The most consequential robotics story of the week isn’t a humanoid doing something uncanny in a living room. It’s a Hong Kong company quietly announcing that the supply side of the robot boom is becoming a business in its own right.
D-Robotics, a provider of robotics development infrastructure, said on Thursday it has closed a US$400 million Series C — a round the company describes as the largest robotics funding round in China in the past four years. Per the press release, it attracted strategic investment from a leading global internet company alongside top-tier investment institutions and follow-on investment from existing shareholders. The company did not name the strategic investor, and trade coverage has been left pointing generally at major platform companies moving into embodied AI.
The numbers behind the round
The claims in the announcement are aggressive, so it’s worth listing them as claims: H1 2026 revenue grew several times year-over-year; cumulative shipments of the Sunrise chip series have exceeded 8 million units; the embodied AI business has entered mass production. The Sunrise S600, the embodied-AI chip the company launched in November 2025, has been adopted by more than 20 customers within six months — including TARS, Spirit AI, X Square Robot, UBTECH, PaXini Tech, Astribot, Fourier and Booster Robotics — with the majority already in mass production, per the release.
That last list is the real news. These are working robotics companies — humanoid builders, industrial wheeled-robot makers, perception suppliers — buying the same compute platform rather than designing their own silicon. If D-Robotics’ numbers hold up, the S600 is becoming something close to a default brain for China’s mid-tier robot industry. A press release is a press release, and “revenue grew several times” is doing a lot of unquantified work. But 8 million cumulative chip shipments is a hardware-scale claim, and companies don’t casually claim it.
Where the money goes tells you the strategy: a full Sunrise chip portfolio spanning low-power chips for home robots up through high-performance humanoid chips, plus a software platform covering the whole robot development chain from data collection through simulation to deployment. Not one robot. Every robot.
The picks-and-shovels layer is consolidating
Every hardware boom eventually stops being about the machines and starts being about what the machines are made of. The robot industry has been heading there for a while — and it’s arriving faster than most people expected. China’s humanoid build-out, which we’ve tracked from the UBTECH factory making 10,000 humanoids a year to robots actually holding up on real factory floors, has created dozens of companies that all need the same thing: dependable edge compute, a perception stack, a development toolchain. Most of them cannot each afford their own chip programme. D-Robotics is betting they won’t have to.
There’s a second layer to it: D-Robotics’ Gravity Program has backed more than 500 robotics innovators across 10 categories, and the company counts over 500 universities and 100,000-plus developers in more than 20 countries building on its platform. That’s not just a supplier relationship — it’s a developer ecosystem, the kind that historically decides which platforms win long before the spec sheets do. And it’s an ecosystem with real reach: China captured 97% of global humanoid shipments in AgiBot’s most recent analysis, which means whoever supplies its brains is supplying the world’s robots.
The interesting comparison is what everyone else is doing at this layer. Figure is building its own silicon-adjacent stack around Helix; Tesla is vertically integrated end-to-end; Nvidia is selling the platform from above with Jetson Thor. D-Robotics is selling it from below and sideways — chips plus toolchain, priced for companies that will never be Nvidia customers. The funding round suggests investors think there’s room for that layer to be a real business, not just a commodity.
The timing also says something. $400 million for infrastructure — not a new humanoid, not a demo — landing four years after China’s last robotics mega-round, in the same month Figure demonstrated zero-shot generalisation in 30 real homes, reads like capital deciding the embodied AI build-out has crossed from research bet to manufacturing problem. Manufacturing problems get picked-and-shovels businesses. They always have.
One caveat belongs here: this is a single-source funding announcement. The round size, the investor description and the shipment figures all come from the company’s own release, and independent verification of the financials hasn’t been published. The claim that it’s the largest China robotics round in four years is the company’s framing too. Treat the trajectory as the story, not the specific numbers.
