2026 World Robot Conference (WRC) Exhibitor Industry Mapping

Organized into five layers by industry-chain position, not by exhibition hall. The hall assignments (Hall A "Smart Integration" / Hall B "Smart Manufacturing" / Hall C "Smart Creation" / Hall D "Smart Living" — the organizer's four official theme halls) correlate only loosely with industry-chain position: both Hall A and Hall D mix components with finished robots.

L0  Compute & Chips              →  Main-control SoCs, AI chips, power/magnetic/storage components
L1  Core Components              →  Reducers, screws, bearings, motors & servos, joint modules, dexterous hands, sensors, batteries, materials
L2  Robot OEMs                   →  Humanoid, quadruped/special-purpose, industrial/collaborative, AMR/logistics, service/commercial, medical/rehabilitation
L3  Embodied Intelligence & Data →  VLA foundation models, simulation & synthetic data, motion capture, motion-control software
L4  Applications & Ecosystem    →  Industrial giants, state-owned enterprises, innovation centers, parks/associations/capital

A single "mature/growth/early" tag conflates corporate maturity with robotics-business maturity, producing the distortion "any listed company that touches robotics = Mature" — companies like Toread, Motic, and Zhongjian Technology are corporately mature, but their robotics business is still at the narrative stage. "Robotics-revenue purity" and "corporate fundamental quality" are two orthogonal axes and should not be collapsed into a single tier.

The framework therefore splits into three independent fields:

Field One · Corporate Stage — using the 🟢🔵⚪ symbols

Field Two · Robotics Business Stage

Field Three · Robotics Exposure

Annotation scope (a deliberate trade-off): Fully tagging all three fields across 300+ companies would significantly hurt readability, and for information-sparse long-tail companies, Fields Two and Three could not be assigned credibly. The main tables therefore keep only the Field One symbol, with all three fields annotated case-by-case wherever doing so would flip the reader's judgment, and §7 collects every "corporately mature ≠ robotics-business mature" case separately under "Listed-Company Robotics Purity Audit."

Within each subcategory, ranking is by industry influence (technical positioning × market share × pricing power × ecosystem spillover) — not by valuation or revenue. Within a layer, T0 = defines the category, T1 = first tier, T2 = effective participant.

Roughly 40–50 small exhibitors in the directory have almost no public information (mostly startups or local supply-cluster companies exhibiting for the first time in 2026). Below, these are grouped uniformly into each category's "⚪ Early / Information Unverified" list, classified only by name and product line, with no fabricated background. This is a deliberate trade-off — a gap is preferable to an invention.

Three structural observations (conclusions first):

① This show carries a far higher "component content" than past editions. Hall B is almost entirely components, chips, and sensors; component density is also very high in Halls A and D. Harmonic reducers appeared from 6 exhibitors (Leader Harmonious Drive, Laifual, Fanzhou, and others, plus Harmonic Drive Systems and Kefeng), screws/linear drives from 6, force/tactile sensors from more than 15, and joint modules from more than 20. Implication: the finished-robot narrative is shifting into the delivery phase of component-level import substitution — consistent with a "narrative → delivery" lifecycle transition signal.

② The single most consequential absence is AgiBot. Per Omdia data, AgiBot ranked first in general-purpose humanoid robots by 2025 global shipment share at 39%, with 2025 revenue of RMB 1.05 billion (up nearly 20x year-on-year), cumulative production passing 10,000 units by March 2026, unit No. 15,000 rolling off the line on June 28, a disclosed post-money valuation of RMB 15 billion for its parent, and a publicly announced Hong Kong listing process launched on July 24. A company that is the global shipment leader being absent from China's largest robotics exhibition is itself an anomaly that calls for explanation.

⚠️ Boundary of attribution: There is currently no public evidence explaining why AgiBot did not exhibit. It may relate to the IPO process, marketing resource allocation, booth logistics, or other factors — the reason is unconfirmed. This note states only the facts (global shipment leader + absent + concurrently pursuing a Hong Kong IPO) without drawing a causal inference.

③ Overseas finished-robot makers are broadly absent; the international companies that did exhibit are concentrated almost entirely upstream, in components. The exhibiting international companies are AMD, Infineon, Murata, Melexis, 3M, Henkel, NSK, Schaeffler, MinebeaMitsumi, Harmonic Drive Systems, SMC, MISUMI, igus, FAULHABER, and EMAG — upstream suppliers, without exception. Tesla, Figure, Boston Dynamics, ABB, FANUC, and KUKA are absent entirely (South Korea's ROBOTIS and Neuromeka, and Japan's XELA, are present but small in scale).

A caution against over-extrapolating: international exhibitors skew structurally upstream, but even upstream is not fully represented — major upstream suppliers such as Nabtesco (RV reducers), THK (linear motion), and SKF (bearings) are likewise absent. The one inference that does hold: China is the world's assembly and shipment center for robot platforms, while key upstream components still depend heavily on international suppliers.

The robot's "brain hardware." A fair number of Chinese companies exhibit in this layer, but almost every global leader is absent — this is the layer with the widest domestic-substitution gap in the entire directory.

➕ Other notable absences in this layer: Qualcomm (RB-series robotics platforms), NXP, Texas Instruments, STMicroelectronics, Cambricon (domestic AI compute).

One of the highest-value line items in the humanoid robot BOM, and the segment where the path to domestic substitution is clearest.

The core of humanoid robot linear joints, and currently one of the segments with the highest technical barriers and the lowest domestic-production rate.

➕ Other absences: Wuxi Best Precision Machinery Co., Ltd. (roller screws), Qinchuan Machine Tool & Tool Group Share Co., Ltd. (screw grinders and products), Jiangsu DINGS' Intelligent Control Technology Co., Ltd. (micro screw motors), THK / NSK (global leaders in linear motion; NSK exhibited as an investment entity, not a product line).

➕ Absent: SKF, Timken outside the Schaeffler parent-company booth, and C&U Group.

The largest subcategory by count. Here we need to distinguish between general-purpose motor manufacturers (with an existing base business) and startups building robot-specific joint motors (pure incremental bets) — the two have completely different risk-return structures.

(a) General-purpose motors and servo drives — with a base business

(b) Robot-specific joint modules and integrated actuators — pure incremental bets

One of the most "crowded" categories at this year's show, and the segment where valuations are expanding fastest.

Assessment: Dexterous hands are a textbook "narrative-pricing" segment — the arms race on degrees-of-freedom specs is fierce, but real orders come mainly from research and demonstration use cases. Companies like Zhaowei and DH-Robotics, which have a manufacturing base business or industrial orders, have a completely different risk structure from startups purely chasing degrees-of-freedom — they should not be compared on the same ranking table.

Absences first, because the absences matter more than who's present:

Platform makers present (ranked by industry influence):

National-team players and innovation centers (quasi-OEM entities, listed separately):

This layer's absences are the most systematic — none of the global "Big Four" are present, except Yaskawa's joint venture.

➕ Other absences: JAKA, AUBO, Topstar, Han's Robot, Universal Robots (the global leader in collaborative robots). (EFORT is not on this absence list — it is present in the form of Qizhi (Wuhu); see the table above.)

This is the relatively "emptiest" layer in this directory — nearly every category leader is absent.

➕ Absent: Gaussian Robotics (a global leader in commercial cleaning), Ninebot, iRobot.

This is the layer with the highest potential value capture in the industry chain, but the least current revenue — a textbook narrative-pricing zone.

This is, in my view, the layer most worth watching at this show, yet the one with the least exposure. Data is embodied intelligence's real bottleneck, and the business model for data collection and synthesis is closer to "selling shovels" than finished hardware is.

Why this section exists: A single "Mature" label tends to conflate corporate maturity with robotics-business maturity — any listed company that puts up a robotics booth is easily stamped Mature. But "the company is Mature" does not mean "the robotics business is Mature." The table below re-tiers exhibiting listed companies by the genuine substance of their robotics business.

How to use this table: In robotics-sector rallies, valuation enthusiasm typically spreads from Group A toward Group C — and Group C sees the deepest drawdown once sentiment turns, with no fundamentals underneath to cushion it. Group A can be valued on a fundamentals framework; Group C only on a narrative framework — the two must not be mixed.

Valuation can be anchored directly to the robotics business.

Robotics is incremental, not the core business, but the supply relationship is verifiable. Valuation should separate the core business from the robotics increment — it should not be re-rated as a whole.

The company itself is mature, but its robotics business has not yet generated verifiable revenue — easily misjudged as Mature by a single-label system.

Framework-integrity note: Under this framework's principle that "framework integrity outweighs comprehensiveness," Group C companies should not remain in any investment-tier ranking. They are listed here to flag risk, not to preserve completeness.

Listed by sub-category below are companies that are "category leaders but did not exhibit" — an omission that must be filled back in when analyzing this directory:

① An exhibitor directory ≠ the industry landscape — this directory has a systematic bias. The decision to exhibit is jointly determined by booth cost, marketing needs, IPO quiet periods, local-government group organizing, and similar factors. Listed or soon-to-list companies actually have less incentive to exhibit (Unitree is the exception, being at a brand-visibility peak). Using an exhibitor directory to rank the industry will systematically overweight startups and underweight mature companies. The correct use of this mapping is to "discover new faces," not to "assess the landscape."

② The real watershed is in Unitree's prospectus numbers. Revenue RMB 159 million → RMB 1.699 billion (over three years); non-GAAP net profit RMB -18.02 million → RMB 591 million; gross margin 44.22% → 60.13%; in-house R&D-and-production ratio >90%. This is the first case in the entire industry to translate "narrative" into a "computable valuation anchor." Under narrative-trading-framework logic, Unitree has entered the Delivery stage — analysis of Unitree should switch decisively to a fundamentals framework; narrative-based pricing no longer applies. The other 95% of finished-robot makers in this directory remain in the narrative stage. These two groups of companies should not be valued with the same method.

Worth noting is a signal buried in the prospectus: Q1 2026 non-GAAP net profit fell 52.55% year-on-year (R&D expense up RMB 38.328 million y/y, plus a sharp rise in sales expense from Spring Festival Gala brand promotion); H1 2026 revenue is guided at RMB 1.052–1.128 billion (+35.6% to +45.4% y/y), but non-GAAP net profit is guided down 21.97% to 6.43% y/y. Revenue growth decelerating from a three-year 226% CAGR into the 40% range, with margins under pressure at the same time, is the textbook signature of a transition from hyper-growth into normal growth — valuation methodology needs to adjust accordingly.

③ Pricing has been set, but the valuation anchor is a "scarcity anchor," not a "fundamentals anchor." At the RMB 150.80 issue price, issue-date market cap is RMB 60.993 billion, implying roughly 71x 2025 non-GAAP P/E and roughly 25x P/S; but on a fully diluted basis, the issue P/E is 219.23x — 5.7 times the industry average of 38.56x. At the same time, H1 2026 profit is guided down year-on-year.

This is a tension worth stating plainly: the market is assigning a 219x issue P/E in the same period profit is declining. What supports this pricing is not current fundamentals but three non-fundamental factors — category scarcity (the first A-share humanoid-robot listing), a small free float (day-one tradable market cap of roughly RMB 4.49 billion, 7.36% of total market cap), and the institutional premium environment on the STAR Market, where not a single new listing broke issue price in H1 2026.

Implication: Unitree's fundamentals have entered the Delivery stage, but its secondary-market pricing in the early days of trading will still be dominated by narrative and liquidity. This is a rare case of two frameworks coexisting on the same name — the fundamentals framework sets the valuation range, the narrative framework sets the trading rhythm, and neither should be used to explain the other's price action.

④ The divergence between upstream scarcity and downstream assembly is very clear in this directory. Assets with genuine physical scarcity (harmonic-reducer precision, planetary-roller-screw process know-how, six-axis force-sensor calibration, miniature-bearing yield, magnetic sensing chips) have few participants, hard barriers, and strong pricing power. In contrast, the number of participants in joint modules, humanoid platforms, and dexterous hands surged within a single year (>13 joint-module startups, >20 humanoid-platform startups), and commoditization is outrunning barrier-building by a wide margin. Upstream names — Leader Harmonious Drive, SRI, Keli Sensing, Orbbec, Hesai — hold a structurally stronger position than downstream assemblers, even where the latter have larger revenue and more media exposure.

⑤ Data and simulation is a significantly underrated position. Companies such as LightWheel, Dexforce, Noitom, Qingtong Vision, and Moxian Technology run a "sell shovels to every finished-robot maker" business model and don't need to bet on which platform maker wins. But the layer's biggest beneficiary is the absent NVIDIA — Isaac Sim's ecosystem head start means Chinese simulation-data companies' ceiling is externally constrained (no market-share data supports language like "dominant position" here). It's also worth noting that this layer's competitors are not only domestic peers but global Physical AI platforms such as NVIDIA and Scale AI — a structural risk that must be factored into any assessment of this layer.

⑥ Where subtraction is warranted. The roughly 20 humanoid-platform startups founded in 2025–2026, the 13+ joint-module startups, and the 5 regional humanoid innovation centers in this directory can be listed for framework completeness, but should not remain in the tiered ranking of any investment framework. Reasons: unverifiable AI-revenue linkage, no operating cash flow, no evidence of differentiated barriers. Under the principle that framework integrity outweighs comprehensiveness, these should be explicitly excluded rather than parked at a low tier.

Note: Earlier versions of the exhibitor information relied mainly on public sources and outside knowledge, without checking each company's official detail page one by one — a genuine coverage gap. The official exhibitor directory currently lists 313 companies. After individually checking official detail pages (or, where unavailable, an official-page summary from search results) and re-classifying by industry-chain position, this section covers 135 companies (approx. 43%), for which the business-direction description is reliable; the remaining roughly 178 have not yet been individually verified — their business descriptions come from outside sources, and most appear only in the merged "long-tail" entries within each layer. Figures in the table marked "self-reported/self-described" — market share, shipment volume, order value, and the like — all come from the companies' own official-site descriptions, have not been cross-verified, and should not be used directly as a basis for investment decisions. The 15 companies marked "per outside sources" draw their information from two additional China robotics-industry-chain investment research documents supplied by the user (not WRC's official detail pages) — a different source from the rest of this table's entries — and were included only after cross-checking that these 15 are indeed genuine WRC exhibitors.

If on-site information conflicts with this report, the on-site information governs.