Skild AI raises $1.4B at a $14B valuation

What happened

Skild AI, a company that builds general-purpose software for robots, raised a $1.4 billion Series C round that values it at more than $14 billion ^[1]. The round was led by SoftBank, with Nvidia, Macquarie Group and 1789 Capital among the other investors ^[1].

The new valuation is a large step up from the company's previous round: Skild AI last raised at a $4.5 billion valuation in the summer of 2025 ^[1]. Chief executive Deepak Pathak told Bloomberg that the company has now raised more than $2 billion in total ^[1].

What the company does

Skild AI was founded in 2023 and builds general-purpose robotic software and foundation models that can be retrofitted to different robots and tasks ^[1]. Rather than building a single robot, the company works on the software layer — models intended to be dropped onto many kinds of machines and put to many kinds of work.

The company says its models need much less additional training for each new task than conventional approaches ^[1]. That claim is central to the pitch: if a general model can be adapted to a new job with little extra training, deploying robots becomes cheaper and faster than building a bespoke system for every task.

Skild AI also hopes its models can learn from watching humans perform tasks ^[1]. Learning from human demonstration is an alternative to collecting large amounts of robot-specific training data, and it points toward software that could pick up new skills from ordinary activity rather than from carefully staged robot trials.

Why it matters

The size of the round and the investor list — a major technology investor in SoftBank, a chip company in Nvidia, and financial firms in Macquarie Group and 1789 Capital — put Skild AI among the better-funded companies working on general-purpose robot intelligence ^[1]. The jump in valuation over roughly half a year reflects how much money is currently moving toward software that could run on many robots rather than on one.

For the wider field, the bet is that the hard part of robotics is the general-purpose model, not the hardware. Skild AI's approach — one set of models, retrofitted across robots and tasks, learning partly by watching people — is one answer to the question of how robots will acquire new skills without a fresh engineering effort each time ^[1].