AI Jobster LogoAI Jobster
April 11, 2026
AI Jobster Team

AI Neolabs: 50 Startups Building the Future of Intelligence

AI Neolabs research laboratory

A new class of AI startup has emerged: the neolab. These are research labs usually founded by elite researchers from major AI labs, professors, or post-economic entrepreneurs. They are pre-product, sub-revenue scale, but have raised at high — usually unicorn+ — valuations to focus on long-term breakthroughs. Together, these 50 startups represent roughly $200 billion in total market capitalization and are reshaping the AI landscape.

The Top 50 AI Neolabs

Here is the complete list of the 50 AI neolabs, ranked by valuation, along with the year they were founded, their focus area, founders, and what they're known for:

Tier 1: The Giants ($5B+ Valuation)

  • Thinking Machines Lab ($50–60B, 2024) — Frontier lab founded by Mira Murati, John Shulman, Susmith Chitlala. Ex-OpenAI CTO, PyTorch founder, Tixier. — Open Jobs

  • SSI / Safe Superintelligence ($32B, 2024) — Frontier lab by Ilya Sutskever. OpenAI co-founder, huge valuation, secrecy. — Open Jobs

  • Skild AI (>$14B, 2023) — Robotics foundation models by CMU Professors Deepak Pathak and Abhinav Gupta. — Open Jobs

  • Poolside ($12–14B, 2023) — Coding models by Jason Warner and Eiso Kant. GitHub Copilot creator. — Open Jobs

  • Reflection AI ($8B, 2024) — Coding agents by Misha Laskin and Ioannis Antonoglou. American open-source models by ex-DeepMind team. — Open Jobs

  • Project Prometheus (>$6.2B, 2025) — Industrial AI backed by Jeff Bezos and Vik Bajaj. — Open Jobs

  • Physical Intelligence ($5.6B, 2024) — Robotics foundation models by Stanford Professors Sergey Levine, Karol Hausman, and Lachy Groom. — Open Jobs

Tier 2: The Contenders ($1B–$5B Valuation)

  • Unconventional AI ($4.5B, 2025) — AI compute hardware by Naveen Rao. MosaicML founder. — Open Jobs

  • Humans& ($4B, 2025) — RL for Humans by Eric Zelikman. Ex-X AI researcher and 8th employee of Google. — Open Jobs

  • Inflection AI (~$4B, 2022) — General AI by Mustafa Suleyman and Karen Simonyan. Microsoft 'pseudo-acquiree'. — Open Jobs

  • Isomorphic Labs ($3.5B, 2021) — Bio/AI drug discovery by Demis Hassabis. DeepMind/Alphabet drug discovery spinoff. — Open Jobs

  • AMI Labs (~$3.5B, 2025) — World models by Yann LeCun and Alexandre LeBrun. Ex-Meta Gen AI veteran. — Open Jobs

  • Decart ($3.1B, 2023) — World models by Dean Lotterdorf. Mirage + Oasis world models. — Open Jobs

  • Xaira Therapeutics (~$2.7–4B, 2023) — Bio/AI drug discovery by Marc Tessier-Lavigne and David Baker. $1B committed launch, ARCH/Foresite incubated. — Open Jobs

  • Sakana AI ($2.05B, 2023) — Efficient models by David Ha (Hardmaru), Llion Jones, and Ben Ilo. Japan-focused unicorn, AI for science results. — Open Jobs

  • General Intuition (~$2B, 2025) — World models by Pim de Witte. Massive gameplay-video dataset. — Open Jobs

  • Liquid AI ($2B, 2023) — Efficient models by Daniela Rus. MIT-spinoff working on on-device models. — Open Jobs

  • H (The H Company) ($2B, 2024) — Foundation models & agents by Charles Kantor, Karl Toyli, Laurent Sifre, Daan Wierstra, and Julien Perolat. Ex-DeepMind French AI model co. — Open Jobs

  • Magic ($1.5B, 2022) — Coding agents by Eric Steinberger and Sebastian De Ro. Long-context 'frontier' coding from Europe. — Open Jobs

  • Periodic Labs ($1.5B, 2025) — Bio/AI material discovery by Liam Fedus and Ekin Dogus Cubuk. Ex-OpenAI VP and ex-DeepMind materials lead. — Open Jobs

  • Harmonic ($1.45B, 2023) — Math AI by Vlad Tenev. Ex-Robinhood founder, math superintelligence. — Open Jobs

  • AI21 Labs ($1.4B, 2017) — Enterprise LLMs by Ori Goshen, Yoav Shoham, and Amnon Shashua. Israeli team incl. Stanford professor. — Open Jobs

  • Lila Sciences (>$1.3B, 2023) — Bio/AI lab automation by Geoffrey von Maltzahn. Huge Series A. — Open Jobs

  • Chai Discovery ($1.3B, 2024) — Bio/AI drug discovery by Josh Meier. OpenAI-backed biotech unicorn. — Open Jobs

  • Flapping Airplanes (>$1B, 2025) — Frontier lab by Ben Spector and Asher Spector. Post people, talented young team. — Open Jobs

  • Recursive (>$1B, 2025) — AI research by Richard Socher. Richard Socher from You.com. — Open Jobs

  • World Labs (>$1B, 2024) — World models by Fei-Fei Li. Stanford AI Professor building world models. — Open Jobs

  • EvolutionaryScale (>$1B, 2024) — Bio/AI drug discovery by ex-Meta research team. ESM3. — Open Jobs

  • AAI (>$1B, 2023) — Frontier lab by Amnon Shashua and Shai Shalev-Shwartz. Stealth lab, Lightspeed-led round. — Open Jobs

  • Kyutai (>$1B, 2023) — Voice AI by Xavier Niel and Rodolphe Saadé. Eric Schmidt backed EU co with ~$330M budget. — Open Jobs

  • Goodfire ($1B, 2024) — Interpretability by Eric Ho and Tom McGrath. Interpretability researchers from all top labs. — Open Jobs

  • Imbue ($1B, 2020) — Reasoning agents by Kanjun Qiu and Josh Albrecht. Early AI lab unicorn. — Open Jobs

  • Reka ($1B, 2022) — Multimodal AI by Yi Tay, Cyprien de Masson d'Autume, and Dani Yogatama. Ex-Google team. — Open Jobs

  • Essential AI ($1B, 2023) — Open-source LLMs by Ashish Vaswani and Niki Parmar. Transformer-author founders, Ramanujam. — Open Jobs

  • Zyphra ($1B, 2021) — Open-source LLMs by Kithik Pullaiah and Danny Martinoli. Fantastic small model results, Mara. — Open Jobs

  • Nous Research ($1B, 2022) — Open-source LLMs by Jeffrey Quesnelle and Karan Malhotra. Decentralized training and open source, Psyche. — Open Jobs

  • Aaru ($1B, 2024) — Simulation by Cameron Fink, Neal Koh, and John Kissler. 'Headline' $1B Series A. — Open Jobs

  • Simile ($1B, 2025) — Simulation by Joan Park, Michael Bernstein, and Percy Liang. Ex-Stanford researcher and professors. — Open Jobs

  • Isara ($1B, 2025) — Frameworks by Edoie Zhang. Ex-OpenAI team with SOTA results. — Open Jobs

  • Moonvalley ($1B, 2024) — Video AI by Naveen Tatavolu, Mateusz Malinowski, and Mik Binkwski. Licensed-data video model Marey for filmmakers. — Open Jobs

  • Hark ($1B, 2025) — Continual learning by Brett Adcock. Ex-Figure AI founder uses $100M personal funds. — Open Jobs

Tier 3: Rising Stars ($300M–$1B Valuation)

  • Prime Intellect (<$1B, 2024) — Decentralized AI training by Vincent Weisser and Johannes Hagemann. Distributed model training, inference and RL. — Open Jobs

  • Ndea (<$1B, 2025) — Program synthesis by François Chollet and Mike Knoop. Keras founder and Zapier founder. — Open Jobs

  • Inception Labs (~$500M, 2024) — Diffusion LLMs by Stefano Ermon. Stanford Professors working on Diffusion LLMs. — Open Jobs

  • Adaption Labs (~$500M, 2025) — Continual learning by Sara Hooker and Sudip Roy. Ex-Cohere execs. — Open Jobs

  • Elorian (~$500M, 2025) — Multimodal AI by Andrew Dai and Yinhe Yang. Ex-DeepMind execs, vision reasoning. — Open Jobs

  • Genesis AI (~$500M, 2024) — Robotics by Zhou Xian and Théophile Gervet. French ex-Mistral / Skild researcher, EU robotics. — Open Jobs

  • CuspAI (~$500M, 2024) — Bio/AI material discovery by Chad Edwards and Max Welling. AI for chemistry. — Open Jobs

  • Poetiq (<$500M, 2025) — Frameworks by Shuneet Bhanga and Ian Fischer. Ex-DeepMind team with SOTA ARC-AGI results. — Open Jobs

  • Axiom Math (~$300M, 2025) — Math AI by Carina Hong. Solving IMO Putnam 2025 problems. — Open Jobs

What LLMs and Models Are They Building?

While each neolab has its own research agenda, the models being developed can be broadly categorized into several themes:

Frontier Foundation Models

Labs like Thinking Machines Lab, SSI, and Flapping Airplanes are working on next-generation frontier models that push the boundaries of general intelligence. SSI, led by Ilya Sutskever (OpenAI co-founder), is famously secretive about its approach but has attracted $32B in valuation on the promise of safe superintelligence. Thinking Machines Lab, founded by ex-OpenAI CTO Mira Murati, is the highest-valued neolab at $50–60B.

Coding-Focused LLMs

Poolside and Reflection AI are building specialized large language models optimized for software engineering. Poolside, created by GitHub Copilot creator Jason Warner, is developing coding models valued at $12–14B. Reflection AI focuses on open-source coding agents, while Magic is pursuing long-context frontier coding from Europe.

World Models and Multimodal AI

AMI Labs (co-founded by Yann LeCun), Decart, General Intuition, and World Labs (founded by Fei-Fei Li) are all building 'world models' — AI systems that develop an internal understanding of how the physical world works. These models go beyond text and image to simulate and predict real-world physics, spatial reasoning, and embodied intelligence.

Efficient and Open-Source Models

Sakana AI, Liquid AI, Essential AI, Zyphra, and Nous Research are focused on making powerful AI more accessible. Sakana AI is building efficient models for science applications from Japan. Liquid AI (MIT spinoff) is developing on-device models. Essential AI, founded by Transformer paper co-authors Ashish Vaswani and Niki Parmar, along with Zyphra and Nous Research, are building competitive open-source LLMs.

Bio/AI and Drug Discovery Models

Isomorphic Labs (DeepMind spinoff led by Demis Hassabis), Xaira Therapeutics, Chai Discovery, Periodic Labs, and EvolutionaryScale are training specialized models for biological research. EvolutionaryScale's ESM3 is a notable protein language model. These labs are at the intersection of AI and biotechnology, using large-scale models to accelerate drug discovery and materials science.

Reasoning and Math Models

Harmonic (founded by ex-Robinhood founder Vlad Tenev) and Axiom Math are pushing the frontier of mathematical reasoning in AI. Harmonic is working toward math superintelligence, while Axiom Math is tackling competitive math problems at the IMO and Putnam level.

Notable Omissions

Several major players are intentionally excluded from the neolab list because they have crossed into different categories:

  • OpenAI, Anthropic, xAI, AI2 — classified as 'big labs' or 'non-profits'

  • Mistral, Cohere — classified as 'enterprise / revenue scale'

  • Stability, Luma, Runway, Pika — classified as 'video / revenue scale' or 'image / revenue scale'

  • Cartesia, Black Forest Labs — classified as 'video / revenue scale' or 'image / revenue scale'

  • Modular, LMArena — classified as 'infra / not model' or 'benchmark / not model'

  • Silicon hardware cos, Robotics hardware cos — classified as 'hardware / not model'

What This Means for AI Job Seekers

The rise of neolabs is creating unprecedented demand for AI talent. These 50 companies are actively hiring researchers, engineers, and applied scientists across machine learning, robotics, biology, mathematics, and infrastructure. From frontier model training to MLOps, from AI safety to drug discovery — the career opportunities span an incredible range of disciplines.

If you're looking to work at the cutting edge of AI, many of these neolabs are hiring right now. The skills most in demand include deep learning research, distributed systems engineering, reinforcement learning, and domain expertise in areas like biology, chemistry, and mathematics.

Find Your Next AI Role on AI Jobster

Ready to join the AI revolution? AI Jobster aggregates open roles from top AI companies, startups, and research labs — including many of the neolabs listed above. Whether you're a seasoned ML researcher or an engineer looking to break into AI, browse hundreds of curated positions across machine learning, data science, AI engineering, robotics, and more.

Visit aijobster.work to explore open roles, upload your resume for AI-powered job matching, and set your job preferences to get personalized recommendations. The future of AI is being built right now — and you could be part of it.