Bedrijven GIDS

Skild AI en Robotics Foundation-modellen

Skild AI bouwt een universeel robotica-basismodel, genaamd Skild Brain, zodat verschillende robots één aangeleerd besturingssysteem kunnen delen voor alle taken en hardware.

Luister naar deze gidsAI-verteld · ~8 min

Op deze pagina9 minuten lezen
  1. Belangrijkste inzichten
  2. Diepe duik
  3. Tracing how an omni-bodied brain recovers from a locked knee
  4. Implementatie in de echte wereld
  5. De toekomst van Skild AI en Robotics Foundation-modellen
  6. Het in de praktijk brengen
  7. Bronnen en verdere literatuur
  8. Blijf verkennen
  9. Veelgestelde vragen

Belangrijkste inzichten

  1. Skild AI was founded in Pittsburgh in 2023 by former Carnegie Mellon professors Deepak Pathak and Abhinav Gupta to build one robot brain for many bodies.
  2. The Skild Brain pairs a slow high-level policy for task decisions with a fast low-level policy that outputs joint angles and motor torques.
  3. Skild AI trained one model across about 100,000 simulated robot bodies and reported zero-shot adaptation to lost limbs, locked knees and jammed wheels.
  4. Its S1 model learns new manipulation tasks from one video demonstration and reported about 66% success on unseen tasks versus 9% for a language-prompted baseline.
  5. Skild AI raised $1.4 billion at a valuation above $14 billion in January 2026 and reported over $100 million in recurring revenue that September.
  6. Nearly all published Skild AI results come from company blog posts rather than peer-reviewed papers or shared public benchmarks.

Diepe duik

Skild AI is a robotics software company that sells one thing: a robot brain. Instead of writing a separate controller for each machine and each task, it trains a single large neural network, the Skild Brain, intended to run quadrupeds, humanoids, tabletop arms and mobile manipulators. In AI terms this is a robot foundation model: a model pretrained on broad data that can then be adapted, prompted or fine-tuned for many downstream jobs, the way large language models are for text. The company was started in Pittsburgh in 2023 by Deepak Pathak and Abhinav Gupta, both former Carnegie Mellon University professors, as Maginative reported when its first major funding round was announced.

The core problem Skild AI targets is data. Language models learn from trillions of words of internet text, but there is no comparable archive of robot actions. Collecting real robot data means people teleoperating machines one demonstration at a time, which is slow and expensive. Skild AI's July 2025 post argues that many existing robotics foundation models start from a vision-language model, a model trained on images and text, and add a small share of robot data, which it puts at under 1%. Its view is that such models understand objects and words but lack action information and physical common sense, so they mainly show semantic generalization on pick-and-place tasks.

The company's answer is a three-part data recipe. Pretraining uses large-scale simulation and internet video. Post-training adds targeted real-world robot data. Human video is treated as data from one more kind of body, since a person picking up a cup is solving the same physical problem a robot arm faces. In January 2026 Skild AI said it could fine-tune its model to perform new skills by watching videos alone, using less than one hour of robot data. The design is hierarchical: a high-level policy runs at low frequency and decides manipulation and navigation actions, while a low-level policy runs at high frequency and turns those commands into joint angles and motor torques. The deployed model is fine-tuned and distilled, meaning compressed into a smaller network, for each setting.

The defining claim is that the brain is omni-bodied: it does not need to be told what body it is in. In September 2025 Skild AI described training one model across a simulated universe of 100,000 different robots for the equivalent of roughly a thousand years of simulated time. Test robots were held out of training, so every result was zero-shot. When engineers removed a quadruped's lower legs, it adapted in about seven to eight seconds by swinging its thigh joints, while a specialist controller trained for that one robot flipped over. With knees locked in software it walked on three legs within two to three seconds, and with jammed wheels it switched from rolling to walking. The company compares this to in-context learning in language models: the model infers its situation from recent history rather than from new training.

The milestones came quickly. Skild AI announced a $300 million Series A at a $1.5 billion valuation on 9 July 2024, led by Lightspeed Venture Partners, Coatue, SoftBank Group and Bezos Expeditions. It publicly introduced the Skild Brain on 29 July 2025. On 14 January 2026 it announced a $1.4 billion Series C at a valuation above $14 billion, led by SoftBank with NVentures, the venture arm of NVIDIA, among the participants, and said revenue had grown from zero to about $30 million within a few months of 2025. In August 2026 it released S1, a manipulation model that learns tasks from a single video demonstration, and in September 2026 it reported crossing $100 million in annual recurring revenue about ten months after its first commercial deployment, with more than 60 paying customers.

S1 shows how the approach compares with the main alternative, a vision-language-action (VLA) model prompted with text instructions. In Skild AI's own tests after 100,000 hours of pretraining, S1 succeeded on about 66% of unseen tasks when shown one demonstration, against 9% for a language-prompted VLA policy. On seen tasks it reached about 96%. The company estimates that one in-context demonstration is worth about 380 conventional post-training demonstrations, which would take 50 to 100 hours of teleoperation to collect. The comparison against classic robotics is starker still: a hand-engineered controller is tuned for one machine and usually fails when the machine changes, which is exactly the case the omni-bodied tests probe.

The limitations are significant. Almost all published evidence comes from Skild AI's own blog posts and showcase videos, not from peer-reviewed papers or shared public benchmarks, and the September 2025 post discloses no architecture, model size or compute budget. Results such as the 66% unseen-task rate depend on the company's chosen tasks and baselines. Real deployments remain hard: the founders wrote that teaching the model to cook an egg took a week for the first egg and two months before it worked reliably across different eggs and setups. Robot soccer trained by self-play, shown in September 2026, is still far below human level by the company's own account.

In practice, using the Skild Brain is a commercial relationship rather than a download. There are no open weights or public model cards; customers deploy it on robots the company supports and integrates. About 90% of its reported revenue comes from manipulation work and about 10% from mobility, a share that includes autonomous mobile robots from its acquisition of Zebra Technologies' robotics business. Buyers should expect integration work, on-site tuning and the safety engineering any industrial robot requires.

03Uitgewerkt voorbeeld

Tracing how an omni-bodied brain recovers from a locked knee

  1. Start with a quadruped walking normally; the model has never seen this exact robot, so it is inferring the body from its recent observations and actions.

  2. Lock one knee in software, as in Skild AI's September 2025 test, so commanded torques on that joint no longer move the leg.

  3. Note what the low-level policy now observes: commanded and measured joint angles disagree, and body tilt and foot contacts change.

  4. Because this new history sits in the model's context, its estimate of the body shifts toward a three-legged robot without any weight update.

  5. The policy redistributes weight and changes step timing; Skild AI reports stable walking after about two to three seconds.

  6. Compare with a specialist controller tuned for the intact robot, which has no mechanism to update its assumptions and typically falls.

Wat het laat zien

A step-by-step account of in-context adaptation for a robot body. To check understanding, predict what should happen when the knee is unlocked: the history changes again, and the model should return to a four-legged gait.

Implementatie in de echte wereld

Dual-arm manipulators running the Skild Brain to assemble NVIDIA Blackwell systems with NVIDIA and Foxconn, a deployment Skild AI described in its September 2026 revenue update.

Sumitomo Wiring Systems working to automate wire harness manufacturing with S1, Skild AI's in-context manipulation model, according to the company's September 2026 post.

Mitsui and Co. piloting S1-powered robots in commercial kitchens, one of more than 60 paying customers Skild AI reported across warehouses, factories, data centers, security and delivery.

A quadruped controlled by the omni-bodied brain that kept walking on three legs after its knees were locked in software, adapting in about two to three seconds in Skild AI's September 2025 tests.

Teaching S1 an unseen task such as pour-over coffee or plant potting from a single video demonstration, with no fine-tuning or change to the model's weights.

De toekomst van Skild AI en Robotics Foundation-modellen

The open problem for Skild AI and its peers is evidence. As of late 2026 the strongest claims, from 100,000-body training to 66% success on unseen tasks, are company-reported, and the field still lacks agreed public benchmarks for general-purpose robot brains, so outside researchers cannot easily compare approaches. A second question is how far human video can substitute for robot data. The January 2026 post claims new skills from video plus under an hour of robot data, but the egg-cooking account in September 2026 shows that reliability across real-world variation still takes months of work. A third is whether learning can exceed human skill: the self-play work argues that models pretrained on human data are capped near human ability, and the company has said it plans to try self-play on construction, factory and household tasks. Commercially, the roughly $100 million in recurring revenue reported in September 2026 suggests real demand, but it is too early to know whether a single omni-bodied model will outperform specialized systems on cost, safety certification and uptime in the long run.

Het in de praktijk brengen

Waarom het ertoe doet

  • Leveranciersstrategie. Roadmaps van leveranciers beïnvloeden welke functies uw team vervolgens kan bouwen.
  • Kosten en budget. Commerciële voorwaarden en implementatieopties zijn van invloed op de kosten en risico's op de lange termijn.
  • Risico en veiligheid. Bedrijfsprikkels bepalen productgebreken, veiligheidshouding en openheid.

Pas op voor

  • Lanceringsaankondigingen kunnen de stabiliteit in echte productieworkflows overtreffen.
  • API-prijzen of beleidswijzigingen kunnen van de ene op de andere dag de aannames doorbreken.
  • De afhankelijkheid van één leverancier verhoogt de lock-in- en migratiekosten.

Volgende stappen

  1. Evalueer providers met behulp van uw eigen taken en datasets.
  2. Controleer de privacy-, beveiligings- en juridische voorwaarden vóór de integratie.
  3. Onderhoud een noodplan voor alle modellen of leveranciers.
  4. Houd de release-opmerkingen in de gaten, zodat wijzigingen in de routekaart teams niet verrassen.

Bronnen en verdere literatuur

  1. Skild AIBuilding the general-purpose robotic brain
  2. Skild AIThe case for an omni-bodied robot brain
  3. Skild AIIntroducing S1: In-Context Learning for Robotics
  4. Skild AIAnnouncing Series C
  5. Skild AIThe Hidden Pillar of Robotics
  6. Skild AIPhysical Self-Play
  7. Skild AILearning by watching human videos
  8. MaginativeSkild.ai Raises $300M Series A with $1.5B Valuation

Blijf verkennen

Gratis nieuwsbrief

Ontvang de dagelijkse AI-briefing

Drie verifieerde AI-verhalen elke weekdagmorgen, geschreven in eenvoudig Engels. Gratis voor altijd, geen advertenties.

Eén e-mail elke weekdag. Afmelden met één klik. We verkopen of delen je adres nooit.

Test jezelf

Take the Skild AI and Robotics Foundation Models quiz

Directe feedback op elke antwoord, en een deelbare certificaat met een verifieerbare ID zodra je een cursus hebt afgerond.

Quiz starten

Ondersteun gratis AI-onderwijs. AI Understanding is een 501(c)(3) non-profit — geen advertenties, geen betaalmuur, nooit. Doe een donatie

Veelgestelde vragen

What is Skild AI and Robotics Foundation Models?

Skild AI bouwt een universeel robotica-basismodel, genaamd Skild Brain, zodat verschillende robots één aangeleerd besturingssysteem kunnen delen voor alle taken en hardware.

Wat is het kernproduct dat Skild AI bouwt?

Skild AI bouwt de Skild Brain, een enkel algemeen basismodel dat bedoeld is om veel verschillende robots te besturen.

Van welke universiteit kwamen de oprichters van Skild AI?

Skild werd mede opgericht door Carnegie Mellon-professoren Deepak Pathak en Abhinav Gupta.

Wat betekent het dat Skilds model 'embodiment-agnostisch' is?

Uitvoeringsonafhankelijk betekent dat hetzelfde model in veel verschillende robotlichamen kan worden ingezet.

Which individual investor participated in Skild AI’s 2024 Series A round?

Skild AI’s 2024 company announcement lists Jeff Bezos through Bezos Expeditions among the Series A participants.

Which task areas did Skild AI list for its general-purpose robot model?

Skild AI’s company announcement describes tasks including manipulation, locomotion, and navigation across robot embodiments.