Universities (31%) and startups and other companies (25%) employ the largest shares of 2020–2026 entrants
Data as of
- 2020–2026 entrants with a current org
- 819
- At startups and other companies
- 25%
- At frontier labs
- 23%
- At universities
- 31%
Where do the newest people entering AI work, and how does their split across sectors compare with people who entered earlier? Across 6,853 people with a dated work, founding or visiting role (6,230 with a known current organization), we group each person by the start year of their earliest recorded non-education role and compare the 872 people who entered in 2020–2026 (819 with a known current organization) against the 5,981 earlier entrants whose first recorded role started before 2020 (5,411 with a known current organization).
We estimate that 60% of 2020–2026 entrants with a known current organization (488 of 819) now work in a commercial sector: 25% (204) at startups and other companies, 23% (188) at frontier labs and 13% (105) at big-tech companies outside frontier-lab units (9 people work in two). Among earlier entrants who started before 2020, the combined commercial share is 35% (1,906 of 5,411); the 2020–2026 commercial share is 1.7 times that of earlier entrants. At universities, the share is 31% (257 of 819) for 2020–2026 entrants and 66% (3,590 of 5,411) for earlier entrants. Research institutes, nonprofits and government bodies employ 22% (184) of 2020–2026 entrants and 22% (1,166) of earlier entrants.
Key findings
- Universities and startups and other companies lead among 2020–2026 entrants: 31% (257 of 819) work at universities and 25% (204) work at startups and other companies; frontier labs employ 23% (188), 3.0 times the frontier-lab share of earlier entrants (8%, 408 of 5,411).
- The university share of earlier entrants is 2.1 times that of 2020–2026 entrants: 66% of pre-2020 entrants (3,590 of 5,411) currently work at a university, compared with 31% (257 of 819) of 2020–2026 entrants.
- Inside startups and other companies, 51% of 2020–2026 entrants work at AI startup organizations or companies they founded: of the 204 2020–2026 entrants in this sector, 104 (51%) either work at a company carrying the AI startup topic (49) or founded the company where they work (83, with 28 in both groups).
- Universities lead first recorded roles in 2020–2026 at 37%: 326 of 872 2020–2026 entrants started their first recorded role at a university, compared with 56% (967 of 1,717) in the 2010s and 74% (1,183 of 1,593) before 1990.
- Google DeepMind leads individual employers of 2020–2026 entrants: employs 48 2020–2026 entrants, followed by Anthropic (32) and NVIDIA (29).
Current workplaces of 2020–2026 entrants and earlier entrants
Paired horizontal bars compare current workplace sectors for 819 entrants whose first recorded role started in 2020–2026 with 5,411 earlier entrants whose first role started before 2020.
Source: People of Superintelligence career records from biographies and OpenAlex, as of 22 October 2026. Organization topics as of 22 October 2026.
Data table
| Workplace sector | 2020–2026 entrants (current) | Share of 2020–2026 | Earlier entrants (current) | Share of earlier | 2020–2026 first-role starts | Earlier first-role starts |
|---|---|---|---|---|---|---|
| Universities | 257 | 31.4% | 3,590 | 66.3% | 326 (37.4%) | 4,176 (69.8%) |
| Startups and other companies | 204 | 24.9% | 1,045 | 19.3% | 192 (22.0%) | 767 (12.8%) |
| Frontier labs | 188 | 23.0% | 408 | 7.5% | 117 (13.4%) | 52 (0.9%) |
| Institutes | 184 | 22.5% | 1,166 | 21.5% | 154 (17.7%) | 795 (13.3%) |
| Big tech | 105 | 12.8% | 487 | 9.0% | 141 (16.2%) | 393 (6.6%) |
Universities employ 31.4% (257 of 819) of 2020–2026 entrants, below 66.3% (3,590 of 5,411) for earlier entrants. Within the 204 2020–2026 entrants in startups and other companies, 49 (6% of all 2020–2026 entrants with a current organization, compared with 2% or 118 among earlier entrants) work at organizations carrying the AI startup topic outside frontier labs, led by Black Forest Labs (8) and Hugging Face (4), and 83 (10% of 2020–2026 entrants, compared with 6% or 347 among earlier entrants) founded the company where they work (28 of them also in the first group). Among other companies in this sector, NVIDIA leads with 29 2020–2026 entrants.
Frontier labs employ 23.0% (188 of 819) of 2020–2026 entrants today, above 7.5% (408 of 5,411) among earlier entrants. Big-tech companies outside frontier-lab units (Google DeepMind, Meta Superintelligence Labs and Qwen Team count with frontier labs) employ 12.8% (105 of 819) of 2020–2026 entrants and 9.0% (487 of 5,411) of earlier entrants.
Where entrants started by entry cohort
Lines track the share of entrants starting their first recorded role in each sector across 5 entry cohorts from Before 1990 through the 2020s.
Source: People of Superintelligence career records from biographies and OpenAlex, as of 22 October 2026. Organization topics as of 22 October 2026. A person whose earliest recorded year has roles in more than one sector counts in each.
Data table
| Entry cohort | Sector | Started in sector | Share of cohort starts | Work in sector now | Share of cohort with current org |
|---|---|---|---|---|---|
| Before 1990 (n = 1,593 starts; 1,200 current) | Frontier labs | 0 | 0.0% | 12 | 1.0% |
| Before 1990 (n = 1,593 starts; 1,200 current) | Big tech | 43 | 2.7% | 54 | 4.5% |
| Before 1990 (n = 1,593 starts; 1,200 current) | Startups and other companies | 171 | 10.7% | 143 | 11.9% |
| Before 1990 (n = 1,593 starts; 1,200 current) | Universities | 1,183 | 74.3% | 1,002 | 83.5% |
| Before 1990 (n = 1,593 starts; 1,200 current) | Institutes | 232 | 14.6% | 234 | 19.5% |
| 1990s (n = 1,204 starts; 1,146 current) | Frontier labs | 0 | 0.0% | 30 | 2.6% |
| 1990s (n = 1,204 starts; 1,146 current) | Big tech | 49 | 4.1% | 84 | 7.3% |
| 1990s (n = 1,204 starts; 1,146 current) | Startups and other companies | 119 | 9.9% | 179 | 15.6% |
| 1990s (n = 1,204 starts; 1,146 current) | Universities | 923 | 76.7% | 908 | 79.2% |
| 1990s (n = 1,204 starts; 1,146 current) | Institutes | 152 | 12.6% | 261 | 22.8% |
| 2000s (n = 1,467 starts; 1,424 current) | Frontier labs | 0 | 0.0% | 77 | 5.4% |
| 2000s (n = 1,467 starts; 1,424 current) | Big tech | 81 | 5.5% | 169 | 11.9% |
| 2000s (n = 1,467 starts; 1,424 current) | Startups and other companies | 150 | 10.2% | 290 | 20.4% |
| 2000s (n = 1,467 starts; 1,424 current) | Universities | 1,103 | 75.2% | 946 | 66.4% |
| 2000s (n = 1,467 starts; 1,424 current) | Institutes | 180 | 12.3% | 322 | 22.6% |
| 2010s (n = 1,717 starts; 1,641 current) | Frontier labs | 52 | 3.0% | 289 | 17.6% |
| 2010s (n = 1,717 starts; 1,641 current) | Big tech | 220 | 12.8% | 180 | 11.0% |
| 2010s (n = 1,717 starts; 1,641 current) | Startups and other companies | 327 | 19.0% | 433 | 26.4% |
| 2010s (n = 1,717 starts; 1,641 current) | Universities | 967 | 56.3% | 734 | 44.7% |
| 2010s (n = 1,717 starts; 1,641 current) | Institutes | 231 | 13.5% | 349 | 21.3% |
| 2020s (n = 872 starts; 819 current) | Frontier labs | 117 | 13.4% | 188 | 23.0% |
| 2020s (n = 872 starts; 819 current) | Big tech | 141 | 16.2% | 105 | 12.8% |
| 2020s (n = 872 starts; 819 current) | Startups and other companies | 192 | 22.0% | 204 | 24.9% |
| 2020s (n = 872 starts; 819 current) | Universities | 326 | 37.4% | 257 | 31.4% |
| 2020s (n = 872 starts; 819 current) | Institutes | 154 | 17.7% | 184 | 22.5% |
Universities were the starting sector for 74% of entrants before 1990 (1,183 of 1,593), 77% in the 1990s (923 of 1,204) and 75% in the 2000s (1,103 of 1,467). In the 2010s, the university share of first roles fell from 75% to 56% (967 of 1,717), startups and other companies rose from 10% to 19% (327) and big tech rose from 6% to 13% (220).
In the 2020s (2020–2026), startups and other companies (22.0%, 192 of 872) trailed universities (37.4%, 326) among first recorded roles. Over the same cohort, 13.4% (117) started their first recorded role at a frontier lab, 17.7% (154) at an institute and 16.2% (141) at a big-tech company.
Among the 303 2020–2026 entrants with a known current organization whose first recorded role was at a university, 195 (64%) still work at a university today and 108 (36%) work outside universities: of those 108, 37 (34%) work at an institute, 35 (32%) at a frontier lab, 27 (25%) at a startup or other company and 13 (12%) at a big-tech company (4 people work in two of those sectors). Of the 108 2020–2026 entrants with a known current organization who started at a frontier lab, 89 (82%) still work at a frontier lab, and 99 2020–2026 entrants now at a frontier lab started in another sector.
Organizations employing the most 2020–2026 entrants
Horizontal bars show the 15 organizations employing the most entrants (top 15 with ties) whose first recorded role started in 2020–2026.
Source: People of Superintelligence career records from biographies and OpenAlex, as of 22 October 2026. Organization topics as of 22 October 2026.
Data table
| Organization | Sector | 2020–2026 entrants | Earlier entrants (before 2020) | Dated current people | 2020–2026 entrant share |
|---|---|---|---|---|---|
| Google DeepMind | Frontier labs | 48 | 183 | 231 | 20.8% |
| Anthropic | Frontier labs | 32 | 71 | 103 | 31.1% |
| NVIDIA | Startups and other companies | 29 | 80 | 109 | 26.6% |
| Mistral AI | Frontier labs | 26 | 24 | 50 | 52.0% |
| DeepSeek | Frontier labs | 25 | 15 | 40 | 62.5% |
| Meta | Big tech | 24 | 65 | 89 | 27.0% |
| Vector Institute | Institutes | 23 | 47 | 70 | 32.9% |
| OpenAI | Frontier labs | 21 | 65 | 86 | 24.4% |
| Allen Institute for AI (Ai2) | Institutes | 21 | 33 | 54 | 38.9% |
| Alibaba | Big tech | 20 | 19 | 39 | 51.3% |
| Microsoft | Big tech | 19 | 98 | 117 | 16.2% |
| Cornell University | Universities | 15 | 42 | 57 | 26.3% |
| Mila | Institutes | 13 | 47 | 60 | 21.7% |
| Big tech | 11 | 149 | 160 | 6.9% | |
| Meta Superintelligence Labs | Frontier labs | 11 | 18 | 29 | 37.9% |
Among individual employers, Google DeepMind leads with 48 2020–2026 entrants: examples include Adam Brown and Adrian Bolton at Google DeepMind. Anthropic follows with 32 (Arthur Conmy and Benjamin Mann), and NVIDIA has 29.
Among the 15 largest employers, 2020–2026 entrants make up the largest share of dated current staff at DeepSeek (62.5%, 25 of 40), followed by Mistral AI (52.0%, 26 of 50) and Alibaba (51.3%, 20 of 39). At the other end of the table, 2020–2026 entrants make up 6.9% of dated current staff at Google (11 of 160), and 26.3% (15 of 57) at Cornell University, the only university in the chart. Within the big-tech rows, 6 of Meta's 24 2020–2026 entrants are recorded at FAIR (Meta), 12 of Microsoft's 19 2020–2026 entrants are recorded at Microsoft Research and 2 of Google's 11 2020–2026 entrants are recorded at Google Research. Outside Vector Institute, the next largest institute employer of 2020–2026 entrants is Allen Institute for AI (Ai2) with 21.
Career data comes from the biographies on this site and from OpenAlex author records, as of 22 October 2026. People whose OpenAlex record may be incomplete or mixed are left out of citation figures.