●●
Director And Principal Scientist For Human-AI Int…
A●● F●●
Research Director
F●● Y●●
Principal Engineer (Director)
K●● M●●
Distinguished Engineer
S●● M●● A●● E●●
Director, Principal Scientist
Team-lead tier (Manager / Lead) · 84
R●● E●● · Team LeadS●● T●● E●● · Research And Eng Lead, Ma…E●● B●● · Cybersecurity Research Te…D●● M●● · Design Strategy UX Resear…M●● M●● · Lead Ai UX EngineerA●● D●● · Software Engineering Mana…M●● B●● · Agentic Evaluation Simula…C●● O●● · Data Science Lead, Gemini…K●● G●● · Software Engineering Mana…A●● W●● · Engineering Manager (L6)A●● C●● · Business And Corporate De…J●● L●● · Engineering Manager
IC depth
687 Staff/Principal571 Senior1423 Other ICs
Note: leadership sketches the senior landmarks of each direction; the team-lead tier (Manager/Lead) reflects the distribution of the core workforce; IC depth reflects team size and tier structure. OpenAI/Anthropic's many MTS carry no level information, so capability assessment has to come back to model-contribution rosters and papers. Names are masked by default in the public version.
Notable People

06Representative Profiles

Three groups of representative profiles, selected from the list by level, direction and strength of track record. Profile facts come from the Metix AI database; those marked "publicly verified" have had their current role confirmed against 2025-2026 public sources (frontier-lab profiles update with a lag, so public sources take priority). All profiles in this section come from public professional records. Group A is industry public figures, Group B is senior technical backbone, Group C is crossover and high-potential profiles. Names are masked by default in the public version.

Group A · Helm & landmarks (industry public figures)

M●● C●● Publicly verified
OpenAI · Head Of Frontiers Research (San Francisco)
14+ yrs experience · National Experimental High School · Massachusetts Institute of Technology
OpenAI Chief Research Officer (CRO). MIT undergrad, a former quant trader, joined in 2018 and led Codex, DALL·E, GPT-4 vision and o1. During Meta's 2025-06 talent grab he sent the "someone has broken into our home and stolen something" internal note and led the comp recalibration.
J●● P●● Publicly verified
OpenAI · Chief Scientist (San Francisco)
9+ yrs experience · Carnegie Mellon University · University of Warsaw (PhD)
OpenAI Chief Scientist. Succeeded Ilya Sutskever in 2024, co-leads research and sets the technical roadmap alongside Mark Chen, and is a key architect of the o1/o3 reasoning line.
J●● K●● Publicly verified
Anthropic · Co-Founder And Chief Science Officer (Pacifica)
22+ yrs experience · Stanford University · Harvard University (PhD)
Anthropic co-founder and Chief Science Officer (CSO). A core author of the Scaling Laws, he left OpenAI in 2021 to co-found Anthropic.
L●● W●● Publicly verified
Thinking Machines · Co-Founder (unknown)
12+ yrs experience · The University of Hong Kong · Peking University (PhD)
Thinking Machines Lab co-founder. Peking University EECS undergrad, former OpenAI VP of safety systems / applied research, and author of the technical blog Lil'Log. A core research force in Mira Murati's team.
F●● L●● Publicly verified
World Labs · Cofounder, CEO (Stanford)
43+ yrs experience · Caltech · Princeton University (PhD)
World Labs co-founder and CEO. Co-director of Stanford's Human-Centered AI institute (HAI), creator of ImageNet, the "godmother of AI." Working on spatial intelligence / world models, she closed a roughly $1 billion (10 × $100M) round in 2026-02.
k●● k●● Publicly verified
Google DeepMind · VP Of Research (unknown)
36+ yrs experience · New York University (PhD)
Google DeepMind CTO; in 2025-06 he also became Google's first chief AI architect (SVP, reporting directly to Pichai), embedding Gemini across all Google products. Having relocated from London to Mountain View, he symbolizes the westward shift of the research center of gravity.
D●● Z●● Publicly verified
Google DeepMind · Principal Scientist Research Director, Google DeepMind (unknown)
23+ yrs experience · Chinese Academy of Sciences (PhD)
Google DeepMind principal scientist / research director, founder of the Gemini Reasoning Team, and a foundational researcher in chain-of-thought (CoT) and LLM reasoning. Peking University / mathematics background.
J●● L●● Publicly verified
Meta (AI research track) · Research Scientist (San Francisco)
9+ yrs experience · Tsinghua University · Massachusetts Institute of Technology (PhD)
Profile current role: Meta AI Research Scientist. MIT PhD, core to GPT-4o/o4-mini multimodality and efficiency, a member of Meta's 2025 recruitment roster.
J●● R●● Publicly verified
Meta (AI research track) · AI Research Scientist (unknown)
12+ yrs experience · University of Bristol · Carnegie Mellon University (PhD)
Profile current role: Meta AI Research Scientist. Gemini pretraining tech lead (drove the Gopher/Chinchilla scaling work), hired from Google DeepMind into Meta MSL in summer 2025.
C●● L●●
Meta (AI research track) · AI Research Scientist Director (Cambridge)
32+ yrs experience · Massachusetts Institute of Technology (PhD)
Meta AI Research Scientist Director. MIT PhD (computer vision), former Microsoft Research, a senior researcher in optical flow / image generation.

Group B · Senior technical backbone (lead / Staff level)

X●● C●●
Meta (AI research track) · AI Research Scientist (Mountain View)
6+ yrs experience · Shanghai Jiao Tong University · University of California, Berkeley (PhD)
Profile current role: Meta Superintelligence Labs Research Scientist. Former Google DeepMind (code generation/reasoning, author of "LLM as Optimizers"), hired into MSL in 2025-07; a scarce research-track profile.
W●● C●●
Meta (AI research track) · Director Of Engineering (San Francisco)
16+ yrs experience · University of Science and Technology of China · Washington University in St. Louis (PhD)
Meta Director of Engineering. USTC undergrad with a Washington University background, working in AI infrastructure/engineering.
F●● Y●●
Google DeepMind · Principal Engineer (Director) (Mountain View)
38+ yrs experience · Tsinghua University · Tsinghua University · Ecole polytechnique fédérale de Lausanne (PhD)
Google DeepMind Principal Engineer (director level). Tsinghua undergrad, a long-tenured senior engineering leader at Google, working in multimodality.
H●● Z●●
Google DeepMind · Engineering Director At Google DeepMind (San Francisco)
29+ yrs experience · Shanghai No 8 High School · Pasadena City College (PhD)
Google DeepMind Engineering Director. A senior leader in computer vision / multimodal engineering.
陈●● J●● C●●
Google DeepMind · Senior Staff Software Engineer And Tech Lead Manager (San Francisco)
30+ yrs experience · The Affiliated High School of South China Normal University · University of Science and Technology of China · University of Kentucky (PhD)
Google DeepMind Senior Staff Software Engineer and tech lead manager. A senior systems engineer who came up through Willow Garage/Cuil — an engineering veteran.
J●● S●●
Meta (AI research track) · Principal ML Engineer (Sunnyvale)
Stanford University (PhD)
Meta Principal ML Engineer. Stanford PhD with a track record spanning Google/Microsoft; a senior ML-engineering profile.
Y●● X●●
Meta (AI research track) · Applied Research Scientist, Uber Tech Lead (San Francisco)
44+ yrs experience · Tsinghua University · University of Electronic Science and Technology of China · University of Southern California (PhD)
Meta Applied Research Scientist and tech lead. Tsinghua undergrad, working in graph computing / recommender systems, active in the IEEE.
E●● H●●
Meta (AI research track) · Engineering Manager (Los Angeles)
30+ yrs experience · University of California, Los Angeles · Harvey Mudd College (PhD)
Meta Engineering Manager. UC PhD with a track record including Google/Adobe.
Y●● W●●
Anthropic · Research Manager, Interpretability (Newark)
24+ yrs experience · University of Virginia (PhD)
Anthropic Research Manager in Interpretability. This is Anthropic's flagship research direction, and team-lead-tier figures in it are extremely scarce in public profiles.
P●● H●●
xAI · Inference Lead (Palo Alto)
4+ yrs experience · National Taiwan University · University of California, Berkeley
xAI Inference lead. NTU background, former LinkedIn, a mid-level tech lead who stayed on after xAI's co-founders departed one after another; in a period of organizational reshuffling.
A●● L●●
Mistral AI · Runtime Co-lead (Cambridge)
7+ yrs experience · University of Cambridge
Mistral AI Runtime Co-lead. Cambridge background, a core mid-level engineering figure at Europe's sovereign-AI standard-bearer.
B●● W●●
OpenAI · Sr. Director Of AI And Data Platform (unknown)
University of Washington
OpenAI Senior Director of AI and Data Platform. Washington University, former Microsoft, a manager in the platform/infrastructure direction.

Group C · Crossover & rising talent (structurally scarce profiles)

P●● Z●● Publicly verified
OpenAI · Member Of Technical Staff (Pasadena)
14+ yrs experience · Tsinghua University · Caltech (PhD)
Profile current role: OpenAI Member of Technical Staff (MTS). A Tsinghua alum who returned from Meta to OpenAI in 2026-02 to work on world models/robotics — one of the two researchers OpenAI hired back from Meta in 2026.
Y●● B●●
Anthropic · Member Of Technical Staff (San Francisco)
8+ yrs experience · University of Toronto · Princeton University (PhD)
Anthropic Member of Technical Staff (MTS). A first-author-level contributor to the Constitutional AI paper, a rare identifiable researcher under Anthropic's culture of secrecy.
How to use this.All profiles in this section come from public professional records and are presented for industry representativeness only. Group A is founders, executives and public technical leads, Group B is senior technical backbone, Group C is crossover and high-potential profiles.
Compensation

07Compensation: the level market and the list market

Frontier-lab compensation has split into two markets: the "level market" you can look up on levels.fyi, and the "list market" that bypasses the leveling system ($10M to $1.5B, as reported). Figures are the median of self-reported samples retrieved in 2026-06, not official company numbers.

LabRegular-level median TCSenior referenceEquity mechanismNotes
OpenAIL5 $819K / L6 $1.23MMTS sample: $300K base + ~$500K/yr PPUPPU profit-participation units, 4-year linear vesting, liquidity via tenderAs reported in 2025-08: ~1,000 people × $1.5M retention grant (2-year vest)
AnthropicSWE median $665KLead median $785KStandard private-company equity + tenderMedian cash below OpenAI, yet first on retention (SignalFire)
Google DeepMindL6 RS $750K-1ML7 $950K-1.4MGSU public stock, best liquidityRS equity 5-15% higher than SWE at the same level
Meta (MSL)E7 median $1.30MList market $10M-100M+/yrRSU + list-based special packageRuoming Pang $200M+ and Tulloch up to $1.5B/6 years are both as reported
xAISWE $205-640KSmall samplePrivate-company options (swapped after the SpaceX acquisition)Folded into the SpaceX system from 2026-02
Mistral (Paris)Paris median €89.5KResearchers $490-950K (estimated)BSPCE optionsEuropean engineering roles trail the US by an order of magnitude; research roles carry a 3-5x premium

Researcher premium and quant bidding wars

The split between the research and engineering tracks has hardened: at DeepMind, same-level research roles carry 5-15% more equity; at Mistral, research total comp is roughly 3-5x engineering; and Meta's list-based packages occur only at the researcher / research-leadership level. Quant funds (Jane Street/Citadel alumni) and the labs bid each other up, and Anthropic/OpenAI proactively host mixers to hire entry-level quants, with named cases already appearing (two Jane Street people moved to Anthropic in 2025).

What it means for recruiters

① The level market can be negotiated against the table; the list market can only be competed for with mission, equity upside and compute freedom; ② the liquidity gap between OpenAI's PPU and DeepMind's GSU is a practical lever in recruiting pitches (PPU depends on the company organizing a tender, GSU can be sold anytime); ③ Europe (Mistral/Kyutai/H) is a compensation depression at equivalent talent density, well suited for budget-constrained buyers building an R&D site.

Playbook

08An action checklist for three reader types

Turn the map into action: VCs watch spinout signals, recruiters watch windows and channels, HR watches defense.

VCs: track spinout signals

① The founding-rate ranking (Section 4.4) points to the high-frequency sources of spinouts; ② several of xAI's co-founders departed in early 2026 with destinations undisclosed — a team-forming signal worth watching; ③ LeCun (AMI Labs $1.03B seed) and Tworek (Core Automation) confirm the "roughly 6 months from executive departure to a closed round" cadence; ④ tenure structure (3.7) can be used to anticipate: if Meta MSL's high-priced 2025 cohort loosens in H2 2026, a new wave of team-forming windows opens with it.

Recruiters: mobility windows and channels

① The mutual-flow matrix (4.1) shows which flow channels have been historically proven; ② tenure windows layered with organizational reshuffles help locate the more mobile groups; ③ the MTS title carries no level, so capability assessment has to come back to public work (model-contribution rosters/papers/systems); ④ the technical pool has a 36.3% PhD rate, and alumni chains + public work are the highest-hit outreach surface.

AI-company HR: a retention reference

① Locate yourself against the retention benchmarks in 4.2; ② the industry's established retention tools — equity refreshes, retention grants and mission narrative — with OpenAI's case giving a price reference ($1.5M × 1,000 people); ③ watch the group with "36+ months of tenure and slowing promotion" (the mirror image of the 3.7 mobility window); ④ non-competes and garden leave are largely ineffective in the US market, so you keep people by giving them direction, not by locking them in.

Turn the map into a list with Metix AI

The retrieval, profiling and flow analysis in this report were all done by Metix AI. We can generate a custom map for any company on the same methodology: full long-list export, org reconstruction, talent-flow monitoring, email unlock and multi-channel outreach — billed on a "pay only for qualified interviews" basis. No interview, no charge.

860M+ global talent profiles11,914-person lab technical-pool long listQuarterly monitoring of the talent-flow networkPay only for qualified interviews
Appendix

09Appendix: methodology, full data and method limitations

9.1 Methodology and approach

The 13 labs covered

OpenAI, Anthropic, Google DeepMind, Meta (AI research track), xAI, SSI, Thinking Machines, Mistral AI, Reflection AI, World Labs, Physical Intelligence, H Company, Kyutai. Geographic scope = profiles based in the US/UK/France. Google DeepMind is counted as a standalone entity, so a small number of people holding a DeepMind title under the Google parent are undercounted.

Meta's special methodology

Meta is enormous and MSL is not a standalone entity, so this report's "Meta (AI research track)" = people currently at Meta whose title/headline is identifiable as AI research/engineering related (including profiles explicitly tagged Superintelligence Labs/FAIR/GenAI). It is an identifiable subset rather than all of Meta AI, so the absolute numbers run conservative.

Function methodology

Technical talent pool = Research Scientists + Research Engineers + MTS (unspecified) + Engineering + Safety & Alignment. OpenAI/Anthropic/SSI make heavy use of the Member of Technical Staff title without splitting research from engineering, broken out separately as MTS, so the research-track share is a lower-bound methodology. Solutions/product/GTM are not counted in the technical pool.

Data recency

The talent database is as of roughly the first half of 2026; frontier-lab profiles update with a lag, so the key profiles have been re-checked one by one against public information, and the latest 2025-2026 role changes are annotated according to public sources.

9.2 Full lab table (Metix AI database methodology, US/UK/France)

Labcurrent-employee profilesTechnical poolResearch-track sharePhD rate
Meta (AI research track)4,2284,09273.3%59.8%
Google DeepMind4,1232,83750.2%36.9%
OpenAI5,4952,6289.6%16.0%
Anthropic2,7401,2516.4%16.0%
xAI1,7177105.1%16.1%
Mistral AI52826351.7%17.5%
Thinking Machines88641.6%43.8%
Reflection AI41210.0%42.9%
H Company321963.2%10.5%
World Labs24128.3%33.3%
Physical Intelligence25911.1%33.3%
Kyutai13875.0%37.5%
SSI300%0%

9.3 Method limitations (essential reading)

Data recency: the talent database is a static snapshot (roughly the first half of 2026); frontier labs see extremely rapid staff turnover and a pronounced profile-update lag (high-frequency job-hoppers often don't update their profiles for months), so all figures are directional rather than real-time; the key profiles have been re-checked and annotated one by one against public sources.

Uneven coverage: this report aggregates public professional records, so coverage is limited for secretive teams like SSI, for DeepMind (counted as a standalone entity) and for Meta (taken as the identifiable AI-research-track subset). Comparisons across labs lean on shares and structure, with absolute numbers secondary.

Function and level inference: based on title/headline keywords; flat titles like MTS make both the research-track share and the team-lead-tier headcount lower bounds.

Outflow and retention: leaver retrieval has a sample cap, so outflow runs conservative; this report's "visible retention rate" differs from SignalFire's 2-year cohort methodology, so read them side by side rather than as substitutes for each other.

Dollar-figure methodology: compensation packages are all as reported by the media and mostly unconfirmed by the companies; please keep the "as reported" qualifier when citing them.

Data and compliance statement.All personal information in this report comes from public professional records, aggregated and organized through the Metix AI database, and is used solely for talent-market research and industry reference; this report contains no judgment of any individual's intent to leave or job performance. If you are an individual mentioned in this report and would like to correct your information or be removed, please contact jc.dai@metix.ai and we will handle it promptly. Industry facts are subject to the cited sources, and compensation is a public-market reference, not an offer.
FAQ

Questions this report answers

How many frontier AI labs and current profiles does this panorama cover?
On Metix AI database methodology (data as of roughly the first half of 2026), 19,057 current-employee profiles at 13 frontier AI labs based in the US, UK or France include an 11,914-person technical pool (research + engineering + MTS + safety).
What share of the 13-lab technical pool sits at Meta and DeepMind?
Meta (AI research track) at 4,092 and Google DeepMind at 2,837 together account for 58% of the visible 11,914-person technical pool. OpenAI (2,628) and Anthropic (1,251) follow.
How should this report be cited?
Metix AI Talent Intelligence, 2026-06-11. Frontier AI Lab Talent Landscape & Flow Map | Metix AI. https://metix.ai/reports/mapping/frontier-ai-labs-talent-2026
Metix AI · Mira | Frontier AI Lab Talent Landscape & Flow Map | 2026-06-11 Talent analytics powered by Metix AI
中文版