Klobot (KOSDAQ: 466100): Robot Fleet Orchestration Software
Every article about robotics is about the robot. Klobot sells the boring thing that decides whether the robot is useful: the software that gets a delivery robot, a patrol robot and a cleaning robot from three different manufacturers to share one elevator without deadlocking. It licenses that layer rather than building hardware.
Who they are
Klobot was founded in May 2017 and is headquartered in Seongnam, just south of Seoul. It listed on KOSDAQ in 2024 and remains founder-led: Kim Chang-gu, who holds roughly 15.5% of the company, became sole CEO in December 2025 when the co-CEO structure was dissolved — a governance simplification the company framed as concentrating accountability for long-term strategy.
The company sits in a category Korean industry calls "robot software," which undersells it. What Klobot actually solves is a coordination problem that gets worse, not better, as robotics succeeds. A hotel with one delivery robot has a product. A hotel with fifteen robots from four vendors, sharing two elevators and a service corridor, has a traffic control problem — and no robot manufacturer's own app will manage a competitor's fleet.
What they do
Two products carry the business. CHAMELEON is a general-purpose autonomous navigation stack licensed to robot manufacturers — mapping, localisation, path planning and obstacle avoidance that a hardware maker can port onto its own chassis instead of writing from scratch. CROMS is the heterogeneous fleet management and control system: a cloud platform that ingests robots from different vendors and presents them as one managed fleet, with integration into building systems like elevators, doors and security.
Alongside the licences, Klobot runs a robot services business, deploying and operating fleets for public agencies and enterprises across guidance, delivery, patrol, logistics automation, manufacturing automation, and cleaning and disinfection. In practice the services work generates the reference sites that make the software credible, and the software is what scales.
One 2026 contract shows the range: the Korean National Police Agency's counter-terrorism and crisis management division contracted Klobot for autonomous platform robots and quadruped robots, with the contract period running from June 2026. Quadrupeds for counter-terrorism response is a long way from hotel room service, and the common layer between them is the software.
How they make money
The revenue model is genuinely two-sided. CHAMELEON earns a per-robot licence fee plus a porting fee for each new robot model — so revenue scales with both the number of manufacturers using it and the number of units they ship. CROMS is priced by site and by robot count, plus service fees for optimisation work and integration with a customer's existing systems. The robot services business bills for consulting, design, deployment and operation.
2025 results, consolidated: revenue of KRW 41.25 billion, up 23.4%, with the operating loss narrowing 58.4% to KRW 3.12 billion and the net loss narrowing 69.4% to KRW 2.06 billion. Growing 23% while cutting the loss by more than half is the shape you want from a company at this stage — still unprofitable, but converging.
Where it sits in the value chain
Klobot occupies the software end of Stage 5 — Compute & robot foundation models. It does not compete with NVIDIA or the vision-language-action model developers; it sits above them. Foundation models decide how a single robot moves; orchestration decides which robot goes where, in what order, with what priority, and what happens when one of them stops responding in a corridor.
This layer becomes structurally more valuable as the stage below it commoditises. If every humanoid ends up running similar open models on similar chips — and NVIDIA publishing an open reference humanoid design in 2026 points that way — then differentiation moves to deployment, fleet operations and integration with the systems a building already runs. That is exactly where Klobot lives.
The bigger trend
Korea has an unusual structural position in robotics: the highest robot density per worker in the world, a government-backed humanoid programme, and large conglomerates buying into robot platforms — Samsung through Rainbow Robotics, LG through Bear Robotics, Hyundai through Boston Dynamics. All three will eventually have mixed fleets from multiple sources operating in the same buildings, and none of them will accept a rival's control plane.
The bear case is equally simple to state. Orchestration software is a thin, hard-to-defend layer that hyperscalers, robot OEMs and building-management incumbents all have reasons to absorb. Klobot's revenue is still service-heavy, which means gross margins look less like software than the SaaS framing suggests. The interesting question over the next few years is the mix: how much of the top line comes from licences that scale, versus deployment projects that don't.
What to watch
- Licence revenue as a share of the total — the difference between a software company and a systems integrator.
- Number of robot manufacturers that have ported CHAMELEON onto their own hardware, and whether any are outside Korea.
- Whether the operating loss keeps narrowing toward break-even at the current growth rate.
- Public-sector contract renewals, including the police agency work that began in June 2026.
- Any humanoid platform adopting CROMS — today's fleets are wheeled robots, and humanoids would be a genuinely new category for the control layer.
Related companies
🇰🇷Korean robot platform maker — the kind of fleet that needs an orchestration layer.
🇨🇳Supplies the perception hardware whose output this software consumes.
🇺🇸The vertically integrated alternative: builds and orchestrates its own fleet in-house.
Also in this stage but not yet profiled: NVIDIA (Nasdaq: NVDA), Alphabet (Nasdaq: GOOGL), Qualcomm (Nasdaq: QCOM) and Samsung Electronics (KRX: 005930).
FAQ
What does Klobot actually sell?
Two licensed software products and a services business. CHAMELEON is an autonomous navigation stack that robot manufacturers license per robot, with a porting fee per model. CROMS is a cloud platform that controls robots from different manufacturers as a single fleet, priced by site and robot count. The services arm deploys and operates fleets for public agencies and enterprises.
Is Klobot profitable?
Not yet, but the losses are shrinking. 2025 consolidated revenue was KRW 41.25 billion, up 23.4%, with an operating loss of KRW 3.12 billion — 58.4% smaller than the prior year — and a net loss of KRW 2.06 billion, down 69.4%.
Why would a robot maker license navigation software instead of writing its own?
Time and cost. Mapping, localisation and path planning are solved problems that still take years to make reliable across real buildings, and a hardware maker's advantage usually lies elsewhere. Licensing also gets a new model to market faster, which matters in a category where product cycles are short.
Does Klobot build robots?
No. It is a software and services company; the hardware in its deployments comes from other manufacturers. That vendor neutrality is the core of its pitch for multi-brand fleet control, since a robot maker's own management software has little incentive to manage a competitor's machines well.