What happened
Dealroom reports that Skild AI, a Pittsburgh- and San Francisco-based robotics software startup, raised a $1.4 billion Series C led by SoftBank. The round reportedly values the three-year-old company at more than $14 billion and brings its total capital raised to over $2 billion. Dealroom says Skild’s software, called the Skild Brain, is designed to control different kinds of robots rather than a single hardware platform. The report says the system is operating on hundreds of robots and that the company generated roughly $30 million in revenue in under six months during 2025. These claims have not been independently confirmed from the supplied source.
Dealroom reports that Skild AI closed a $1.4 billion Series C led by SoftBank, with NVIDIA, Jeff Bezos through Bezos Expeditions, Macquarie Capital and 1789 Capital also participating. The report says strategic backers Samsung, LG and Schneider Electric joined the round. Dealroom says the financing values Skild at more than $14 billion and brings the company’s total capital raised to over $2 billion. The supplied report does not provide a term sheet, investor ownership details, liquidation preferences, or an independently verifiable financing announcement, so those aspects remain unknown.
According to Dealroom, Skild sells software rather than manufacturing robots. Its central product, the Skild Brain, is described as an “omni-bodied” AI system intended to control quadrupeds, humanoids, tabletop robotic arms and mobile manipulators. The company’s stated premise is that one intelligence layer can support multiple robot forms and tasks, avoiding a separate system for each machine. Dealroom attributes the phrase “Any task, any hardware, one brain” to co-founder and president Abhinav Gupta. The report does not explain the system’s architecture, training data, latency, hardware requirements, or failure-recovery mechanisms.
Dealroom reports that Skild’s software is running on hundreds of robots in factories, data centers and logistics hubs. It identifies NVIDIA’s Houston factory as one deployment, mentions several unnamed wiring and construction companies, and describes a trial at LaGuardia. The report also says ABB Robotics, Universal Robots and Mobile Industrial Robots are embedding the Skild Brain into their machines. The source does not specify how many robots are active at each site, which tasks they perform, how long deployments have lasted, or whether the systems operate autonomously or under close human supervision.
The report says Skild booked roughly $30 million in revenue in under six months in 2025, calling that period its first meaningful commercial stretch. Dealroom attributes the company’s founding to Gupta, who previously built Meta’s FAIR Robotics Lab, and Carnegie Mellon professor Deepak Pathak. It says the company was founded in 2023. Dealroom presents the funding as a bet on industrial and warehouse robotics, rather than home robots, and reports Gupta’s view that a broader “GPT moment” for physical AI may develop over the next year to a year and a half. That timeline is a forecast attributed to Gupta, not an independently established result.
Read the primary source: app.dealroom.co ↗
Why it matters
The reported financing is a major capital commitment to an AI approach focused on physical systems rather than text or image generation. Skild is positioning its software as a general intelligence layer for factories, logistics hubs and other industrial settings, where a common control system could reduce the need to develop separate AI for every robot design and task. The report does not establish how reliably the system performs, how broadly it is available, or whether the reported valuation reflects durable commercial demand.
If the reported figures are accurate, the round would place Skild among the most highly valued private companies focused on AI for physical robots. That matters because robotics companies have historically faced a fragmented development problem: software trained for one body, sensor package or task may not transfer cleanly to another. A system that genuinely generalizes across robot types could make deployments faster and reduce duplicated engineering work. The source reports this as Skild’s strategy, but supplies no independent performance evidence showing that the promised generality has been achieved.
The potential public and industrial impact is practical rather than primarily conversational. In factories, logistics facilities and data centers, a common AI control layer could affect how companies automate material handling, inspection, construction-related work and other repetitive operations. It could also change the bargaining position between robot manufacturers and software providers if intelligence becomes a separately supplied layer. However, the report does not say whether Skild’s deployments have replaced human labor, improved productivity, reduced injuries, or lowered costs, and those effects should not be inferred from the financing alone.
The reported OEM relationships are significant because integration into robot manufacturers’ products could give Skild a distribution route beyond direct customer projects. ABB Robotics, Universal Robots and Mobile Industrial Robots are named by Dealroom as partners embedding the software. If confirmed and expanded, those integrations could make a common control system available across a larger installed base. The source does not establish whether these are commercial agreements, pilot integrations, or arrangements with limited geographic or technical scope.
The financing also illustrates investor confidence in “physical AI,” but valuation is not proof of technical or commercial success. A $14 billion valuation can reflect expectations about future markets, strategic competition and access to capital as well as current performance. Dealroom’s reported revenue figure is notable, yet the source does not identify its accounting basis, customer concentration, margins, recurring share or comparison with operating costs. There is likewise no independent confirmation in the supplied material of the round, valuation, deployments or revenue.
What to watch next
The key test is whether Skild can turn reported deployments into repeatable, measurable commercial performance across different robot bodies and work environments. Important unanswered questions include the identity and scale of most customers, the tasks robots perform, error and safety rates, human oversight requirements, operating costs, and whether the reported revenue is recurring. Further scrutiny should also examine the terms of the financing, the independence of the reported valuation, and whether OEM integrations lead to broad availability.
The most important evidence will be deployment data that can be checked outside company or investor claims. Useful disclosures would include task-completion rates, intervention frequency, downtime, safety incidents, performance across different robot bodies, and results under changing lighting, layouts and object conditions. Without such information, “universal brain” remains a positioning claim rather than a demonstrated capability. Dealroom’s report does not provide benchmarks or test protocols.
Customer and partner detail will determine whether the reported scale is durable. The source names NVIDIA’s Houston factory and several robot manufacturers but leaves most end users unnamed. Follow-up reporting should establish whether the hundreds of robots are paid production systems, limited pilots, research units or a mixture. It should also clarify the LaGuardia trial’s purpose, duration and outcome, and whether the wiring and construction deployments are still active.
Safety and accountability deserve particular attention because errors in physical systems can damage equipment or injure people. Questions include how Skild handles uncertain commands, unexpected objects, sensor failures, network outages and changes to a worksite; who can stop a robot; and how incidents are logged and investigated. The source mentions industrial deployment but does not report safety metrics, certification, insurance arrangements, human-oversight rules or regulatory approvals.
The financing itself should be checked against primary corporate, investor or regulatory records when available. Verification should cover the amount raised, the valuation, participating investors, the company’s total funding and the reported 2025 revenue. Future coverage should also watch whether Skild releases software broadly through its OEM partners, adds more customers, or reports measurable improvements. Until then, the supplied Dealroom account supports treating the funding and strategy as reported developments, not independently confirmed proof that Skild has solved general-purpose robot control.
