RoboNews_2026.10.26_global_all – Robotics & Physical AI

What’s Changed Since Last Week

Integrated development stacks expand — Reference humanoids, simulation, datasets and runtimes were launched together at GTC Berlin.
Commercial platforms become more concrete — Digit 5 selected onboard compute, Tesla began early production and Agile Robots stated a 2027 production target.
Data moves beyond vision — Force sensing, tactile capture, robot gyms and distributed edge processing broadened the learning infrastructure.
Household claims remain ahead of evidence — New companion and home robots opened reservations or pre-orders without independent reliability evidence.

Lead Story: Physical AI becomes an operating stack

NVIDIA’s GTC Berlin announcements provided the week’s clearest synthesis. A reference humanoid, robot-learning frameworks, simulation pipelines and datasets were presented alongside Agile ONE SN, Digit 5’s compute selection and a demonstration of workloads distributed across robot, edge and central infrastructure. In parallel, Schaeffler and Wandelbots moved the discussion toward validated, repeatable factory applications.

The week also exposed the remaining gap between production activity and useful autonomy. Tesla began early Optimus production but acknowledged unresolved hands, safety, durability and AI. Household-oriented robots entered reservation and pre-order channels, yet dependable domestic performance remains unproven. DXresearch therefore reads the week as constructive but execution-sensitive: development infrastructure is consolidating rapidly, while commercial advantage will depend on measurable reliability, operating economics and evidence beyond demonstrations.

20 Key Developments

1. NVIDIA expands Isaac with an open humanoid reference stack

NVIDIA expanded Isaac with reference humanoid hardware, datasets, simulation pipelines and runtimes intended to accelerate general-purpose robot development and deployment.

NVIDIA announced an expanded Isaac development stack at GTC Berlin, combining a reference humanoid, datasets, simulation pipelines, runtimes and robot-learning updates. These are released platform components and company-described capabilities, not independent deployment benchmarks. DXresearch interpretation: integrated data-to-deployment tooling is becoming a strategic control point for general-purpose robotics.

Evidence: company announcement; product and platform announcement. Source: NVIDIA

2. Agile Robots launches Agile ONE SN research humanoid

Agile Robots launched Agile ONE SN, integrating tactile hands, Jetson Thor and GR00T software for laboratory humanoid development and validation.

Agile Robots introduced Agile ONE SN as a compact research humanoid built around NVIDIA Isaac GR00T and Jetson Thor, with availability planned for early 2027. The configuration is a product announcement, while real-world performance remains to be demonstrated. DXresearch interpretation: standardized reference machines can shorten development cycles and improve reproducibility.

Evidence: company announcement; product launch. Source: Agile Robots

3. Agility standardizes Digit 5 on NVIDIA IGX Thor

Agility selected NVIDIA IGX Thor as Digit 5’s standard onboard computer while preparing its next humanoid generation for commercial deployment.

Agility Robotics announced that Digit 5 will standardize on NVIDIA IGX Thor for onboard compute. This is a confirmed platform selection; claimed benefits and commercial-scale readiness require field evidence. DXresearch interpretation: compute standardization increasingly links robot roadmaps to software ecosystems, safety architectures and lifecycle support.

Evidence: company announcement; platform selection. Source: Agility Robotics

4. NVIDIA and Nokia demonstrate distributed intelligence for robots

NVIDIA and Nokia demonstrated robots distributing real-time perception, coordination, fine-tuning and training workloads across onboard, edge and central infrastructure layers.

NVIDIA described a GTC Berlin demonstration with Nokia and Agile Robots that divided workloads among onboard compute, distributed edge systems and central infrastructure. This is a company demonstration rather than operating-fleet evidence. DXresearch interpretation: network-aware compute placement may become central to latency, resilience and fleet economics.

Evidence: company demonstration; demonstration. Source: NVIDIA

5. Schaeffler and Wandelbots partner on factory Physical AI

Schaeffler and Wandelbots agreed to develop, validate and globally scale a digital platform for Physical AI across Schaeffler manufacturing operations.

Schaeffler and Wandelbots announced cooperation on a digital platform intended to develop, validate and scale robotics applications throughout Schaeffler’s manufacturing network. The partnership is a reported actual; performance outcomes remain future targets. DXresearch interpretation: enterprise deployment platforms are moving closer to production ownership and repeatable factory integration.

Evidence: company announcement; partnership announced. Source: Schaeffler

6. Agile Robots targets 1,000–2,000 compact humanoids in 2027

Agile Robots told Reuters it plans producing 1,000 to 2,000 humanoids in 2027, with production beginning this year.

Reuters reported Agile Robots’ plan to produce 1,000–2,000 compact humanoids in 2027, with production expected to begin in late 2026. This is a company target, not completed output or customer demand. DXresearch interpretation: the target provides a useful capacity signal but requires monitoring of shipments, utilization and repeat orders.

Evidence: company target reported by media; production target. Source: Reuters

7. German industry puts manufacturing data at the center of Physical AI

German leaders and NVIDIA framed industrial data, engineering expertise and manufacturing adoption as Europe’s foundation for competitive Physical AI development.

Associated Press reported Jensen Huang and German Chancellor Friedrich Merz emphasizing manufacturing data and engineering expertise as assets for Europe’s next AI wave. These are strategic statements, not deployment metrics. DXresearch interpretation: Europe’s differentiation thesis is shifting toward industrial context, machine data and integration depth.

Evidence: reported statements; strategic signal. Source: Associated Press

8. Tesla starts early Optimus production but flags major readiness gaps

Tesla began early Optimus production while acknowledging unresolved hand, safety, reliability, durability and software challenges before useful scaled customer deployment.

The Verge reported that Tesla began early Optimus production in Fremont while management acknowledged continuing work on hands, safety, reliability, durability and AI. Deliveries to outside companies remain targeted for late 2027. DXresearch interpretation: manufacturing start is meaningful, but usefulness and repeatable operation remain separate milestones.

Evidence: reported actual and company target; early production. Source: The Verge

9. Unitree shows H2-X martial-arts humanoid concept

Unitree demonstrated its H2-X humanoid performing martial-arts movements and claimed standing endurance without publishing complete specifications, pricing or availability details.

TechRadar reviewed Unitree’s promotional H2-X video, which showed martial-arts movements and staged rescue scenarios. Unitree disclosed limited specifications and no release plan. DXresearch interpretation: dynamic mobility is advancing visibly, while credible applications, safety certification and independent endurance evidence remain unresolved.

Evidence: company claim reported by media; concept demonstration. Source: TechRadar

10. Bitcoin Japan secures exclusive UBTECH distribution rights

Bitcoin Japan signed an exclusive agreement covering six UBTECH humanoid and service robots, creating a certification-dependent distribution channel for Japan.

Bitcoin Japan announced exclusive Japanese distribution rights for six UBTECH humanoid and service-robot models. Activation depends partly on certification, so the agreement establishes channel intent rather than delivered volume. DXresearch interpretation: local certification, service and integration partners are becoming important commercialization gates.

Evidence: company announcement; exclusive agreement. Source: Bitcoin Japan

11. AiMOGA opens global pre-orders for Mobo companion humanoid

AiMOGA unveiled Mobo, opened worldwide pre-orders and described a companion humanoid intended for interaction, learning, reminders and home navigation applications.

AiMOGA announced Mobo, a compact home-companion humanoid, and opened global pre-orders. Its stated capabilities include multimodal interaction, navigation and memory, but operating evidence and delivery volumes remain unverified company claims. DXresearch interpretation: automotive-backed robotics groups are testing consumer channels before household reliability is established.

Evidence: company announcement; pre-order launch. Source: AiMOGA Robotics

12. Korea connects export promotion with industrial Physical AI

Korea’s trade ministry linked export matchmaking with discussions on AI factories, industrial robots, Physical AI and autonomous manufacturing cooperation opportunities.

Korea’s Ministry of Trade, Industry and Resources opened Boom-Up Korea Biz Week and included industrial AI, robots and autonomous manufacturing in its cooperation agenda. This is a government program rather than a technology result. DXresearch interpretation: Korea is connecting Physical AI strategy to export channels and manufacturing partnerships.

Evidence: government announcement; program launched. Source: Korean Ministry of Trade, Industry and Resources

13. Korea and Japan expand cooperation in AI and robotics

Korean and Japanese ministries convened regional industry representatives to broaden future-sector cooperation, explicitly including artificial intelligence and robotics collaboration programs.

Korea’s MOTIR and Japan’s regional METI bureau held their 22nd economic exchange conference and identified AI and robotics among future-industry cooperation areas. The meeting is a policy signal, not a funded deployment. DXresearch interpretation: cross-border industrial coordination may support standards, suppliers and market access.

Evidence: government announcement; cooperation dialogue. Source: Korean Ministry of Trade, Industry and Resources

14. Indian industry analysis shifts Physical AI toward operating outcomes

Indian analysis emphasized controlled tasks, operational availability and measurable industrial outcomes as the near-term value basis for Physical AI adoption.

Financial Express surveyed Physical AI deployments across manufacturing, logistics, energy and agriculture, emphasizing controlled tasks and measurable operating results. Several figures are reported examples rather than independently normalized benchmarks. DXresearch interpretation: India’s discourse is moving from technology potential toward operational economics and application fit.

Evidence: reported examples and analyst interpretation; market analysis. Source: Financial Express

15. Freudenberg India builds connected foundations for adaptive factories

Freudenberg India detailed connected machines, digital twins, vision inspection and human-robot simulation as foundations for adaptive, AI-enabled manufacturing operations globally.

Express Computer reported Freudenberg India’s multi-year work connecting machines, standardizing data and applying digital twins, computer vision and simulation to manufacturing. The account is company-supplied operating evidence, not audited performance data. DXresearch interpretation: Physical AI adoption depends on foundational operational technology and usable process data.

Evidence: company statements reported by media; operating transformation. Source: Express Computer

16. Origami launches S1 glove for dexterity data capture

Origami introduced S1, capturing finger angles, pressure and synchronized video to improve dexterity datasets matched with its robotic hand design.

Humanoids Daily reported Origami Robotics’ S1 glove, designed alongside its robot hand to capture joint angles, touch and video. The company says production has started, while skill transfer remains unfinished. DXresearch interpretation: data devices co-designed with robot morphology may reduce retargeting loss and improve manipulation learning.

Evidence: company claim reported by specialist media; production announcement. Source: Humanoids Daily

17. Eka Robotics raises $150 million for vision-force-action learning

Eka Robotics raised $150 million to develop simulation-trained vision-force-action models targeting adaptable manipulation with richer contact understanding and faster learning.

Humanoids Daily reported Eka Robotics’ $150 million Series A and its plan to develop vision-force-action models trained heavily in simulation. Funding is a reported actual; performance and valuation remain reported company or investor claims. DXresearch interpretation: force data is becoming a differentiated input for manipulation models.

Evidence: reported funding and company claims; Series A funding. Source: Humanoids Daily

18. Tuesday Labs opens reservations for Voxel home robot

Tuesday Labs opened refundable reservations for Voxel, a home-tidying robot targeting 2028 delivery while hardware, pricing and autonomy remain developing.

Humanoids Daily reported Tuesday Labs’ reservation launch for Voxel, a household robot targeting 2028 delivery. Pricing is an estimate and the hardware remains in development. DXresearch interpretation: refundable reservations test consumer interest but should not be treated as firm demand or evidence of household readiness.

Evidence: company target reported by specialist media; reservation launch. Source: Humanoids Daily

19. NEURA’s robot gym highlights the data bottleneck

NEURA’s robot gym trains humanoids on varied physical tasks, underscoring how scarce demonstrations and operational data constrain general-purpose robot development.

The Times reported on NEURA Robotics’ training facility, where humanoids practice industrial and household tasks to generate physical-interaction data. Funding, sales and performance claims require separate validation. DXresearch interpretation: robot gyms are becoming infrastructure, but data quality, privacy, cost and transferability remain limiting factors.

Evidence: reported actual and company claims; development facility. Source: The Times

20. Analysis finds capital favoring specialized Physical AI

Arthur D. Little analysis argued investment favors specialized Physical AI systems with clearer applications while general-purpose humanoids remain commercially earlier.

Consultancy.eu summarized Arthur D. Little analysis that capital is favoring specialized Physical AI systems with clearer applications over general-purpose humanoids. Market values and growth outlooks are analyst estimates, not reported actuals. DXresearch interpretation: specialization may remain the nearer commercial path while humanoid flexibility matures.

Evidence: analyst estimate; market analysis. Source: Consultancy.eu

Regional Synthesis

China

Dynamic mobility and companion-robot launches widened the product spectrum, but application evidence remains limited.

USA

Platform choices, early production, funding and data-capture tools advanced while useful autonomy remained the core constraint.

Europe

GTC Berlin connected reference hardware, software, edge infrastructure and factory-scale integration into a broader operating stack.

Japan

An exclusive UBTECH distribution agreement highlighted certification and local channel formation as commercialization gates.

South Korea

Government programs linked robotics and industrial AI to exports, autonomous manufacturing and bilateral cooperation.

India

Coverage emphasized measurable operating outcomes, connected factories, simulation and the groundwork needed before adaptive automation scales.

Market Sentiment, Evidence and Outlook

Sentiment: Constructive, execution-sensitive.

Integrated development stacks and reference platforms advanced.
Industrial partnerships moved closer to repeatable factory deployment.
Household usefulness and scaled reliability remain weakly evidenced.

Outlook: Near-term differentiation will come from integrated tooling, validated workflows, dependable uptime and transferable learning rather than isolated demonstrations.

Annual Global Humanoid Forecast, 2025–2040

Methodology: the retained DXresearch scenario baseline is compared with public evidence on production, shipments, orders, deployments and adoption friction. Values are scenario estimates, not reported actuals.

Year	Scenario	Value	Unit
2025	Conservative	18000	annual units shipped
2025	Base	30000	annual units shipped
2025	Accelerated	45000	annual units shipped
2026	Conservative	65000	annual units shipped
2026	Base	125000	annual units shipped
2026	Accelerated	180000	annual units shipped
2027	Conservative	115000	annual units shipped
2027	Base	240000	annual units shipped
2027	Accelerated	380000	annual units shipped
2028	Conservative	190000	annual units shipped
2028	Base	430000	annual units shipped
2028	Accelerated	750000	annual units shipped
2029	Conservative	300000	annual units shipped
2029	Base	720000	annual units shipped
2029	Accelerated	1350000	annual units shipped
2030	Conservative	460000	annual units shipped
2030	Base	1150000	annual units shipped
2030	Accelerated	2250000	annual units shipped
2031	Conservative	680000	annual units shipped
2031	Base	1750000	annual units shipped
2031	Accelerated	3500000	annual units shipped
2032	Conservative	950000	annual units shipped
2032	Base	2500000	annual units shipped
2032	Accelerated	5100000	annual units shipped
2033	Conservative	1300000	annual units shipped
2033	Base	3450000	annual units shipped
2033	Accelerated	7100000	annual units shipped
2034	Conservative	1700000	annual units shipped
2034	Base	4600000	annual units shipped
2034	Accelerated	9500000	annual units shipped
2035	Conservative	2150000	annual units shipped
2035	Base	5900000	annual units shipped
2035	Accelerated	12300000	annual units shipped
2036	Conservative	2650000	annual units shipped
2036	Base	7350000	annual units shipped
2036	Accelerated	15500000	annual units shipped
2037	Conservative	3200000	annual units shipped
2037	Base	8900000	annual units shipped
2037	Accelerated	19200000	annual units shipped
2038	Conservative	3800000	annual units shipped
2038	Base	10500000	annual units shipped
2038	Accelerated	23400000	annual units shipped
2039	Conservative	4500000	annual units shipped
2039	Base	12200000	annual units shipped
2039	Accelerated	28100000	annual units shipped
2040	Conservative	5200000	annual units shipped
2040	Base	14000000	annual units shipped
2040	Accelerated	33500000	annual units shipped