Robot brains: $14 billion of value, $30 million of sales
The teams building universal brains for robots say they are still in their GPT-2 era. Their investors are already paying the price of success: a valuation above $14 billion against roughly $30 million of revenue. What that ratio really tells you about how companies get priced.
The fact
The companies building universal “brains” for robots admit they are still in their GPT-2 era: models that impress in a demo and do not yet hold up in production. Investors are not treating them as prototypes. Skild AI raised close to $1.4 billion in January 2026, lifting its valuation above $14 billion, against revenue the company puts at around $30 million.
Why it deserves attention
That works out at roughly 470 times revenue. No sector multiple gets you there: a mature software publisher rarely trades above ten times sales. So the price does not come from the accounts, it comes from scarcity. Few teams are judged capable of cracking the problem, and the number is set by competition between investors for a place on the cap table, not by what the company collects.
Pricing by scarcity is not new. What is new is who carries it. These rounds are no longer venture capital alone: asset managers, sovereign investors and industrial groups now sit in them (Samsung, LG, Schneider and Nvidia at Skild AI; Tether at Germany's Neura Robotics, up to $1.4 billion in June 2026), while in China Unitree has gone further still: listed in Shanghai since 19 August 2026, it closed its first session 460% above its offer price, near $50 billion, then gave part of that back in the days that followed. The risk is leaving the closed circuit of specialist funds and moving into long-term savings and the automation plans of real operating companies. If robotics reaches its ChatGPT era five years late, a handful of funds will no longer be the ones absorbing the gap.
To understand how a price gets attached to a company that sells almost nothing, read the Fundamental: “Business valuation: how companies are priced.”
You'll learn the three families of methods (discounted cash flow, market multiples, adjusted net assets), the techniques built for companies with no revenue such as the Berkus and VC methods, and how to read the next valuation you come across.
Read the Fundamental →
