Key takeaways
- Gausium (高仙机器人), founded in 2013 in Shanghai, has delivered more than 40,000 commercial cleaning robots and serves over 6,500 customers in 70+ countries.
- The company says it reached its first break-even in 2024, a rare claim of profitability among Chinese service-robot startups.
- Its Series C was ≈ US$169 million / HK$1.31 billion, co-led by Capital Today and SoftBank Vision Fund in late 2021.
- The core thesis is labor substitution: one scrubber can replace 1.5–2 cleaners where staffing is hardest.
- A single navigation stack is reused across indoor, industrial, and outdoor machines, which is the real engineering moat.
- Scaling still hinges on after-sales coverage and the slow cultural shift of facilities managers toward autonomous cleaning.
The least romantic robot in the building
Nobody films the robot that mops the airport at 3 a.m. There is no crowd, no backflip, no keynote. Yet the quietest machine on the floor may be the one actually paying its way. In a terminal or a shopping center, a cleaning robot does not need to be inspiring — it needs to show up every night, dodge luggage carts, and log its square meters. That is the business Gausium (高仙机器人) has built, and it is far harder than it looks.
The company, whose international brand is Gausium, was founded in 2013 by Edward Cheng (程昊天), a Cambridge-trained engineer, and is led technically by CTO Qin Baoxing (秦宝星), a National University of Singapore PhD in autonomous driving.
Verified scale and money
The funding history is well documented. Gausium's Series C of ≈ US$169 million / HK$1.31 billion closed in November 2021, co-led by Capital Today and SoftBank Vision Fund, with Meituan and others participating; this appears in the company's own announcement, a third-party company profile, and an independent funding compilation. A US$50 million D+ round followed in May 2024. The brand was renumbered to "Gausium" for overseas markets in 2022.
On operations, the company states it has delivered more than 40,000 commercial cleaning robots, serves over 6,500 customers, and covers more than 20 billion square meters of cleaned area across 70+ countries and 350+ cities. It also says it reached its first break-even in 2024 — a claim repeated on its official webinar page and in industry analyses. That matters: most humanoid and service-robot peers are still burning cash.
Why cleaning is the safer bet
The demand side is structural, not fashionable. Cleaning is repetitive, low-status, and increasingly hard to staff. The company cites a standard internal benchmark that one scrubber can replace 1.5–2 human cleaners, cutting labor cost by 40–60%. In aging, high-wage markets, that math closes faster than any delivery or greet-and-guide use case.
Gausium's product matrix now spans small building units, large industrial scrubbers, carpet vacuums, and outdoor sweepers. The engineering point worth noting is reuse: a single localization-and-perception stack is shared across machines, with cleaning mechanisms swapped per scenario. That lowers the marginal cost of each new model and is a more durable advantage than any single spec.
The engineering backbone
Behind the product names is a research apparatus the company says includes six R&D centers across Shanghai, Singapore, Beijing, Suzhou, Chengdu, and Shenzhen, with more than 300 engineers, plus academic ties to the National University of Singapore and Tsinghua. The less visible asset is data: years of real cleaning runs across airports, malls, and factories feed the perception and path-planning models. In a category where the differentiator is shifting from hardware specs to task success rate and continuous runtime, that accumulated operational record is harder to copy than any single machine.
From automation to autonomy
The company frames the next step as moving from "fixed-trajectory execution" to "perception-driven task decisions." Its "inspection cleaning" mode, introduced in 2022, scans for dirt and cleans only where needed, raising efficiency severalfold versus full-coverage passes. The 2026 Mira and Marvel models split sweeping and washing into separate mechanisms to avoid turning dry trash into slurry — a small, physical-world detail that separates hardware veterans from algorithm-only entrants.
Honest limitations
- Delivery, customer, and area figures are company-stated; they describe cumulative units and coverage, not active fleets or verified third-party audited counts.
- Break-even in 2024 is a company claim; no public audited financial statements confirm it, and profitability can be narrow or one-off.
- Market-leadership statements (e.g., "No. 1" shares) come from the company or single consultancy reports; treat them as category-defined, not absolute.
- Cleaning robots still need human support for dustbag changes, maintenance, and edge cases, so "unattended" is aspirational, not yet default.
What readers can do now
- Facility managers in high-labor-cost regions should pilot one scrubber against a manual team for a month and measure cost-per-square-meter plus uptime.
- Investors should ask for audited break-even evidence before treating Gausium's profitability claim as a category signal.