Rainbow Robotics Co., Ltd. (KOSDAQ:277810) is worth ₩8.47tn. Over the last four quarters it sold ₩45.1bn of robots.
That is about 188 times sales, and 64 times the ₩132.9bn of equity on its balance sheet. The shares closed at ₩436,500 on September 2, down 3% on the day, in a 52-week range of ₩275,500 to ₩880,000.
Nobody owning this is valuing the robots it sold last year. They're valuing Samsung Electronics' decision to take 35% of the company and consolidate it, and the possibility that a Korean university spinoff ends up manufacturing humanoids at Samsung scale. That is a coherent thing to buy. It's just worth being precise about what the operating business currently shows, because two of the trends in it point the wrong way.
The growth is real and it accelerated. First-half revenue was ₩21.4bn against ₩10.4bn a year earlier, up 105%. The June quarter alone did ₩12.3bn against ₩6.2bn. Full-year 2025 revenue of ₩34.1bn was itself up 76% on 2024.
Now the margin. Gross margin was 52.3% in 2022 and 47.7% in 2023. It fell to 32.6% in 2024, recovered slightly to 34.9% in 2025, and ran about 36% across the first half of 2026.
That direction matters more than the growth rate. Falling gross margin while revenue doubles means the incremental sale is a different, cheaper product than the old one — mobile platforms and hardware rather than the collaborative arms that carried the early margins. A robotics company scaling into its own manufacturing should eventually see the opposite. Seventeen points of gross margin have gone somewhere, and the company's disclosure doesn't break out product mix finely enough to say where.
Below that line, operating losses are widening rather than narrowing. Minus ₩2.98bn in 2024, minus ₩2.48bn in 2025, and minus ₩3.59bn in the first half of 2026 alone. Selling and administrative expense in the June quarter was ₩6.66bn against ₩4.63bn of gross profit.
One warning about the history: 2023 shows ₩51.9bn of SG&A and a ₩44.6bn operating loss, offset by ₩43.7bn of finance income in the same year. Whatever that was, it makes any multi-year comparison of the expense line meaningless. Use 2024 onward.
This is the number that surprised me most. Cash and equivalents were ₩68.5bn at the end of 2024, ₩10.1bn at the end of 2025, and ₩2.5bn at the end of March 2026.
Two and a half billion won. About $1.8 million.
It recovered to ₩15.9bn by June, but look at how: investing activities contributed a positive ₩16.5bn in the June quarter, meaning the company sold financial investments. Operating cash flow was negative ₩3.18bn in the quarter and negative ₩6.46bn across the half. Financing contributed essentially nothing — ₩0.13bn.
Total assets of ₩144.9bn against ₩12.0bn of liabilities means there is no solvency question here; the balance sheet is almost entirely equity and there is more investment left to sell. But a company that let its cash balance reach ₩2.5bn while building ₩41.5bn of property and equipment is running its treasury tight, and at some point the funding for a manufacturing ramp has to come from somewhere. With Samsung at 35%, "somewhere" has an obvious answer, which is part of what the multiple is pricing.
Samsung Electronics raised its stake to about 35%, became the largest shareholder, consolidated Rainbow Robotics as a subsidiary in its own accounts, and set up a Future Robotics Office. That combination is what changed this from a small cobot maker into an ₩8tn stock.
The commercial test is running now. Coupang, Korea's dominant e-commerce operator and one of the most automated logistics networks anywhere, is evaluating the RB-Y1 — a dual-arm mobile manipulator with seven degrees of freedom per arm on a wheeled base — for sorting, transport and handling in a live warehouse. A larger deployment could follow. Rainbow Robotics has also released omnidirectional wheels and a development kit for the platform.
That is exactly the right shape of pilot. Warehouse manipulation is where humanoid economics either work or don't, the customer is demanding, and the deployment is domestic so the support cost is low. What it isn't yet is revenue.
Here's a structural point that explains the valuation better than any robotics argument, and it's specific to Korea.
KOSDAQ operates a technology-track listing route that lets deep-technology companies list on the strength of an external technical evaluation rather than a record of profits. The United States has no equivalent path; a comparable American company either stays private on venture capital or reaches the public market through a merger. Which is why the US humanoid names an investor might want — Figure, Agility, Apptronik — are all private, and Rainbow Robotics is not.
The consequence is a scarcity premium, and it's large. Global capital that wants listed exposure to humanoid robotics has very few doors. Rainbow Robotics is one of them, and it has a ₩132.9bn equity base and 19.4 million shares outstanding. A small float meeting a global theme produces exactly the price you see.
This cuts both ways as an argument. The premium isn't irrational — scarcity is a real feature of an asset. But it is a premium on listing structure rather than on business quality, and it will compress the moment the private names start going public.
The valuation is its own risk and doesn't need elaborating. At 188 times sales, revenue could grow ten-fold and the shares would still be at 19 times sales.
The gross margin trend is the underrated one. If the platform business is structurally a 35% gross margin hardware business rather than a 50% margin specialty one, then even large volumes produce modest gross profit, and the operating leverage everyone assumes arrives late or not at all.
Samsung's involvement carries a subtlety worth naming. Reporting in August indicated Samsung has also been developing humanoid work internally, separate from the RX line. A 35% shareholder that is simultaneously an in-house competitor is a more complicated relationship than a 35% shareholder that isn't, and Rainbow Robotics' minority holders have no visibility into how the group allocates programs between them.
And the pilot might just be a pilot. Warehouse robot trials that don't convert are common, and Coupang evaluating a robot is not Coupang buying thousands of them.
A purchase order. Specifically, a disclosed contract from Coupang or any large logistics operator with a unit count and a delivery schedule attached. Korean rules require a separate filing for a single supply contract above a set share of prior-year revenue, and on 2025 revenue of ₩34.1bn that threshold is tiny — meaning almost any meaningful commercial order becomes a public filing the day it's signed. Watch DART, not the earnings release.
Second, gross margin in the September quarter. Holding near 36% while revenue grows says the mix has stabilized. Falling below 30% says volume is being bought.
Third, how the company funds its next capex step. Property and equipment went from ₩19.5bn to ₩41.5bn in eighteen months on a cash balance that touched ₩2.5bn. Whether the next stage comes from a Samsung capital injection, a rights issue to all shareholders, or asset sales will say a lot about how the largest shareholder views everyone else on the register.
kstock reads DART every morning and writes up what moved — the contract, the buyback, the number that does not add up. The daily post and a Saturday roundup, by email.