Doosan Robotics / 두산로보틱스 (454910.KS)

Asia's leading cobot brand pivoting to an AI-systems-integrator and humanoid story — strong franchise, shrinking revenue, roadmap still on paper.

Investment Research · Photoncap-style deep dive · v1 of "Doosan Robotics" · May 14, 2026

What Doosan Robotics physically does

Doosan Robotics makes collaborative robot arms — "cobots" — and is one of the largest cobot brands in Asia. A cobot is a lightweight, force-limited robot arm designed to work safely alongside humans without a safety cage: it senses contact and stops, so it can be deployed on a factory floor next to a worker rather than fenced off the way a traditional industrial robot arm is. Doosan's cobot range spans payloads from a few kilograms up to the heavier 20kg-plus class, where the company has historically been particularly strong — by some 2022 estimates it held roughly 72% of the global cobot market for payloads of 20kg or more. The use cases are machine-tending, palletising, assembly, welding, screwdriving, food service (Doosan's coffee-and-food robot cells are a visible product) and inspection — the work that sits between fully manual and fully automated.

The technology proposition is deployment speed and flexibility. A traditional six-axis industrial robot from Fanuc or Yaskawa is faster and more precise, but it requires integration engineering, safety fencing and fixed installation. A Doosan cobot is meant to be set up quickly, redeployed across tasks, and programmed by a non-specialist. That has made cobots the fast-growing segment of factory automation and the natural on-ramp for small and mid-sized manufacturers who cannot justify a full automation cell.

The 2025–2026 strategic story is Doosan trying to move beyond being a cobot hardware vendor. The company is repositioning as an AI-powered "intelligent robot solutions" provider and systems integrator — selling modular, fast-to-deploy automation solutions rather than just arms — and has publicly committed to humanoid robot development. It opened the Doosan Robotics Innovation Center in Bundang in September 2025 to consolidate its AI, software and humanoid R&D, and stood up dedicated AI/software and humanoid development teams in the second half of 2025. Doosan does not yet make a shipping humanoid; the humanoid line is a stated strategic intent backed by an R&D centre, not a product. Doosan also does not make its own AI foundation models — like its Korean peers, the intelligence layer is the build-or-partner challenge ahead.

Product roadmap

The shipping product line is the cobot range — the H-, M-, A- and E-series collaborative arms across the payload spectrum, plus packaged solutions like the food-and-beverage robot cells. Doosan has cumulatively shipped over 10,000 cobot units as of late 2025, a real installed base. The product cadence here is incremental: payload extensions, controller and software upgrades, easier programming, and the "intelligent robot solutions" packaging that bundles the arm with vision, software and integration into a faster-deploy product.

The roadmap event that re-rated the stock is the humanoid and AI pivot. The confirmed milestones: a strategic shift announced in 2025 toward "practical humanoids" capable of complex tasks; the formation of dedicated AI/software and humanoid R&D teams in H2 2025; the opening of the Doosan Robotics Innovation Center in Bundang in September 2025; and a "Best of Innovation" award at CES 2026 for the company's intelligent-robot-solutions positioning. What is not on the roadmap is a humanoid product with a name, a launch date and a price. As of May 2026, Doosan's humanoid effort is an R&D programme and a strategic narrative, not a product line — the distinction the company's valuation does not currently make. Treat the cobot installed base and shipment numbers as confirmed; treat the humanoid roadmap as stated intent only, with no dated deliverable.

The financial print

This is where Doosan Robotics gets uncomfortable. FY2025 revenue was approximately KRW 32.98 billion (~$23M USD) — down roughly 29.6% from KRW 46.83 billion in 2024. The cobot business shrank materially in 2025. The company has been loss-making — forward P/E is not meaningful because the company is not expected to print a profit on the near-term consensus horizon, which is why the frontmatter forward_pe is recorded as 0.0 (not applicable). Trailing-twelve-month revenue as of late 2025 was around $22M USD, consistent with the shrinking top line. Against that, the market capitalisation is roughly KRW 6.9 trillion (~$4.9B USD) — a company doing ~$23M of declining revenue, losing money, valued near $5 billion.

The honest framing: Doosan Robotics is valued almost entirely on the humanoid/AI-integrator narrative and on its strategic position within the Doosan Group, not on its printed financials. The earlier attempt to merge Doosan Robotics into the much larger, profitable Doosan Bobcat — at a stock-swap ratio that was widely criticised for overvaluing the loss-making Robotics unit — was cancelled in December 2024 after Korean market turmoil and shareholder opposition; that episode is itself evidence that even the controlling group struggled to find a defensible valuation for the standalone Robotics business. There is no robust multi-broker FY2026 earnings consensus to anchor to, because the earnings are negative and the story is narrative-driven; treat any forward figure as estimate-basis.

The binary event is the Q1 2026 earnings, due around May 15, 2026 — the print that shows whether the FY2025 revenue decline has bottomed and reversed, or whether the cobot core is still contracting while the humanoid story remains pre-product. With the stock +16.2% above its 50-day MA and RSI at 61.5 going in, there is some momentum cushion but also a real "show me" bar.

Customer mix today

Doosan Robotics' customer mix is its genuine strength and is the opposite of the customer-concentration risk that afflicts a name like UBTECH or Rainbow. Doosan sells cobots through a broad global distribution network — distributors and system integrators reselling to small and mid-sized manufacturing end-users across North America, Europe and Asia. The company opened a US headquarters to push North American distribution and has built one of the more extensive cobot channels in the industry. There is no single dominant customer; the revenue is diversified across thousands of end-users and a wide distributor base. Doosan does not disclose a top-customer concentration percentage because the structure is genuinely diffuse — that diffuseness is real, and it is a positive.

But the structural shift in the mix is the wrong direction on volume. FY2025 revenue fell ~29.6%, meaning the diversified channel sold materially fewer cobots than in 2024 — a function of soft global manufacturing capex, intensifying price competition from lower-cost cobot makers, and a destocking cycle in the distributor channel. The humanoid "customer mix" does not exist yet — there are no humanoid customers because there is no humanoid product. So the customer picture is: a healthily diversified but shrinking cobot base, and a humanoid customer base of zero. The valuation prices a future humanoid customer mix that has not begun to form.

What's actually happening in the cobot channel

The mechanism that matters for the near term is whether Doosan's cobot revenue decline is cyclical or structural. The cyclical read: 2025's global manufacturing capex was soft, SMEs deferred automation spend, and distributors worked down inventory — a destocking cycle that mean-reverts when manufacturing capex recovers, and the Q1 2026 print is where a recovery would first show. The structural read: cobots are commoditising. Lower-cost Chinese cobot makers are competing hard on price, the technology is maturing to the point where the hardware is increasingly undifferentiated, and a Doosan cobot now competes not only with Universal Robots and Fanuc's CRX line but with a widening field of cheaper alternatives. A 29.6% revenue decline in a single year is too steep to be purely cyclical — some of it is structural pricing and share pressure.

This is precisely why Doosan is pivoting to "intelligent robot solutions" and AI-integration: the company is trying to escape commoditised hardware competition by moving up the stack into software, vision-integrated solutions and faster modular deployment, where margin and differentiation are defensible. The CES 2026 "Best of Innovation" award and the Bundang Innovation Center are the visible artefacts of that pivot. The honest assessment in May 2026: the pivot is strategically correct but unproven in the numbers — there is no disclosed evidence yet that the solutions/AI mix is offsetting the hardware decline. The Q1 2026 print is the first real test of whether the channel has bottomed and whether the up-the-stack move is generating revenue.

The competitive threat / Universal Robots and the Chinese cobot field

Doosan's direct competitive set on its actual revenue base: Universal Robots (the Danish company, the global cobot market leader with the most comprehensive portfolio and global presence — the benchmark Doosan is measured against), Fanuc's CRX collaborative line (Fanuc bringing its industrial-robot scale, reliability and 8-year-maintenance-free positioning into the cobot segment), Stäubli, and a broadening field of lower-cost Chinese cobot makers competing aggressively on price. Industry evaluations place Universal Robots as the market leader and Doosan as an "emerging leader" — strong, expanding, but not the category leader. On the humanoid side, Doosan competes — on paper — with Hyundai's Boston Dynamics, Rainbow Robotics (its Samsung-backed Korean peer), UBTECH, Unitree and Tesla. On humanoids Doosan is the furthest behind of any of these: it has an R&D centre and a press release where the others have products.

The competitive bear case is the squeeze: in cobots Doosan is being pressured from below by cheap Chinese hardware and from above by Universal Robots' scale and Fanuc's reliability brand, and its revenue is shrinking as a result. In humanoids it is behind everyone with a product. The competitive bull case is narrower — Doosan has a real, large, trusted cobot brand and installed base, a genuine global channel, and Doosan Group backing, and if the AI-solutions pivot works it could leverage that channel to sell higher-value integrated systems rather than commoditised arms. There is no active IP litigation of note involving Doosan Robotics as of May 2026. The competitive risk is not legal — it is being structurally squeezed in the business it actually has, while being structurally behind in the business its valuation prices.

The terminal risk

The terminal risk for Doosan Robotics is being caught between two losing positions. On one side, cobots get commoditised: the hardware becomes a low-margin, Chinese-cost-structure-dominated category, and Doosan's cobot revenue and margin erode structurally rather than cyclically — the 29.6% FY2025 decline is the leading edge of that, not a one-off. On the other side, the humanoid pivot never produces a competitive shipping product: Doosan starts the humanoid race years behind Boston Dynamics, UBTECH and Unitree, without Samsung-scale backing of the kind Rainbow has, and the "practical humanoid" stays an R&D programme. If both happen, Doosan is a shrinking cobot maker with a humanoid story that never ships, and the KRW 6.9 trillion valuation has no support.

The transition timing that governs the downside is shorter than for the pure humanoid names because the cobot-commoditisation pressure is happening now, in the printed revenue, not in some 2030s scenario. The humanoid-parity timing is the same long-tail question as for the rest of the theme. The named alternative beneficiaries of Doosan's squeeze are Universal Robots and Fanuc (taking cobot share with scale and reliability) and the components supply chain (selling into whoever wins). Doosan does have a credible asset — the brand, the 10,000-unit installed base, the global channel — and the AI-solutions pivot is the strategically right response. But "strategically right" and "valued correctly" are different things, and at ~$5B for a shrinking, loss-making, ~$23M-revenue business, the multiple is pricing a turnaround and a humanoid franchise that are both still hypothetical.

Bull / Gap / Optionality

1. Genuine cobot franchise — brand, installed base, global channel. Doosan is one of Asia's leading cobot brands with over 10,000 units shipped and one of the industry's more extensive distribution networks, including a US headquarters built out for North American reach. This is a real asset that a humanoid startup cannot replicate — a trusted brand and a route to market.

2. The AI-solutions pivot is the strategically correct response to commoditisation. Moving up the stack from commoditised hardware to AI-integrated "intelligent robot solutions" and modular systems integration is exactly the right move — it is where margin and differentiation live. The CES 2026 "Best of Innovation" award and the September 2025 Bundang Innovation Center are tangible commitment, not just slides.

3. Cyclical revenue decline could mean-revert. Part of the FY2025 29.6% revenue drop is a soft-manufacturing-capex and distributor-destocking cycle. If global manufacturing capex recovers through 2026, the cobot channel restocks and the top line can bounce — the Q1 2026 print is the first place a recovery would show.

4. Doosan Group backing and strategic intent. Doosan Robotics sits inside the Doosan Group, which has the balance sheet to fund a multi-year AI and humanoid R&D push. The group's strategic commitment — dedicated humanoid teams, a consolidated innovation centre — means the humanoid effort is funded, not starved.

5. Cobots are still a structurally growing category long-term. Even with near-term price pressure, collaborative robots remain the fastest-growing segment of factory automation as SMEs automate. If Doosan defends share and rides the category, the installed base compounds — the FY2025 decline notwithstanding.

Gap

1. Revenue is shrinking — down ~29.6% in FY2025. This is the single most important fact and the valuation ignores it. A ~$23M-revenue company doing less revenue than the year before, valued near $5 billion, is being priced on narrative entirely. A decline this steep is too large to be purely cyclical.

2. The humanoid roadmap is a press release, not a product. As of May 2026 there is no Doosan humanoid with a name, a launch date or a price — only an R&D centre and strategic intent. Doosan is the furthest behind of any humanoid-narrative name in this batch, and it lacks the Samsung-scale captive backing that gives Rainbow its story.

3. Squeezed from both sides in the core business. Cheap Chinese cobot makers compete from below; Universal Robots' scale and Fanuc's reliability brand compete from above. Doosan is an "emerging leader," not the category leader, and its shrinking revenue is the evidence of the squeeze.

4. Even the controlling group couldn't agree a valuation. The cancelled 2024 Doosan Bobcat merger — abandoned partly because the swap ratio was widely seen as overvaluing the loss-making Robotics unit — is direct evidence that pricing this business is contentious even for insiders. The standalone valuation rests on a turnaround that has not started in the numbers.

Optionality

The trade

Doosan Robotics is the weakest risk/reward in this batch and the position reflects it — a small, tactical holding at most, not a conviction name. The franchise is real, but the revenue is shrinking, the humanoid roadmap is still a press release, and the ~$5B valuation prices a turnaround and a humanoid business that are both hypothetical. If you take a position, initiate at KRW 101,460–112,140 (current KRW 106,800 ± 5%) — a melt-up-tape entry, with the stock +16.2% above its 50-day MA and RSI at a not-yet-extended 61.5. Size at 0.5% of risk capital — the smallest sizing in the batch; this is a Bucket C name where the chase risk is high and the fundamental support is thin. Stop at KRW 88,000, roughly 18% below current, beneath the structural support — a wide stop because the name is volatile and narrative-driven, but a hard one because if the Q1 print confirms the cobot decline is structural rather than cyclical, the narrative premium can unwind sharply. The catalyst is the Q1 2026 earnings around May 15, 2026 — the test of whether the FY2025 revenue collapse has bottomed. The cleaner expression of the same thesis: if you want Korean cobot/robotics exposure, Rainbow Robotics (277810) at least has a real strategic owner committing capital and a shipping semi-humanoid product, and if you want the broader robotics theme, the industrial-arm incumbents Fanuc (6954) and Yaskawa (6506) are profitable, growing, and gaining cobot share rather than losing it. Conviction: 4 / 10.

Sources referenced inline throughout. Reference v1 of this template format: _Watchlist/hanmi-photoncap-style.md.

\newpage

RR — Richtech Robotics Inc. · SKIP / WAIT (Tier-3) · Conv 3/10 · Bucket B

ticker: RR name: Richtech Robotics Inc. theme: Robotics bucket: B conviction: 3 entryzonelo: 2.65 entryzonehi: 2.93 currentprice: 2.79 pricedate: 2026-05-14 positionsizepct: 0.4 stoploss: 2.20 thesisoneline: Micro-cap service-robot operator pivoting to a RaaS model and an unproven "Dex" humanoid; large cash pile, tiny revenue, heavy dilution. catalystnext: Dex humanoid rollout (2Q FY2026) + next quarterly print catalystdate: 2026-05-31 deepdivepath: Theme -- Robotics/RR/rr-deep-dive.md lastupdated: 2026-05-14T00:00:00Z rsi: 58.8 vs50ma: 18.7 forwardpe: -12.0 themecycleposition: early customermixsummary: Hospitality/restaurant and casino service-robot deployments; small RaaS recurring base (~$0.3M) plus hardware sales; no single disclosed dominant customer. terminalriskoneline: Service robots are a commoditized hardware category and "Dex" enters a humanoid race already led by far better-capitalized players. bulldriverscount: 4 gapriskscount: 4 optionalitycount: 5 lastearningsdate: 2026-02-13 nextearningsdate: 2026-05-31