Genesis AI

Genesis AI is a robotics-AI foundation-model company founded in late 2024 by Zhou Xian (Carnegie Mellon University robotics PhD) and Théophile Gervet (former research scientist at Mistral AI), with offices in Paris and San Francisco. The company is building a universal robotics foundation model and a horizontal platform for general-purpose physical AI, and in May 2026 announced a full-stack expansion into custom robotic-hand hardware. Genesis AI emerged from stealth in July 2025 with $105 million in seed financing co-led by Eclipse Ventures and Khosla Ventures, with Bpifrance, HSG, Eric Schmidt, Xavier Niel, Daniela Rus, and Vladlen Koltun participating.

At a glance

• Founded: Late 2024, by Zhou Xian and Théophile Gervet. Offices in Paris and San Francisco.
• Status: Private. Seed round in July 2025.
• Funding: $105 million seed financing co-led by Eclipse Ventures and Khosla Ventures with Bpifrance, HSG, Eric Schmidt, Xavier Niel, Daniela Rus, and Vladlen Koltun participating. Subsequent rounds reportedly in process at higher valuations.
• CEO: Co-founder structure with Zhou Xian and Théophile Gervet at the executive helm.
• Other notable leadership: Théophile Gervet (Co-Founder; former Mistral AI research scientist), Zhou Xian (Co-Founder; CMU robotics PhD).
• Open weights: Limited. Genesis AI develops commercial robotics foundation models and synthetic-data infrastructure rather than open-weights model releases.
• Flagship outputs: Universal robotics foundation model (in development); horizontal platform for general-purpose physical AI; custom human-shaped robotic hands announced May 2026.

Origins

Genesis AI was founded in late 2024 by Zhou Xian, who completed a PhD in robotics at Carnegie Mellon University, and Théophile Gervet, a former research scientist at Mistral AI. The founder pair combines academic robotics-AI research credibility (Zhou's CMU robotics-PhD work) with frontier-foundation-model engineering experience (Gervet's Mistral tenure), giving the company a distinctive cross-Atlantic founding profile that few robotics-AI insurgents have matched.

The early product premise was to create high-fidelity synthetic data for training robotics foundation models, with the company building a platform to simulate realistic 3D worlds where objects move and interact in physically accurate ways. The simulation platform was positioned as the data-generation layer that frontier robotics-AI foundation models would consume, with the model layer building on top. Industry coverage at the time of stealth exit emphasized the simulation-and-synthetic-data thesis as distinctive within the robotics-AI insurgent cohort.

The July 2025 stealth-exit and $105 million seed financing was co-led by Eclipse Ventures and Khosla Ventures, with Bpifrance (the French sovereign-investment vehicle), HSG, Eric Schmidt, Xavier Niel, Daniela Rus (the senior MIT robotics professor), and Vladlen Koltun participating. The transatlantic investor cohort, including a French sovereign vehicle and senior US venture capital, reflected the company's Paris-and-San-Francisco operating structure.

In May 2026, TechCrunch and adjacent industry outlets reported that Genesis AI had moved to a full-stack approach, building its own custom robotic hands with the same size and shape as a human hand rather than the two-finger grippers commonly used by other robotics-AI labs. The full-stack expansion was characterized as a recognition that controlling the hardware was necessary to close the gap between simulation and real-world deployment, and put Genesis AI in a different competitive posture relative to the foundation-model-only positioning of Skild AI and Physical Intelligence.

Mission and strategy

Genesis AI's stated mission is to build a universal robotics foundation model and a horizontal platform for general-purpose physical AI. The strategic premise reflects a synthesis of the foundation-model-only thesis pursued by Skild AI and Physical Intelligence and the vertically-integrated-robotics-and-AI thesis pursued by humanoid-robot-and-AI manufacturers.

The strategy combines three threads. First, the simulation-and-synthetic-data platform that generates realistic 3D-world training data for robotics foundation models. Second, the universal robotics foundation model layer trained on the synthetic data and positioned as a horizontal platform for diverse robot hardware. Third, the full-stack robotic-hand hardware announced in May 2026, which gives Genesis AI direct control over the most capability-constrained part of the robot-platform stack while keeping the foundation-model layer horizontal.

The competitive premise positions Genesis AI within the robotics-AI foundation-model insurgent cohort alongside Physical Intelligence, Skild AI, and Rhoda AI, with the simulation-data, universal-foundation-model, and full-stack-hand-hardware combination providing a distinctive multi-layer positioning relative to peers.

Models and products

• Universal robotics foundation model. Genesis AI's flagship model layer, in development. Positioned as a horizontal platform for general-purpose physical AI.
• Simulation-and-synthetic-data platform. Realistic 3D-world simulation for generating physics-accurate training data for robotics foundation models.
• Custom human-shaped robotic hands. Announced May 2026. Same size and shape as a human hand, providing a closer match to real-world conditions than two-finger gripper alternatives.
• Full-stack robotic systems. Stated forward direction following the May 2026 hardware expansion.

Distribution channels are positioned as horizontal foundation-model licensing to hardware partners alongside Genesis AI's own full-stack robotic systems. Specific named hardware partners have not been disclosed publicly.

Benchmarks and standing

Genesis AI's evaluation framework focuses on robot-task-completion metrics across the synthetic-and-real-world data surface rather than horizontal foundation-model leaderboards. Public capability disclosures have been limited to demonstration videos, with the May 2026 full-stack-hand-hardware announcement positioning the company toward more concrete capability demonstrations.

Industry coverage has consistently grouped Genesis AI with the principal robotics-AI foundation-model insurgent cohort, with the founder pair's combined CMU-and-Mistral credibility providing distinctive provenance. The Paris-and-San-Francisco operating structure and the Bpifrance sovereign-investment participation gives Genesis AI a transatlantic positioning that is distinctive within the otherwise US-and-China-dominated robotics-AI insurgent segment.

Leadership

As of May 2026, Genesis AI's senior leadership includes:

• Zhou Xian, Co-Founder. Carnegie Mellon University robotics PhD.
• Théophile Gervet, Co-Founder. Former Mistral AI research scientist.
• Senior research and engineering leadership across the foundation-model, simulation-platform, and robotic-hand programs.

Continued senior research and engineering recruitment has supported the universal robotics foundation model program and the May 2026 hardware-platform expansion.

Funding and backers

• Seed (July 2025): $105 million co-led by Eclipse Ventures and Khosla Ventures with Bpifrance, HSG, Eric Schmidt, Xavier Niel, Daniela Rus, and Vladlen Koltun participating.
• Subsequent rounds: Industry coverage has indicated additional fundraising activity in 2026 at higher valuations.

Cumulative disclosed private capital approximately $105 million as of stealth-exit; subsequent rounds reportedly in process.

The investor cohort is distinctive in three respects. First, the Eclipse-and-Khosla co-lead structure brings two of the senior Silicon Valley AI-and-deep-tech investors, with Eclipse having a particular concentration in industrial-and-physical-AI ventures and Khosla holding Series A positions in Skild AI and adjacent robotics-AI insurgents. Second, the Bpifrance participation gives Genesis AI a French sovereign-investment relationship that few US-and-China-headquartered robotics-AI insurgents have matched, anchoring the company in the broader French AI ecosystem alongside Mistral AI and the broader Paris-based AI scene. Third, the participation of senior individual investors including Eric Schmidt (former Google Chief Executive), Xavier Niel (the Iliad founder and a senior French tech investor), Daniela Rus (the senior MIT robotics professor), and Vladlen Koltun gives the round an unusual research-and-industry-credibility-anchored profile relative to peer robotics-AI rounds.

Industry position

Genesis AI occupies a distinctive position within the robotics-AI foundation-model insurgent cohort, with the simulation-and-synthetic-data platform thesis, the universal-foundation-model layer, and the May 2026 full-stack-hand-hardware expansion combining into a multi-layer positioning that few peers have replicated. The Paris-and-San-Francisco operating structure and the transatlantic investor cohort gives the company a distinctive geographic profile.

The structural risks are several. First, the simulation-to-real-world generalization gap has been a persistent technical challenge in robotics, and whether Genesis AI's synthetic-data approach closes the gap better than peer approaches relying on real-world teleoperation data is unproven at production scale. Second, the May 2026 full-stack-hand-hardware expansion adds hardware-development complexity to the foundation-model and simulation-platform programs, and the multi-layer execution scope is wider than the foundation-model-only positioning of Physical Intelligence and Skild AI. Third, the seed-round capital base is materially smaller than the Series-A-and-beyond capital bases of Physical Intelligence ($400 million Series A), Skild AI ($300 million Series A), and Rhoda AI ($450 million Series A), which may constrain the multi-layer execution scope.

Competitive landscape

• Physical Intelligence, Skild AI. Direct robotics-AI foundation-model peers with foundation-model-only positioning.
• Rhoda AI. 2026-cohort robotics-AI foundation-model peer with video-predictive-control technical bet.
• 1X, Figure AI, Boston Dynamics, Tesla Optimus. Humanoid-robotics hardware peers; the May 2026 full-stack-hand expansion puts Genesis AI in adjacent hardware territory.
• Mistral AI. Co-founder Gervet's prior employer; not directly competitive but a comparable insurgent foundation-model peer in the European frontier-AI segment.
• Google DeepMind (RT-2, Gemini Robotics), NVIDIA Research (GR00T). Frontier-lab robotics-AI research programs.

Outlook

• The pace of universal-robotics-foundation-model technical disclosures and capability demonstrations.
• The execution of the May 2026 full-stack-hand hardware platform and integration with the foundation-model layer.
• The Series A or adjacent fundraising timeline given the comparatively recent seed round and the multi-layer execution scope.
• Continued senior research and engineering recruitment across the Paris and San Francisco offices.
• Whether the simulation-and-synthetic-data approach delivers measurable advantages over peer approaches relying on real-world teleoperation data.

Sources

• TechCrunch: Genesis AI launches with $105M seed funding. July 2025 stealth-exit reference.
• TechCrunch: Khosla-backed robotics startup Genesis AI has gone full stack. May 2026 full-stack hardware expansion.
• Sifted: Ex-Mistral employee raises $105m for AI and robotics startup. Founder background and Paris office reference.
• PR Newswire: Genesis AI Emerges From Stealth. Seed-round announcement.
• Genesis AI official site. Company reference.
• The Robot Report: Genesis AI brings in $105M to build universal robotics foundation model. Industry coverage.