Hermitage Capital wanted exposure to artificial intelligence (AI) robotics in China, but it was wary of volatility in the manufacturing and embodied AI segments: Companies were asking for high valuations despite some being pre-product or pre-revenue, and it was unclear whether they would survive as cutthroat competition clashed with nascent commercialisation.

The technology-focused investor opted for indirect exposure via the infrastructure layer instead, participating in a USD 100m Series A round for D-Robotics, which designs AI chips used in humanoid robots and service robots. The company is a year-old spinout from Horizon Robotics, a Hong Kong-listed chipmaker focused on autonomous driving.

“We are less focused on project-driven businesses that rely on heavy customisation for individual clients. Our investment thesis prioritises scalable technology companies capable of delivering standardised, high-margin products at volume,” Sean Xiang, founder and CEO of Hermitage Capital, explained.

“The ideal candidates typically require 3-4 years of deep R&D to build technical moats before hitting mass production – what we call the ‘hard-tech marathon’ model.”

The D-Robotics Series A comprised multiple tranches. Hermitage contributed USD 8m across two: a first tranche of USD 26m led by 5Y Capital in June 2024, and a second tranche of USD 36.6m in March 2025. Other investors include GL Ventures, Linear Venture, Vertex Growth, Huangpu River Capital, Plum Ventures, and Unity Ventures.

The right fit

With bases in Hong Kong and Shanghai, Hermitage claims to be a global growth-stage investor. It has USD 1.5bn in assets under management, and has invested in 32 companies since its establishment in 2017. The strategy is described as high conviction, with a preference for repeat investments in existing portfolio companies through the acquisition of primary or secondary shares.

For example, Hermitage deployed more than USD 110m across 10 separate investments in financial technology player Airwallex and USD 50m in Horizon via five tranches. It would do the same for D-Robotics, provided the company sticks to its planned growth trajectory.

D-Robotics doesn’t sit in Hermitage’s investment sweet spot – Series B rounds and beyond, when companies have achieved market and technology validation. However, as a spinout from an established business, D-Robotics has the makings of a track record and demonstrated product-market fit. The entry valuation was also attractive, the PE firm added.

Originally Horizon’s non-automotive solutions division, D-Robotics was incorporated as a subsidiary before the parent’s Hong Kong IPO last year. It received R&D services through the end of 2024 and has an agreement to purchase product development solutions from Horizon until the end of 2026. There is an option to extend the contract if both sides agree.

Chip development is D-Robotics’ core business, but it has a full-stack infrastructure offering that also includes algorithms and software. Two series of chips – RDK X3 and RDK X5 – have been released, with 5m units produced to date. Sales are expected to reach 10m by the end of this year. The proceeds will be pumped into further product development and team expansion.

The company has worked on more than 10 types of vertical robotics applications for 200 companies. Key clients include robot vacuum makers Narwal and Ecovacs. Cong Wang, who led the spinout and now serves as CEO, told local media that revenue has tripled since the separation from Horizon.

According to Xiang, the chips are “indispensable for specialised robotics applications” thanks to the combination of superior computing performance with industry-leading energy efficiency. Meanwhile, economies of scale are expected to help reduce costs and ensure competitive profit margins.

Competitive edge?

Broader-level considerations for Hermitage in making its investment decision included the large market size of robotics, China’s engineering dividends, and the push for more domestically developed chips in response to geopolitical tensions.

According to Morgan Stanley Research, the humanoid robot market will be worth USD 5tn by 2050, by which point more than 1bn humanoids will be in use. Henry Zhang, president and a managing partner at Hermitage, expects D-Robotics to be in a stronger position when general robotics – still at an early stage of commercialisation – enter mass production.

“No matter which robotics company becomes the frontrunner, it will not develop its own chips but instead rely on those from Nvidia or domestic suppliers,” Zhang said. “You will hardly find another domestic company besides D-Robotics that truly has the capability to develop computing chips for general robots.”

With D-Robotics aiming for a qualified IPO by mid-2029, the key to achieving the requisite scale is becoming the go-to computing platform for both specialised and general robotics. “Companies in the upstream segment of the value chain must establish a dominant market position by delivering high-value-added solutions that address critical customer pain points,” Xiang added.