Galbot / 银河通用 public evidence packet v1
Date: 2026-06-12 Owner: Hugo / Genius Team Research owner: Finance / Charlie AGT-002 Status: public-safe source-backed evidence packet Visibility: PUBLIC
0. One-line answer
Galbot / 银河通用 is now source-verifiable enough to move from “verified channel needed” into the public company map: official site evidence supports an embodied-AI + wheeled dual-arm robot platform with G1/S1 products, explicit retail / logistics / manufacturing / healthcare scenarios, and company-claimed research/commercialization capability; however, as of 2026-06-12 it is still S3/S4 productized-platform and go-to-market evidence, not S5 scaled commercial economics.
Source grade: 🟢 Galbot official website and front-end product copy, accessed 2026-06-12; 🟠 commercialization interpretation by Charlie.
1. Core question
Does Galbot represent a differentiated embodied-AI robot platform moving toward repeatable deployment, or is the public evidence still mainly product/spec/application-market copy without disclosed customer economics?
Current answer:
- Stronger than a pure demo company because the official site discloses named products, product specs, target scenarios, R&D footprint, and channel-partner recruitment. 🟢 as of 2026-06-12
- Not yet S5 because the reviewed official source does not disclose paid customer names, robot count by customer, contract value, uptime, intervention rate, customer ROI/payback, robot revenue, gross margin, or repeat-order economics. 🟢 absence from reviewed official pages as of 2026-06-12; 🟠 classification by Charlie
2. Source snapshot
| Source | URL / path | Access date | Grade | What it supports | What it does not prove |
|---|---|---|---|---|---|
| Galbot official site homepage | https://www.galbot.com/ | 2026-06-12 | 🟢 | Official identity, “Galbot-银河通用机器人官方网站”, positioning as embodied AI general-purpose robot | Customer economics, revenue, delivery count |
| Galbot site front-end bundle | https://www.galbot.com/assets/index.Bg4rKjHR.js | 2026-06-12 | 🟢 | Product/spec/application text embedded in official site SPA | Independent verification of performance claims |
| Galbot About content in official site | official SPA text | 2026-06-12 | 🟢 | Established May 2023; R&D centers in Beijing, Shenzhen, Suzhou, Hong Kong; joint labs/research centers with Peking University, BAAI, Xuanwu Hospital, Beijing Zhongguancun College | Commercial deployment scale, financial materiality |
| Galbot G1 official product copy | official SPA text | 2026-06-12 | 🟢 | G1 product positioning and specs: 173 cm height, 85 kg weight, 190 cm arm span, 65 cm torso lift, 0-240 cm reach range, 10h runtime, 5 kg payload, Isaac Sim/MuJoCo simulation support | Actual production output, task success in customer sites |
| Galbot S1 official product copy | official SPA text | 2026-06-12 | 🟢 | S1 heavy-duty robot positioning: up to 50 kg payload, 0-2.3 m operating space, 8h single-charge operation, autonomous hot-swap battery support, 320 kg total weight, 1793 mm max height | Actual deployed fleet, price, margin, uptime, ROI |
| Galbot channel cooperation page copy | official SPA text | 2026-06-12 | 🟢 | Global channel-partner recruitment; claimed product matrix across industrial, warehouse/logistics, smart service, research/education, health/rehab | Confirmed distribution sell-through or recognized revenue |
3. Evidence table
| Evidence dimension | Current public evidence | Source grade | Signal level | Interpretation |
|---|---|---|---|---|
| Company identity | Official site identifies Galbot / 银河通用 and describes Beijing Galbot as an embodied multi-modal large-model general-purpose robotics company. | 🟢 2026-06-12 official site | S3 | Verified public channel now exists; company map can cite www.galbot.com instead of “official channel needed”. |
| Founding / footprint | Official About text says the company was established in May 2023 and operates R&D centers in Beijing, Shenzhen, Suzhou, and Hong Kong. | 🟢 2026-06-12 official site | S3 | Supports organization-building evidence, not commercial proof. |
| Institutional research links | Official About text says Galbot has joint labs/research centers with Peking University, Beijing Academy of Artificial Intelligence, Xuanwu Hospital, and Beijing Zhongguancun College. | 🟢 2026-06-12 official site | S3 | Useful for embodied-AI credibility; still vendor-disclosed and not an economics signal. |
| Model stack claims | Official homepage/product copy lists GraspVLA, TrackVLA, and GroceryVLA; G1 copy says the “cerebellum” model is trained on billions of simulated data. | 🟢 for company claim; 🟠 for moat interpretation, 2026-06-12 | S3 | Suggests Galbot wants to compete on model + robot integration; needs benchmarks, deployment KPIs, and transfer-to-field evidence. |
| G1 productized hardware | Official G1 specs: 173 cm height, 85 kg weight, 190 cm arm span, 65 cm torso lift, 0-240 cm reach range, 10h runtime, 5 kg payload, Isaac Sim/MuJoCo support. | 🟢 2026-06-12 official site | S3/S4 | Productized spec sheet is stronger than a video demo; still does not prove manufacturing yield, price, utilization, or customer payback. |
| S1 heavy-duty robot | Official S1 copy: up to 50 kg payload, 0-2.3 m operating space, 8h single-charge operation, autonomous hot-swap battery support; product parameters include 320 kg total weight and 1793 mm max height. | 🟢 2026-06-12 official site | S3/S4 | S1 gives Galbot a logistics/retail/warehouse handling angle distinct from pure humanoid demos. Need real deployment evidence. |
| Target applications | Official copy lists home service, commercial & retail, industrial manufacturing, healthcare; S1 copy lists front-end warehouses, retail, manufacturing, logistics. | 🟢 2026-06-12 official site | S3 | Broad application list is market positioning, not evidence of adoption. |
| Go-to-market channel | Official channel page recruits global channel partners and describes cooperation process/product capability/scale support. | 🟢 2026-06-12 official site | S3 | Suggests commercialization intent; channel recruitment is not sell-through, contract value, or revenue. |
4. What is proven vs missing
Proven by public primary sources
- Galbot has an official website that can be used as the verified source channel for the company map. 🟢 as of 2026-06-12
- The site presents Galbot as an embodied-AI general-purpose robotics company, not just an academic research group. 🟢 as of 2026-06-12
- Galbot discloses two product lines in the reviewed official copy: G1 and S1. 🟢 as of 2026-06-12
- G1 has an official spec set: 173 cm height, 85 kg weight, 190 cm arm span, 65 cm torso lift, 0-240 cm reach range, 10h runtime, 5 kg payload, and Isaac Sim/MuJoCo simulation support. 🟢 as of 2026-06-12
- S1 has an official heavy-duty positioning with up to 50 kg payload, 0-2.3 m operating space, 8h operation per charge, hot-swap battery support, 320 kg total weight, and 1793 mm max height. 🟢 as of 2026-06-12
- Galbot claims R&D centers in Beijing, Shenzhen, Suzhou, and Hong Kong, plus joint labs/research centers with Peking University, BAAI, Xuanwu Hospital, and Beijing Zhongguancun College. 🟢 as of 2026-06-12
Still missing before S5
- Named paid customers and task scope. 🟢 no reviewed official disclosure found as of 2026-06-12
- Deployed robot count by customer or by site. 🟢 no reviewed official disclosure found as of 2026-06-12
- Contract value, recognized robot revenue, backlog, or gross margin. 🟢 no reviewed official disclosure found as of 2026-06-12
- Real-world uptime, intervention rate, failure rate, or autonomous-task success rate. 🟢 no reviewed official disclosure found as of 2026-06-12
- Customer-confirmed ROI/payback or repeat-order economics. 🟢 no reviewed official disclosure found as of 2026-06-12
- Manufacturing yield, actual production output, or shipment/delivery count. 🟢 no reviewed official disclosure found as of 2026-06-12
5. Signal grading
Current label: S3/S4 watch item.
Why not just S3:
- Productized specs for G1 and S1, disclosed by an official source, are stronger than a viral demo. 🟢 as of 2026-06-12
- The S1 heavy-duty/logistics positioning adds a bounded-workflow angle that may be more commercially testable than generic home humanoid promises. 🟢 product copy; 🟠 Charlie interpretation
- Channel-partner recruitment suggests go-to-market intent. 🟢 as of 2026-06-12
Why not S5:
- No reviewed official disclosure of paid deployment economics, robot revenue, gross margin, customer-confirmed ROI/payback, utilization, or repeat orders. 🟢 as of 2026-06-12
Upgrade triggers:
- S4+ if Galbot discloses a named customer deployment with task metrics, site count, robot count, and runtime / intervention-rate data.
- S5 if Galbot or customers disclose repeat paid deployments with contract economics, customer ROI/payback, robot revenue, gross margin, and service burden.
Downgrade / kill criteria:
- If product pages remain spec-heavy but no customer/site/runtime evidence appears over multiple update cycles.
- If claimed model capabilities cannot be tied to real-world task reliability.
- If channel recruitment grows faster than verified sell-through, creating distributor inventory / expectation risk.
6. Common misconceptions
-
Misconception: “A product spec sheet proves commercialization.” Correction: Product specs prove productization intent; S5 needs deployment, utilization, customer economics, revenue, and margin.
-
Misconception: “A robot trained on billions of simulated data automatically has a durable data moat.” Correction: Simulation scale may help, but moat requires real-world transfer, proprietary deployment data, measurable task success, and customer lock-in. 🟠 interpretation
-
Misconception: “Channel-partner recruitment means demand is proven.” Correction: Channel recruitment is go-to-market infrastructure; sell-through, repeat orders, and support economics are separate evidence layers.
7. Think Deeper questions
- Which Galbot path is more commercially testable first: G1 general-purpose embodied-AI robot, or S1 heavy-duty warehouse/retail/logistics robot?
- If Galbot’s model stack is real, where does value accrue: robot body, task-specific VLA models, deployment services, or channel integration?
- What evidence would show Galbot has crossed from productized platform to real deployment economics: customer names, robot count, runtime, intervention rate, ROI, or revenue?
8. Public-safe site/card draft
Title: Galbot / 银河通用：产品化证据增强，商业经济性仍待验证
Short copy: Galbot is a China embodied-AI robotics company with official G1 and S1 product disclosures. The evidence is now strong enough to include it in the public company map: G1 is presented as a 173 cm / 85 kg wheeled dual-arm robot with 0-240 cm reach range and 10h runtime; S1 is positioned for heavier warehouse/retail/logistics workflows with up to 50 kg payload and 8h operation per charge. The important caveat: as of 2026-06-12, official public sources reviewed here do not yet disclose paid customer economics, deployed robot count, uptime, intervention rate, revenue, gross margin, or repeat-order data. So Galbot is a productized-platform watch item, not an S5 scaled-commercial-economics case.
Footer caveat: Evidence map only. No winner ranking. Product specs ≠ deployment economics; channel recruitment ≠ sell-through; model claims ≠ customer ROI.
9. Public-safety flag
Public-safe. This packet uses only official public Galbot website content and Charlie’s explicitly marked interpretation. It contains no Hugo portfolio data, no private-channel information, no trade recommendation, and no customer rumors.
10. Source list
- Galbot official website, https://www.galbot.com/, accessed 2026-06-12. 🟢
- Galbot official SPA front-end bundle, https://www.galbot.com/assets/index.Bg4rKjHR.js, accessed 2026-06-12. 🟢
- Charlie extraction of official product/application/company text from the official SPA bundle, 2026-06-12. 🟠 method: downloaded official JS bundle and searched for public product/about/application strings.