Skild AI — the general-purpose robot foundation model
Skild AI is a Pittsburgh-headquartered robot foundation model company building Skild Brain — a single, omni-bodied AI model designed to run across humanoids, quadrupeds, tabletop arms, and mobile manipulators. Spun out of the Carnegie Mellon University Robotics Institute in 2023, Skild is one of the most highly capitalized players in the current humanoid robotics and general-purpose robot AI boom, alongside Figure AI, Tesla Optimus, 1X, Physical Intelligence, and Boston Dynamics.
- 2023Founded
- Pittsburgh, PAHeadquarters
- Skild BrainFoundation model (2025)
- $2B+ raisedFunding raised
- USACountry
- PrivateStatus
About Skild AI
Skild AI, Inc. was founded in 2023 in Pittsburgh, Pennsylvania by Deepak Pathak and Abhinav Gupta, both faculty members of Carnegie Mellon University's Robotics Institute and School of Computer Science. Pathak — the Raj Reddy Associate Professor at CMU — serves as chief executive; Gupta serves as president. The two co-founders had previously worked together at Meta's Fundamental AI Research (FAIR) lab, and share a decade-long research background in self-supervised learning, embodied intelligence, and robot manipulation. Rather than building a humanoid robot themselves, Pathak and Gupta set out to build the brain — a general-purpose robot foundation model that any humanoid or other embodiment could license and run.
The company's trajectory has been one of the fastest in the current AI cycle. Skild AI raised a roughly $14.5 million seed round in late 2023 co-led by Lightspeed Venture Partners and Sequoia Capital, then formally emerged from stealth in July 2024 with a $300 million Series A at a $1.5 billion valuation — a round led by Lightspeed, Coatue, SoftBank Group, and Jeff Bezos through Bezos Expeditions, with Sequoia, Felicis, Menlo Ventures, General Catalyst, CRV, Amazon, SV Angel, and Carnegie Mellon University also participating. The company disclosed the round specifically to finance training data acquisition, model scaling, and aggressive hiring across AI, robotics, and infrastructure.
Growth accelerated from there. In 2025 Skild AI raised a Series B reported at approximately $500 million at a valuation of roughly $4.5 billion, with SoftBank leading and NVIDIA, Samsung, and LG Technology Ventures joining as strategic investors. In January 2026 the company then closed a Series C of close to $1.4 billion, again led by SoftBank Group, with participation from NVentures (NVIDIA's venture arm), Macquarie Capital, Bezos Expeditions, Disruptive, 1789 Capital, and existing investors Lightspeed, Felicis, Coatue, and Sequoia. Strategic investors in the round included Samsung, LG, Schneider Electric, CommonSpirit, and Salesforce Ventures. The Series C brought Skild's valuation above $14 billion and total funding raised to more than $2 billion — making Skild one of the most highly capitalized AI-first robotics companies in the world, and a rare software-only player in the humanoid stack that sits alongside hardware OEMs on comparable valuations.
Skild AI's competitive position is deliberately different from a Figure, a Tesla Optimus, or a 1X: instead of building both the robot and the brain, Skild positions itself as the foundation-model layer that any robot maker can license. That places Skild alongside a small handful of pure-play robot-AI companies — Physical Intelligence (π), Google DeepMind's Gemini Robotics work, and NVIDIA's GR00T program among them — as the software-first side of the general-purpose robotics AI boom.
Product lineup: Skild Brain and the omni-bodied model family
Unlike a humanoid OEM, Skild AI does not sell a robot. Its product is a single robot foundation model — Skild Brain — that runs across many embodiments, packaged as a licensable software stack for hardware partners. What follows is Skild's publicly disclosed model lineup and how each generation fits its "any robot, any task, one brain" thesis.
Skild Brain (2025) — the omni-bodied foundation model
Introduced in July 2025 as Skild's first named public product, Skild Brain is a large-scale general-purpose robot foundation model. It uses a hierarchical architecture: a low-frequency, high-level manipulation-and-navigation policy that plans and reasons, feeding a high-frequency low-level action policy that closes the loop on the robot's motors. The same trained model works on quadrupeds, humanoids, tabletop arms, mobile manipulators, and legged industrial platforms — a property Skild calls "omni-bodied intelligence."
Skild Brain on partner humanoids (2025–2026)
Skild's most visible go-to-market path is licensing Skild Brain to humanoid and industrial-robot manufacturers. In 2026 the company publicly announced deployments and partnerships with ABB Robotics and Universal Robots, and a broader technology partnership with NVIDIA for training and onboard inference. Skild has also demonstrated the same model driving legged robots, wheeled bases, and dexterous humanoid hands in first-party video.
Research and demonstration platforms
Skild AI has shown internal demonstrations on off-the-shelf quadruped platforms, mobile bases, and third-party humanoid hardware. These are not products — they are the reference embodiments on which Skild Brain is trained and evaluated before being ported to partner platforms.
| Model | Year | Type | Embodiments | Compute target | Runtime | Architecture | Notable |
|---|---|---|---|---|---|---|---|
| Skild Brain | 2025 | Robot foundation model | Humanoids, quadrupeds, arms, mobile | NVIDIA-class GPU (edge + cloud) | Onboard inference | Hierarchical VLA (2-tier) | Omni-bodied; "one brain, any robot" |
| Skild Brain + ABB | 2026 | Deployed model | Industrial arms / mobile | Edge GPU | Onboard inference | Hierarchical VLA | ABB Robotics partnership |
| Skild Brain + UR | 2026 | Deployed model | Collaborative arms | Edge GPU | Onboard inference | Hierarchical VLA | Universal Robots partnership |
Technology stack
Skild AI is software-first — its "product" is a trained model plus the runtime that deploys it onto a partner's hardware. But that runtime touches nearly every subsystem of a modern humanoid robot: perception, planning, control, power delivery, and safety. The hardware constraints of the platforms Skild Brain runs on shape the model as much as the training data does.
Perception
Skild Brain consumes multi-camera vision plus proprioceptive state (joint positions, velocities, torques) and, where the hardware supports it, force-torque and tactile signals from the manipulator. This mix of vision and touch is what allows the model to perform contact-rich tasks — inserting parts, gripping deformable objects, and reacting to unexpected contact — across humanoids, quadrupeds, and industrial arms without per-embodiment retraining.
The Skild Brain foundation model
Skild Brain is publicly described as a hierarchical Vision-Language-Action (VLA) system. The high-level policy runs at low frequency and handles task decomposition, manipulation, and navigation planning conditioned on natural-language instructions. The low-level policy runs at high frequency and issues actuator-level commands to the robot's motors, with the two policies co-trained on a large, diverse corpus of robot experience. Because the same model handles many embodiments, training scales in a way that per-robot policies cannot.
Actuators, power, and compute
Although Skild AI does not build hardware, Skild Brain has to reason about it. The model is trained and deployed on partner robots that overwhelmingly use electric brushless DC (BLDC) actuators — no hydraulics — driven by high-torque motor controllers, sensored with high-resolution encoders and IMUs, and powered by lithium-ion battery packs targeting multiple hours of continuous operation. Onboard compute is dominated by NVIDIA-class GPU accelerators, running Skild Brain's low-level policy at high control rates entirely on the robot.
Power management is a particularly demanding subsystem for the platforms Skild Brain deploys on. Each joint's servo controller must deliver hundreds of watts of instantaneous torque without thermally throttling, which drives the use of high-efficiency wide-bandgap semiconductors (SiC and GaN) in the motor-driver stages and requires tightly matched precision passives on every gate-drive and current-sense channel. Onboard battery packs integrate cell balancing, thermal management, and multi-stage protection ICs — subsystems whose reliability is often the practical ceiling on how long a humanoid platform can run Skild Brain between charges.
Applications and commercial deployments
Skild AI's go-to-market is the humanoid and industrial-robotics stack — licensing Skild Brain to OEMs and integrators rather than selling directly to end-customers. Publicly announced partners, pilots, and target markets include:
- ABB Robotics. A publicly announced 2026 partnership brings Skild Brain to ABB's industrial and mobile-robot platforms, targeting flexible manufacturing and warehouse automation where general-purpose behaviors have been hardest to program with classical control.
- Universal Robots. Skild Brain has been announced as a foundation-model option for Universal Robots' collaborative arm family, giving UR customers a path from hand-scripted programs to language-directed, contact-rich manipulation.
- NVIDIA. A strategic technology and investment partnership provides the training and onboard-inference stack for Skild Brain — including NVIDIA-class GPU compute on the robot itself — and is a corner of NVIDIA's broader Isaac / GR00T push into physical AI.
- Humanoid manufacturers. Skild AI positions Skild Brain as the brain layer any humanoid OEM can license. The company has not disclosed every named humanoid partner, but its public messaging is explicit that humanoids are a primary target embodiment.
- Additional target markets. Skild has signaled interest in warehouse logistics, elder-care and consumer robotics, and heavy industrial automation as longer-horizon rollouts of the Skild Brain platform.
The robotics supply chain — where Hybrid Electronics fits in
A modern humanoid robot is one of the most component-dense products ever built. A single Skild-Brain-class humanoid platform depends on thousands of individual electronic components: precision BLDC motor drivers, high-torque servo modules, IMUs and 6-DoF force-torque sensors, LiDAR and time-of-flight modules, GPU compute boards, battery-management ICs, wide-bandgap SiC/GaN power semiconductors, high-speed board-to-board connectors, precision resistors, and thousands of MLCC capacitors distributed across every power and signal rail.
Building that kind of hardware puts enormous pressure on the electronic-components supply chain. Prototype pilots and low-volume production runs frequently hit allocation, EOL notices, and long lead times on the exact parts they need, and R&D and field-service teams routinely have to support older platforms that use semiconductors no longer in mainstream distribution. During the current humanoid robotics and general-purpose robot AI boom — a boom Skild AI's own funding rounds illustrate — this pressure is amplified: dozens of well-funded teams are chasing the same short list of high-torque BLDC drivers, precision IMUs, GPU compute modules, and automotive-grade power semiconductors, and franchise stock on many key parts is now measured in months of lead time rather than weeks.
Hybrid Electronics has been sourcing electronic components for industrial and embedded systems since 1995. For robotics teams — humanoid, mobile, industrial, and research — we specialize in finding the parts you need when mainstream distribution can't:
Obsolete and end-of-life parts
Legacy semiconductors, sensors, connectors, and passives that are no longer in mainstream distribution — the parts that keep older platforms running during transitions to newer hardware, and the long-tail items that block a BOM from closing.
Allocated and hard-to-find active parts
High-torque BLDC motor drivers, IMUs and MEMS sensors, SiC/GaN power devices, GPU compute modules, precision resistors, and MLCC capacitors caught in allocation or long-lead times. Send us a BOM or a shortage list — we'll confirm what we can source and at what lead time.
Frequently asked questions
Who founded Skild AI?
Skild AI was founded in 2023 by Deepak Pathak and Abhinav Gupta, both faculty members at Carnegie Mellon University's Robotics Institute and School of Computer Science. Pathak serves as CEO and Gupta as president. The company is headquartered in Pittsburgh, Pennsylvania.
What is Skild Brain?
Skild Brain is Skild AI's general-purpose robot foundation model. It is designed as an omni-bodied system — a single trained model that runs across humanoids, quadrupeds, tabletop arms, and mobile manipulators — and uses a hierarchical architecture combining a low-frequency high-level manipulation-and-navigation policy with a high-frequency low-level action policy.
Where is Skild Brain deployed?
Skild AI has publicly announced partnerships with humanoid, industrial, and mobile-robot manufacturers to run Skild Brain on their platforms. Named partners include ABB Robotics, Universal Robots, and NVIDIA. Skild AI is a software-first robot foundation model provider rather than a humanoid OEM.
How much has Skild AI raised?
Skild AI closed a $300 million Series A in July 2024 at a $1.5 billion valuation, led by Lightspeed Venture Partners, Coatue, SoftBank, and Jeff Bezos through Bezos Expeditions. A subsequent Series B in 2025 valued the company at approximately $4.5 billion, and a Series C of close to $1.4 billion in January 2026 brought total funding above $2 billion and the valuation above $14 billion.
Is Hybrid Electronics affiliated with Skild AI?
No. Hybrid Electronics is an independent electronic-components sourcing company. We are not a distributor, reseller, or partner of Skild AI. All trademarks are property of their respective owners. Hybrid supports robotics developers by sourcing obsolete, allocated, and hard-to-find semiconductors, sensors, connectors, and passives for R&D, prototype, and low-volume production builds.
Can Hybrid help with components for humanoid robotics builds?
Yes. We regularly support R&D and integrator teams with hard-to-find semiconductors, BLDC motor-driver ICs, IMUs and other MEMS sensors, connectors, SiC/GaN power devices, and precision passives for the industrial and embedded boards used in robotics platforms. Send us a BOM or a short list of shortages and we'll confirm availability and pricing.
Trademark and affiliation notice. Skild AI® and Skild Brain™ are trademarks of Skild, Inc. Hybrid Electronics is an independent electronic-components sourcing company and is not affiliated with, endorsed by, or authorized by Skild, Inc. All information on this page is compiled from public sources for educational and reference purposes.