What is Skild AI?
The universal brain for every robot — a single foundation model that controls any embodiment, any task.
- Category
- Robotics AI Foundation Model
- Headquarters
- Pittsburgh, PA
- Founded
- 2023
- Employees
- ~64–100 (est. Q2 2026)
- Total Funding
- $1.81B across 4 rounds
- Valuation
- $14B+ (January 2026, Series C)
What is Skild AI?
Skild AI is a Pittsburgh-based physical AI company building the world's first unified robotics foundation model — the Skild Brain — that serves as a general-purpose intelligence layer for any type of robot, across any task, without hardware-specific retraining. Founded in 2023 by Carnegie Mellon University robotics professors Deepak Pathak and Abhinav Gupta, the company has raised $1.81B in disclosed equity (with CEO Deepak Pathak citing more than $2B total to Bloomberg) and reached a $14B+ valuation as of January 2026, making it the most valuable pure-play robotics software startup in the world.
The Skild Brain is an 'omni-bodied' foundation model trained on trillions of synthetic simulation episodes generated via NVIDIA Isaac Sim and NVIDIA Cosmos world foundation models, combined with millions of real-world human manipulation and locomotion video frames ingested at internet scale. It controls quadrupeds, humanoid robots, tabletop arms, mobile manipulators, and autonomous mobile robots (AMRs) through a single shared intelligence layer — delivered via cloud API and embedded firmware — without requiring operators to re-engineer policies for each new robot body or task. Enterprise customers deploy the Brain across security and facility inspection, warehouse pick-and-pack, last-mile delivery, data center operations, construction, and precision factory assembly.
Skild grew from zero to approximately $30M in live revenue within just a few months of its first commercial deployments in 2025, demonstrating rapid enterprise contract wins rather than a slow land-and-expand cycle. Specific named customer counts have not been publicly disclosed, though the company has referenced multiple enterprise deployments across warehouse, manufacturing, and inspection verticals. Key partnerships announced in March 2026 with NVIDIA, Foxconn, ABB Robotics, and Universal Robots — and the April 2026 acquisition of Zebra Technologies' Robotics Automation business (including the Symmetry Fulfillment fleet-orchestration platform) — position Skild as the first provider of a complete end-to-end warehouse automation stack: humanoids for picking, robotic dogs for inspection, arms for packing, AMRs for material movement, and an orchestration layer managing them all from a single AI control plane.
What does Skild AI offer?
Skild AI's core offering is the Skild Brain foundation model, delivered as a cloud API and embedded software for robotics OEMs and enterprise operators. The platform spans multiple vertical modules, embodiment types, and delivery modes.
- Skild Brain· Core Platform
- Omni-Bodied Foundation Model· Core Platform
- Cloud API / Robot-Brain-as-a-Service· Delivery
- Embedded Firmware for OEMs· Delivery
- Mobile Manipulation Platform· Vertical Module
- Security & Inspection Robot Platform· Vertical Module
- Autonomous Packing Module· Vertical Module
- Symmetry Fulfillment Orchestration (Zebra acquisition)· Fleet Management
- NVIDIA Isaac Lab / Isaac Sim Integration· Simulation & Training
- NVIDIA Cosmos Synthetic Data Pipeline· Simulation & Training
- Humanoid Robot Support· Embodiment
- Quadruped / Robot Dog Support· Embodiment
- Tabletop Arm Support· Embodiment
- AMR / Mobile Robot Support· Embodiment
- Warehouse Automation· Industry Vertical
- Manufacturing / Factory Assembly· Industry Vertical
- Healthcare Facility Operations· Industry Vertical
- Construction & Inspection· Industry Vertical
- Data Center Operations· Industry Vertical
How does Skild AI make money?
Skild AI operates a pure B2B software licensing and API-access model — it sells recurring access to the Skild Brain foundation model to robotics OEMs and enterprise operators, generating software revenue without manufacturing hardware. Specific per-seat or per-robot pricing tiers are not publicly disclosed; the company has not published a pricing page, and enterprise contracts are closed on a negotiated basis calibrated to fleet size and deployment complexity.
The primary revenue stream is foundation model licensing: robot hardware manufacturers such as ABB and Universal Robots embed the Skild Brain into their robot firmware or access it via cloud API, paying recurring fees. A second stream — vertical-specific software modules (mobile manipulation, inspection, autonomous packing) — allows enterprises to deploy turnkey solutions layered on top of the base Brain without custom AI engineering. The April 2026 acquisition of Zebra's Symmetry Fulfillment platform adds a third stream: SaaS fleet orchestration revenue, bringing recurring software fees from existing Zebra enterprise customers into the Skild stack.
The unit economics are structurally attractive: software licensing scales without proportional incremental cost and carries AI-infrastructure gross margins. Skild's central competitive flywheel is data compounding — every robot deployment, regardless of hardware OEM or task, feeds anonymized real-world sensor and action data back into the global Skild Brain training pipeline, improving model quality for all customers and widening the moat over time. This network effect is structurally similar to what made large language model companies defensible: the more hardware partners and enterprise deployments Skild accumulates, the more diverse training data its model ingests, and the more capable (and sticky) it becomes relative to single-OEM AI stacks.
Future monetization levers include private-cloud AI-factory deployments — packaging GPU training and inference infrastructure as a managed service for large enterprise customers who cannot expose proprietary operational data to a shared cloud environment — consumer robotics licensing as the technology matures into home and light-commercial use cases, and data-sharing arrangements with OEM partners who want to fund dedicated model improvements for their hardware. The company's January 2026 Series C at a $14B valuation implies a revenue multiple of roughly 467x the 2025 run-rate, a premium that prices in the winner-take-most dynamics of robotics operating systems.
Who leads Skild AI?
Skild AI is co-founded and led by two former Carnegie Mellon University professors who are among the most-cited researchers in robot learning and computer vision globally. The broader team has drawn senior talent from Meta AI Research (FAIR), Google, Tesla, Uber, Amazon, and NVIDIA.
- Deepak PathakCo-Founder & CEO2023–presentRaj Reddy Assistant Professor at CMU Robotics Institute (on leave); PhD from UC Berkeley (AI), BS with Gold Medal from IIT Kanpur (CS). Pioneered 'artificial curiosity' in reinforcement learning — a widely cited framework for self-directed robotic exploration. Previously a researcher at Facebook AI Research (FAIR) and Google; co-founded VisageMap Inc. (acquired by FaceFirst). Recipient of MIT TR35, Sloan Fellowship, Okawa Research Award, and multiple Best Paper Awards at ICRA and CoRL.
- Abhinav GuptaCo-Founder & President2023–presentAssociate Professor at CMU Robotics Institute (on leave), joint with CMU School of Computer Science; PhD in Computer Science from University of Maryland (2009); IIT Kanpur undergraduate. Founding research scientist at Facebook AI Research (FAIR). Known for large-scale unsupervised learning, robotic manipulation, and locomotion at scale. 25+ combined years of AI/robotics research with Pathak. Drives Skild Brain technical architecture and research direction.
- Miriam ElmasryHead of Talent Acquisition2023–presentLeads people operations and recruiting for Skild AI, based in San Francisco. Previously Senior Technical Recruiter at Tesla (Autopilot and Tesla Bot / Optimus full-cycle recruiting) and recruiter at Uber. Responsible for scaling the team from founding hires by drawing talent from top AI labs including Meta, Google, Tesla, Amazon, and NVIDIA.
How do you contact Skild AI's leadership?
The dominant verified email pattern at Skild AI is first-name only (e.g., deepak@skild.ai), used by approximately 58% of employees per RocketReach and LeadIQ analysis. Secondary patterns include last-name only (~17%) and first-initial + last name (~13%). The official press contact is press@skild.ai (published on the company website). Executive emails below follow the verified first-name pattern; co-founder addresses are inferred from the pattern, not individually confirmed from a public source.
deepak@skild.aiHow much funding has Skild AI raised?
Skild AI has raised approximately $1.81B in disclosed equity across four rounds since its 2023 founding — with CEO Deepak Pathak telling Bloomberg the total exceeds $2B when including all tranches. The January 2026 Series C, led by SoftBank Group at a $14B+ valuation, tripled the company's valuation from $4.5B to over $14B in just seven months, making Skild the highest-valued pure-play robotics software startup in the world.
The company's earliest capital was a $14.5M seed round in 2023, co-led by Lightspeed Venture Partners and Sequoia Capital, with participation from Felicis Ventures, SV Angel, and Carnegie Mellon University itself. This funded proof-of-concept Skild Brain development and founding team formation out of CMU's Robotics Institute.
In July 2024, Skild emerged from stealth with a landmark $300M Series A — one of the largest early-stage rounds Lightspeed has led — co-led by Lightspeed Venture Partners, Bezos Expeditions (Jeff Bezos' personal vehicle), Coatue, and SoftBank Group, at a $1.5B post-money valuation. Additional participants included Felicis, Sequoia, Menlo Ventures, General Catalyst, CRV, Amazon Industrial Innovation Fund, and the Alexa Fund, instantly minting Skild as a unicorn and funding the initial commercialization of the Skild Brain.
In June 2025, Skild closed a $100M Series B led by SoftBank Group at a $4.5B valuation, with NVIDIA contributing $25M and Samsung committing $10M as strategic co-investors. This round reflected strong early traction: live revenue grew from zero to approximately $30M within months of first commercial deployments. The January 2026 Series C was a $1.4B round led again by SoftBank Group Corp., with participation from NVentures (NVIDIA's venture arm), Macquarie Capital, Bezos Expeditions, 1789 Capital, IQT (In-Q-Tel), and Disruptive. Returning investors Lightspeed, Felicis, Coatue, and Sequoia all increased their positions. New strategic investors joining the cap table included Samsung, LG, Schneider Electric, CommonSpirit Health, Salesforce Ventures, and IQT — signaling early defense, healthcare, and industrial enterprise interest in general-purpose robotic intelligence.
How did Skild AI get here?
Skild AI went from academic spinout to $14B robotics software platform in under three years, anchored by landmark funding rounds, rapid revenue inflection, marquee industrial partnerships, and a transformative acquisition.
- 2023Founded & Seed Round ($14.5M)Deepak Pathak and Abhinav Gupta leave Carnegie Mellon University professor roles to found Skild AI in Pittsburgh, PA. Lightspeed and Sequoia co-lead a $14.5M seed round — with CMU itself as a participating investor — to fund early Skild Brain model development and team formation.
- July 9, 2024Emerges from Stealth — Series A ($300M, $1.5B valuation)Skild launches publicly and closes a $300M Series A co-led by Lightspeed, Bezos Expeditions, Coatue, and SoftBank at a $1.5B post-money valuation — instantly becoming a unicorn. The Skild Brain omni-bodied foundation model is formally announced. Additional investors include Felicis, Sequoia, Menlo Ventures, General Catalyst, CRV, Amazon Industrial Innovation Fund, and the Alexa Fund.
- June 12, 2025Series B ($100M, $4.5B valuation) & First Live RevenueSkild closes a $100M Series B led by SoftBank at a $4.5B valuation, with NVIDIA investing $25M and Samsung committing $10M as strategic co-investors. Live revenue grows from zero to approximately $30M within months, with multiple enterprise customers deploying the Skild Brain across warehouse and inspection applications.
- January 14, 2026Series C ($1.4B, $14B+ valuation)SoftBank leads a $1.4B Series C with co-investment from NVentures (NVIDIA), Macquarie Capital, Bezos Expeditions, 1789 Capital, and IQT. Returning investors Lightspeed, Felicis, Coatue, and Sequoia all increase positions. New strategic investors include Samsung, LG, Schneider Electric, CommonSpirit Health, Salesforce Ventures, and IQT (In-Q-Tel). Valuation triples in seven months to over $14 billion.
- March 17, 2026ABB Robotics, Universal Robots & NVIDIA Partnerships; Foxconn DeploymentSkild announces expanded collaborations with ABB Robotics and Teradyne's Universal Robots and Mobile Industrial Robots (MiR) to embed Skild Brain across global industrial cobot fleets. Concurrently, Skild deploys its omni-bodied Brain on Foxconn's NVIDIA Blackwell GPU assembly lines in Houston, performing complex dual-arm assembly operations on server rack production.
- April 15, 2026Acquires Zebra Technologies' Robotics Automation BusinessSkild acquires Zebra's Robotics Automation unit — including the Symmetry Fulfillment fleet-orchestration platform and its enterprise customer relationships — in exchange for cash plus an equity stake for Zebra in Skild. This creates the first integrated end-to-end warehouse automation stack: humanoids for picking, robotic dogs for inspection, arms for packing, AMRs for material movement, and Symmetry orchestrating them all.
Who are Skild AI's competitors?
Skild competes at the foundation-model layer against other cross-embodiment software companies, and indirectly against vertically integrated hardware-plus-software robotics companies that build their own proprietary AI stacks rather than licensing from a third party.
- Physical IntelligenceNearest direct competitor building cross-embodiment robot foundation models (π0, π0.7); raised ~$1.07B total and pursuing a further $1B round at an $11B valuation (March 2026). Partially open-sourced π0 in February 2025, a strategy that could commoditize the foundation layer where Skild competes, though Physical Intelligence has not yet demonstrated equivalent industrial OEM breadth.
- Field AIBuilds adaptive robotic software for unstructured outdoor, energy, and construction environments; raised $405M at a $2B valuation in 2025. Competes in the learn-from-observation software stack but targets field operations verticals (energy, infrastructure) rather than the warehouse and manufacturing verticals Skild has prioritized.
- 1X TechnologiesVertically integrated humanoid robot company (Eve, Neo) targeting a $10B valuation in its next round. Bundles hardware and AI policy stack, monetizes across hardware sales and software licensing, and is shipping Neo humanoids to EQT portfolio companies at scale — an alternative architecture to Skild's pure-software OEM licensing model.
- Figure AIBuilds the Figure 02 humanoid robot with an in-house AI policy stack targeting automotive and logistics. Reached a $39B valuation in September 2025 following a $1B+ Series C. A vertically integrated competitor whose OEM customers and enterprise buyers could alternatively run Skild Brain instead of Figure's proprietary control system.
- NVIDIA Robotics (Isaac / GR00T)NVIDIA's Isaac Lab, Cosmos world foundation model, and Project GR00T provide training infrastructure and a general-purpose robotics model — positioned as an ecosystem platform that could subsume the foundation-model role Skild occupies. However, NVIDIA is simultaneously a Series C investor in Skild and a key infrastructure partner, making the relationship more co-opetitive than adversarial.
- Google DeepMind (Gemini Robotics)DeepMind's RT-2 and Gemini Robotics programs develop vision-language-action models that generalize across robot embodiments — a long-horizon threat from a hyperscaler with near-unlimited compute and training data. Not yet commercialized as a licensable OEM product as of June 2026, but represents the most formidable potential entrant given DeepMind's research pedigree.
Skild AI — frequently asked questions
