Aldebaran Robotics

From Paris pioneer to Chinese acquisition: how the world's most-studied social robot company spent twenty years building a research institution and failed to build a business.

---

How to Read This Report

This report separates four categories of evidence throughout. Readers should weight them accordingly.

Inline citations use bracketed numerals keyed to the Sources list in §14. Where the research dossier is thin on a topic, this report says so plainly rather than filling space with speculation. Module placeholders (e.g., <!-- module: products -->) are rendered by the Max Robotics platform with live database content; the surrounding prose is written to read coherently with or without that panel visible.

---

01Executive Overview

Aldebaran Robotics occupies a peculiar position in the history of robotics: it is simultaneously one of the most academically influential robot companies ever to exist and one of the most commercially unsuccessful. The NAO humanoid robot, first sold in 2008, became the default research platform for human-robot interaction (HRI) globally, accumulating more than 3,600 independent research studies ¹¹. Pepper, launched in 2014, briefly captured the imagination of the retail and hospitality sectors. Together, the two products sold approximately 37,000 units across more than 70 countries ⁷¹¹. Neither figure translated into a viable business.

The company's trajectory is a case study in the gap between scientific utility and commercial sustainability. Founded in Paris in 2005 by Bruno Maisonnier, Aldebaran attracted early venture backing including a $13 million round led by Intel Capital ¹⁵, was acquired by SoftBank for approximately $100 million around 2012 ¹⁷, rebranded as SoftBank Robotics Europe, and then sold again in 2022 to URG (United Robotics Group), a German holding company, which restored the Aldebaran name ³⁴. Under each successive owner, the fundamental problem remained unresolved: the robots were expensive to manufacture, prone to hardware failures, limited in autonomous capability, and priced at a level that made large-scale commercial deployment economically irrational for most buyers.

By February 2025, Aldebaran had entered safeguard proceedings, followed by formal bankruptcy ⁸. The Paris Commercial Court approved liquidation in June 2025, terminating 106 employees and leaving debts exceeding €60 million ³¹³. The accumulated net deficit from 2019 to 2022 alone reached €156 million ¹². In July 2025, Shenzhen-based Maxvision — a company whose primary business is biometrics, security systems, and smart transportation — acquired the core intellectual property and engineering assets, with plans to establish a new entity, NAO Robotics SA, staffed by approximately 59 engineers ⁷⁹.

What Maxvision has acquired is substantial in research terms and uncertain in commercial terms. The NAO platform has genuine scientific credibility, an established developer ecosystem, and a ROS2-compatible software stack ²¹. What it does not have is a demonstrated path to profitability. Pepper production halted in 2020–2021 ¹⁸; NAO7 was reportedly in development at the time of liquidation but has not shipped; and the broader market for social robots in retail and service environments has contracted sharply since the mid-2010s enthusiasm that drove Pepper's initial deployment.

This report examines what Aldebaran actually built, what the evidence supports about its technology, why the business failed despite genuine scientific achievement, and what the Maxvision acquisition realistically means for the NAO platform's future.

Latest news

This module is being compiled — no data to show yet.

---

02The Aldebaran Robotics Story

Founding and Early Years (2005–2012)

Aldebaran Robotics was founded in Paris in 2005 by Bruno Maisonnier, an entrepreneur with no prior robotics engineering background but a clear conviction that humanoid robots would become consumer and educational products within a decade ³⁴. The company's early years were characterised by genuine engineering ambition on a startup budget. The first NAO prototype was demonstrated publicly at RoboCup 2007, and the robot was selected as the standard platform for the RoboCup Standard Platform League — a decision that would prove transformative for the company's research footprint ³⁴.

The RoboCup selection was not merely a marketing win. It meant that robotics research teams worldwide were required to purchase NAO units to compete, creating a captive institutional market and generating a feedback loop of academic publications, software contributions, and platform familiarity among the next generation of robotics engineers. By the time NAO went on general sale in 2008, it had already accumulated a research community that no competitor could easily replicate.

In 2010, Aldebaran raised $13 million in a funding round led by Intel Capital ¹⁵, described at the time as positioning the company as a "world leader in humanoid robotics." The framing was premature but not entirely without basis: NAO was genuinely the most capable commercially available small humanoid at that price point, and the developer ecosystem was growing.

The SoftBank Era (2012–2022)

SoftBank's acquisition of Aldebaran — with an initial transaction around 2012 and full consolidation by approximately 2015 ¹⁷³⁴ — brought capital, manufacturing scale, and the ambition to turn social robots into a mass-market consumer and enterprise product. The acquisition price of approximately $100 million ¹⁷ reflected SoftBank's belief, shared by Masayoshi Son personally, that humanoid robots were on the cusp of mainstream deployment.

The most visible product of this era was Pepper, announced in 2014 and manufactured by Foxconn in Taiwan ³⁴. Pepper was a fundamentally different product from NAO: taller (approximately 1.2 metres), wheeled rather than bipedal, equipped with a chest-mounted tablet screen, and designed explicitly for customer-facing service interactions in retail, banking, and hospitality. SoftBank deployed Pepper in its own Japanese retail stores and struck agreements with HSBC and other large enterprises ³⁴. The robot attracted enormous media coverage and generated genuine commercial interest.

The commercial reality was more sobering. A 2015 survey reportedly found that only 15% of Pepper customers planned to renew their contracts ³⁴ — a figure that, if accurate, suggests the robot was failing to deliver measurable value in service deployments. Pepper's limitations were structural: it could not carry objects, climb stairs, or perform physical tasks; its conversational capability was scripted rather than genuinely adaptive; and the novelty effect that drove initial deployments wore off quickly. Production was halted in 2020 or 2021 ¹⁸, though the exact year is slightly uncertain across sources.

The company was rebranded SoftBank Robotics Europe during this period ¹⁶, reflecting its position as the European arm of a larger SoftBank Robotics Group that also operated in Japan and the United States.

The URG Interlude and Collapse (2022–2025)

In 2022, United Robotics Group (URG), a German robotics holding company, acquired SoftBank Robotics Europe and restored the Aldebaran name ³⁴. The strategic rationale was to consolidate European humanoid robotics capabilities under a single entity and pursue profitability through a combination of cost reduction and new product development.

The URG period was marked by what employees later described as unrealistic expectations: a two-year target for profitability that the company's cost structure and market position made essentially unachievable ³⁴. RAG-Stiftung, a German foundation that had been providing funding, stopped its financial support in August 2024 ³⁶. Without that backing, the company's cash position deteriorated rapidly.

In January 2025, Aldebaran entered a safeguard procedure ⁸. By February 2025, formal bankruptcy proceedings had begun ⁸. The Paris Commercial Court approved liquidation in June 2025, terminating all 106 employees and leaving creditors facing debts exceeding €60 million ³¹³. The accumulated losses from 2019 to 2022 had reached a net deficit of €156 million, with an operating loss of €26 million recorded in 2023 alone ¹².

The liquidation prompted immediate concern among the thousands of institutions worldwide that had deployed NAO robots and depended on Aldebaran's software infrastructure, cloud services, and technical support. RobotLAB, the primary North American distributor, publicly pledged continuity of service to its customers ¹¹ — an acknowledgement that the cloud dependency risk, long noted by independent observers, had become acute.

The Maxvision Acquisition (July 2025)

Maxvision, a Shenzhen-based company whose primary business encompasses biometrics, security systems, and smart transportation technology, acquired Aldebaran's core intellectual property and engineering assets on approximately 19 July 2025 ⁷⁹. The acquisition included the NAO platform IP, the Choreographe software ecosystem, and — critically — a reported 59 engineers from the original team ⁹.

The planned successor entity, NAO Robotics SA, is described as maintaining a French research and development presence alongside customer service operations ⁷. NAO V7, which was reportedly in development at the time of liquidation, is expected to continue under the new ownership ⁹.

EDITORIAL INFERENCE: Maxvision's acquisition of Aldebaran is strategically coherent from a Chinese industrial policy perspective — acquiring established Western robotics IP, an international brand with genuine research credibility, and a trained engineering team at distressed-asset prices. Whether Maxvision has the robotics product management capability to commercialise NAO more successfully than three previous owners managed is an open question. The company's core competencies are in security hardware and systems integration, not consumer or educational robotics. The retention of French engineers is a positive signal for platform continuity but does not resolve the fundamental commercial challenges that defeated Aldebaran under SoftBank and URG.

---

03Product Portfolio: What Aldebaran Robotics Actually Sells

NAO

NAO is the product that defined Aldebaran and, in many respects, defined the field of small humanoid robotics for research and education. The current production version is NAO6; NAO7 was reportedly in development at the time of liquidation ⁹.

Verified hardware specifications [1][24][25]:

NAO is bipedal, which distinguishes it from Pepper and from most competing educational robots. The 5 kg mass and low-power motors mean that falls — which occur with some regularity given the robot's balance limitations — do not typically cause injury to nearby humans or catastrophic damage to the robot itself ³⁰. This is a genuine safety advantage in educational and research settings.

The software ecosystem is NAO's strongest asset. Choreographe, Aldebaran's proprietary visual programming environment, allows non-engineers to create behaviours through a graphical interface. The full SDK supports Python and C++ development. ROS2 compatibility has been demonstrated through the Open Access NAO (OAN) framework ²¹, enabling integration with the broader robotics research ecosystem independently of manufacturer APIs. Speech recognition and synthesis support eight languages via Nuance integration, and the platform supports face detection, object recognition, and — more recently — GPT-based conversational capability ¹²¹.

Pricing (VERIFIED across multiple independent sources [4][5]):

The Robots-as-a-Service pricing model introduced by RobotLAB represents an attempt to lower the barrier to institutional adoption, though the total cost of a 36-month RaaS contract at $289/month ($10,404) is comparable to outright purchase at the lower end of the range.

Pepper

Pepper was Aldebaran's attempt to move beyond the research niche into mass-market service robotics. The design choices reflected this ambition: a wheeled base (more stable and reliable than bipedal locomotion for service environments), a chest-mounted tablet for displaying information and accepting touch input, and an emphasis on emotional recognition and conversational interaction.

Verified hardware characteristics [34][18]:

• Height: approximately 1.2 metres
• Locomotion: wheeled (omnidirectional)
• Display: chest-mounted tablet screen
• Sensors: cameras, microphones, emotion/face recognition capability
• Manufacturer: Foxconn, Taiwan
• Commercial release: 2014
• Production halt: 2020–2021

Pricing (VERIFIED across multiple independent sources [4][5]):

Pepper's pricing placed it firmly in the enterprise budget category. At $25,000–$30,000 per unit, a meaningful retail deployment of ten robots represented a $250,000–$300,000 capital commitment before any software integration, content development, or maintenance costs. The 15% contract renewal rate cited from 2015 ³⁴ — if accurate — suggests that most enterprise customers concluded the return on that investment was insufficient.

Production was halted in 2020–2021 ¹⁸. No new Pepper units have been manufactured since. Existing deployed units continue to operate where they remain functional, but the platform is effectively end-of-life.

Plato

Plato is a wheeled serving assistant, distinct from both NAO and Pepper. It is mentioned in the official product documentation ¹ but receives minimal coverage in independent sources. Specific hardware specifications, pricing, and deployment scale are UNKNOWN from the available dossier. Its existence suggests Aldebaran made at least one attempt to address the food service and hospitality automation market, but the absence of independent coverage implies it achieved negligible commercial traction.

NAO V7 (In Development)

NAO V7 was reportedly in development at the time of liquidation ⁹. No specifications, pricing, or release timeline have been publicly confirmed. Whether Maxvision will continue this development under NAO Robotics SA is stated as an intention ⁹ but remains UNKNOWN in terms of concrete milestones or delivery dates.

Portfolio Summary

Products & versions

NAO

Bipedal humanoid robot (58 cm, 5 kg, 25 DOF) designed for education, research, and HRI; first sold in 2008 with current version NAO6 and NAO7 in development at liquidation.

Pepper

Wheeled humanoid robot (~1.2 m tall) with a chest tablet screen, emotion and face recognition, used in retail, service, and social interaction contexts; released 2014, production halted 2020–2021.

Plato

Wheeled serving assistant robot developed by Aldebaran for indoor service and hospitality applications.

---

04Technology Stack: Strengths and the Work That Remains

Software Architecture

Aldebaran's software stack is more mature than the company's financial history might suggest. The Choreographe visual programming environment has been refined over more than fifteen years and represents a genuine contribution to accessible robot programming. Its graphical behaviour-tree interface allows educators and researchers without deep programming expertise to create complex multi-step behaviours, which is a primary reason NAO became the dominant educational robotics platform globally.

At the developer level, the NAO SDK supports Python and C++ with well-documented APIs for motor control, sensor access, speech, vision, and navigation. This documentation quality — accumulated over years of active developer community engagement — is a significant asset that competitors building new platforms must spend years replicating.

The Open Access NAO (OAN) framework, documented in a 2024 arXiv paper ²¹, demonstrates ROS2 integration that operates independently of Aldebaran's proprietary cloud APIs. This is technically significant: it means that even if Aldebaran's cloud infrastructure were to go offline — a risk that became acute during the liquidation — researchers and developers can maintain functional ROS2-based operation of existing NAO hardware. The OAN framework provides standardised ROS2 interfaces for NAO's actuators, sensors, and perception systems, enabling integration with the broader ROS2 ecosystem of navigation, manipulation, and AI packages.

Perception and Sensing

NAO's perception capabilities are adequate for its intended use cases and limited for anything more demanding. The dual-camera system (up to 1280×720 at 60 fps) supports face detection, object recognition, and basic visual navigation ¹. The four-microphone array enables sound localisation and supports speech recognition in eight languages via Nuance integration ¹.

A 2013 research paper on vision-guided robot hearing ²² demonstrated NAO-based audio-visual integration for sound source localisation — an example of the kind of perception research the platform has enabled. The work is methodologically sound but reflects the research-grade rather than production-grade nature of NAO's sensing: the capabilities demonstrated in controlled laboratory conditions do not straightforwardly transfer to noisy, unstructured real-world environments.

The 2016 arXiv paper on autonomous sensorimotor context discovery ²⁰ describes NAO-based research into unsupervised motor learning — the robot learning relationships between its motor commands and sensory outcomes without supervisory signals. This is legitimate and interesting research, but it is important to be precise about what it demonstrates: experimental capability in controlled conditions, not deployed autonomous operation in commercial settings.

Locomotion and Physical Capability

This is where the gap between NAO's research utility and its commercial limitations is most stark. NAO is a bipedal robot, which makes it scientifically interesting and practically fragile. It cannot climb stairs, open doors, or carry objects of any meaningful weight ³⁴. Its gait is slow and computationally expensive. Falls occur with sufficient regularity that the fall manager — an inertial sensor system that detects impending falls and adjusts motor stiffness to minimise damage — is a standard operational feature rather than an edge-case safety system ¹³⁰.

The 5 kg mass means falls are low-risk to humans, and the robot's construction is designed to absorb impact ³⁰. However, hardware failures are a persistent operational problem. Independent sources consistently cite ribbon harness tears as a common failure mode ³⁴ — a structural design issue that makes field repair difficult and expensive. The combination of fragile internal wiring, bipedal instability, and limited physical capability means NAO requires regular maintenance and is unsuitable for unsupervised deployment in environments where it might encounter unexpected physical obstacles.

Pepper's wheeled locomotion is more reliable for service environments, but Pepper's physical capability is similarly constrained: it cannot manipulate objects, navigate stairs, or perform any physical task beyond moving between locations and displaying information on its chest tablet.

Artificial Intelligence and Autonomy

EDITORIAL INFERENCE with evidence basis: The autonomy characterisation of NAO and Pepper requires careful framing. Both robots execute their primary tasks — social interaction, educational demonstrations, scripted service interactions — without a human performing those tasks in real time. In that narrow sense, they are autonomous. However, the nature of that autonomy is scripted behaviour execution with sensor-triggered branching, not general-purpose autonomous decision-making.

The GPT-based conversational capability noted in recent documentation ¹ represents a genuine upgrade to the conversational layer, enabling more naturalistic dialogue than the earlier rule-based systems. However, this capability depends on cloud connectivity, which introduces the dependency risk noted by independent observers ³⁰³¹.

A 2025 arXiv paper ²³ describes a reinforcement learning pipeline for text-driven motion generation on NAO — using large language models to generate motion descriptions and reinforcement learning to optimise the resulting motor commands. This is technically sophisticated research that demonstrates the platform's continued relevance as a research testbed. It does not, however, represent deployed production capability: the paper describes an experimental pipeline, not a feature available in shipping NAO units.

Cloud Dependency Risk

The cloud dependency issue deserves specific attention because it represents a systemic risk that the liquidation made concrete. Several of NAO's AI functions — including some speech recognition and conversational capabilities — depend on network connectivity to Aldebaran's cloud infrastructure ³⁰. When a company enters liquidation, cloud services are among the first casualties. RobotLAB's public pledge of service continuity ¹¹ and the OAN framework's cloud-independent ROS2 operation ²¹ are partial mitigations, but they do not fully resolve the risk for institutions that have deployed NAO units relying on Aldebaran's proprietary cloud APIs.

EDITORIAL INFERENCE: This is a structural vulnerability in the social robotics business model more broadly, not unique to Aldebaran. Robots sold as hardware products with cloud-dependent AI capabilities create long-term support obligations that are difficult to sustain through multiple ownership transitions. The Maxvision acquisition will need to address cloud infrastructure continuity explicitly if NAO Robotics SA is to retain the confidence of the existing installed base.

Technology Strengths and Gaps: Summary

---

05Research, Papers, Authors and Labs

Research Footprint

NAO's selection as the RoboCup Standard Platform League robot in 2007 created a self-reinforcing research ecosystem that no other small humanoid has matched. The figure of 3,600+ independent research studies ¹¹ is a company-cited number, but its broad plausibility is supported by the volume of NAO-related publications visible in academic databases and the consistent independent references to NAO as the dominant HRI research platform.

The research use cases span a wide range: human-robot interaction methodology, autism spectrum disorder therapy, physical rehabilitation assistance, elderly care and cognitive stimulation, STEM education effectiveness, robot soccer (RoboCup), speech and audio processing, computer vision, reinforcement learning, and social robotics ethics. This breadth reflects NAO's genuine versatility as a research instrument — a 58 cm bipedal robot with a full sensor suite, open SDK, and ROS2 compatibility is a useful testbed for a wide range of robotics research questions.

Selected Research Evidence

The dossier contains four research papers directly relevant to NAO's technical capabilities:

[20] Autonomous sensorimotor context discovery (arXiv:1608.00737, 2016): Describes unsupervised learning of sensorimotor contingencies on NAO — the robot discovering relationships between motor commands and sensory outcomes without supervisory signals. This is foundational cognitive robotics research. The methodology is sound; the results are research-grade demonstrations, not production deployments.

[21] Open Access NAO (OAN): ROS2-based software framework for HRI (arXiv:2403.13960, 2024): Describes a complete ROS2 software stack for NAO that operates independently of Aldebaran's proprietary APIs. This is practically significant: it provides a cloud-independent operational path for existing NAO hardware and integrates with the broader ROS2 ecosystem. The paper is recent (2024) and directly addresses the cloud dependency risk.

[22] Vision-Guided Robot Hearing (arXiv:1311.2460, 2013): Audio-visual integration for sound source localisation on NAO. Demonstrates the platform's utility for multimodal perception research. The work is methodologically solid within its scope.

[23] Text-Driven Motion Generation on NAO via Reinforcement Learning (arXiv:2506.05117, 2025): The most recent paper in the dossier. Describes a pipeline combining large language model text-to-motion generation with reinforcement learning optimisation for NAO motor control. This is technically ambitious and demonstrates that NAO remains a relevant research platform even as the company has collapsed. The paper is from 2025, post-liquidation announcement, suggesting the research community continues to invest in NAO-based work regardless of the company's fate.

Research Community Resilience

EDITORIAL INFERENCE: The persistence of NAO-based research through Aldebaran's financial difficulties and into the post-liquidation period is significant. It suggests that the platform's value to the research community is sufficiently high that researchers are willing to work around corporate instability — developing cloud-independent software stacks ²¹, continuing to publish NAO-based results ²³, and actively discussing strategies to prevent NAO hardware from becoming obsolete ³¹³²³³. This community resilience is an asset that Maxvision/NAO Robotics SA inherits, but it is not unconditional: if the new entity fails to maintain software compatibility, provide hardware support, or continue NAO development, the research community will migrate to alternative platforms.

The concern about universities accumulating obsolete robots ³¹³²³³ is directly relevant here. Institutions that have invested in NAO fleets — sometimes dozens of units — face a genuine risk that their hardware becomes unsupported. The OAN ROS2 framework ²¹ partially mitigates this by providing a manufacturer-independent operational path, but hardware maintenance, spare parts, and firmware updates remain dependent on whoever controls the NAO IP.

Company-linked papers

This module is being compiled — no data to show yet.

Authors & labs

This module is being compiled — no data to show yet.

Code & simulation

• Open Access NAO (OAN)arXiv / GitHub (referenced)

  ROS2-based open-source software framework enabling HRI applications on the NAO robot independently of manufacturer APIs.

Datasets & benchmarks

This module is being compiled — no data to show yet.

---

06Media Evidence Library: What the Videos Prove

The dossier contains six video sources. Four are directly relevant to Aldebaran products; two (sources ²⁶ and ²⁷) concern unrelated robots (Unitree G1 and an unspecified home robot) and are excluded from this analysis. Sources ²⁸ and ²⁹ concern NEURA Robotics and Jibo respectively and are similarly excluded.

What the Videos Actually Show

[19] Aldebaran Robotics Increases Rate of Innovation Using AWS (YouTube): This is a corporate promotional video describing Aldebaran's use of Amazon Web Services for development infrastructure. It is marketing content, not a technical demonstration. It confirms cloud infrastructure dependency and AWS as a development partner, but provides no evidence about robot capability.

[24] Aldebaran Robotics' Nao (YouTube): An official product demonstration video. Shows NAO performing scripted behaviours: walking, gesturing, speaking, and responding to simple commands. The movements are pre-programmed. The video demonstrates that NAO can execute these behaviours reliably in a controlled studio environment. It does not demonstrate autonomous operation in unstructured environments, generalised task completion, or robustness under real-world conditions.

[25] NAO Next Gen: the new robot of Aldebaran Robotics (YouTube): A product launch video for what appears to be an earlier NAO generation. Shows similar scripted demonstrations — walking, dancing, interacting with a human demonstrator. The production quality is high; the behaviours shown are clearly choreographed. The video is useful for confirming the robot's physical form factor and basic motion capabilities.

Evidence Assessment

EDITORIAL NOTE: This report treats choreographed demonstration videos as evidence of scripted behaviour execution capability only. The behaviours shown in ²⁴ and ²⁵ — walking, gesturing, responding to cues — are genuine capabilities of the NAO platform. They are not evidence of autonomous task completion in unstructured environments, which independent sources consistently describe as beyond NAO's practical capability ³⁴³⁰.

The absence of independent third-party video evidence of NAO or Pepper performing sustained autonomous work in real commercial deployments — as opposed to controlled demonstrations — is itself informative. The most credible evidence of real-world deployment comes from written accounts of institutional use (universities, hospitals, HSBC branches) rather than video documentation of autonomous operation.

Media library

This module is being compiled — no data to show yet.

---

07Commercial Reality

Revenue and Financial Performance

The financial record of Aldebaran under its various owners is one of sustained, substantial loss. The verified figures from the dossier paint a consistent picture:

These figures represent losses across the SoftBank Robotics Europe and URG/Aldebaran eras. The €156 million cumulative deficit from 2019–2022 is particularly striking: it implies that even at the height of Pepper's commercial deployment and with SoftBank's backing, the company was burning cash at a rate that no realistic product revenue could offset.

UNKNOWN: Annual revenue figures are not publicly disclosed in the available dossier. The company was privately held throughout its existence, and no revenue figures appear in the cited sources. The financial picture is therefore one-sided: losses are documented; revenues are not.

Unit Economics and Deployment Scale

Approximately 20,000 NAO units and 17,000 Pepper units were sold across the company's history ⁷¹¹. The 40,000+ figure cited in some company materials [COMPANY CLAIM] likely includes Plato or rounds up; the 37,000 figure from convergent independent sources is more defensible.

At verified pricing:

• 20,000 NAO units at an average of €7,500 = approximately €150 million gross revenue from NAO over ~15 years
• 17,000 Pepper units at an average of €18,500 = approximately €315 million gross revenue from Pepper over ~7 years

EDITORIAL INFERENCE: These are rough estimates based on mid-range pricing and do not account for volume discounts, reseller margins, subscription revenue, or software licensing. But even at these estimates, cumulative gross revenue of approximately €465 million over the company's lifetime would need to cover manufacturing costs (Foxconn for Pepper; presumably contract manufacturing for NAO), R&D, sales, support, and overhead for a company that at various points employed hundreds of people. The €156 million net deficit from 2019–2022 alone suggests that the cost structure was fundamentally misaligned with achievable revenue at the volumes the company was selling.

Customer Base and Deployment Evidence

The deployment evidence is strongest for the education and research segments:

• Universities and research institutions: NAO is the dominant small humanoid research platform globally, with 3,600+ research studies ¹¹ and deployment across hundreds of universities. RobotLAB reports 6,500+ NAO units deployed in North America alone ¹¹.
• RoboCup: NAO has been the Standard Platform League robot since 2008, creating a sustained institutional demand from competing university teams worldwide ³⁴.
• Healthcare: Documented deployments in autism therapy, physical rehabilitation, and elderly care settings, primarily in research or pilot contexts ³⁴.
• Retail and service: HSBC bank branches and SoftBank retail stores in Japan are the most frequently cited commercial deployments ³⁴. The scale of these deployments is UNKNOWN from the available dossier.

The 15% Pepper contract renewal rate cited from 2015 ³⁴ — if accurate — is the single most damning commercial data point in the dossier. It implies that approximately 85% of early Pepper enterprise customers concluded, after their initial contract period, that the robot was not delivering sufficient value to justify renewal. This figure is from a community/podcast source ³⁴ and is not independently verified, but it is consistent with the broader pattern of Pepper's commercial failure and production halt.

Distribution and Channel

RobotLAB is the primary North American distributor and has been the most publicly active in managing customer concerns following the liquidation ¹¹. The company's pledge of "zero disruption" to servicing ¹¹ reflects both its commercial interest in retaining customers and the genuine risk that Aldebaran's collapse could strand thousands of deployed units without support.

The distribution model — selling through educational technology resellers and direct enterprise sales — is appropriate for the institutional market but limits the company's ability to achieve the volume and margin profile of a consumer electronics business. Each NAO sale to a university involves significant pre-sales support, software integration assistance, and ongoing maintenance — cost elements that erode margin on a €7,500 unit.

The Fundamental Commercial Problem

EDITORIAL INFERENCE: Aldebaran's commercial failure was not primarily a technology failure. NAO is a genuinely capable research platform that has enabled thousands of studies and trained a generation of robotics researchers. The failure was a

08Markets and Use Cases

Aldebaran's commercial footprint was built on four distinct market verticals, each with different economics, adoption drivers, and failure modes. Understanding how each vertical actually performed — rather than how it was pitched — is essential to assessing both the company's collapse and the residual value of its IP under Maxvision ownership.

Education and STEM

Education was, by a considerable margin, the most durable and defensible market Aldebaran ever served. NAO's combination of a programmable SDK, Choreographe visual programming environment, ROS2 compatibility, and a robust academic community made it a genuine research and teaching instrument rather than a novelty. The approximately 6,500 NAO units deployed in North America alone through RobotLAB ¹¹ — a figure from a partner press release and therefore to be treated as a ceiling rather than a floor — suggests meaningful penetration into university engineering departments, computer science programmes, and secondary STEM curricula.

The robot's use in special education deserves particular attention. NAO's consistent, patient, and non-judgemental interaction style proved genuinely useful in autism spectrum disorder (ASD) therapy contexts, where the predictability of a scripted robot is a feature rather than a limitation ³⁴. Multiple peer-reviewed studies have used NAO in ASD intervention settings, and this application area generated some of the most credible independent validation of the platform's utility. The robot's small size (58 cm, 5 kg) and low-power motors also meant that safety concerns in paediatric settings were manageable ³⁰.

The economics of the education market, however, were structurally problematic. Universities and schools purchase robots through capital budgets, not recurring revenue streams. A NAO unit priced at €5,000–€10,000 ⁴⁵ is a one-time sale, and the installed base of approximately 20,000 units ⁷ represents a ceiling on addressable revenue from existing customers rather than a recurring annuity. Aldebaran's attempts to introduce subscription-based pricing (RobotLAB lists $239–$289/month RaaS over 36 months ⁵) were a belated recognition of this structural problem, but the model was never adopted at sufficient scale to change the company's financial trajectory.

The risk of robot obsolescence in university settings is now well-documented. Multiple independent analyses published in 2025 warned that institutions face accumulating inventories of NAO units that are difficult to repair, lack manufacturer support, and are increasingly outpaced by newer platforms ³¹³²³³. Ribbon harness failures — a known hardware defect — and the difficulty of sourcing replacement parts compound this problem. The Maxvision acquisition may partially address support continuity, but the 59-engineer team planned for NAO Robotics SA ⁹ is a fraction of what would be required to service a global installed base of 20,000 units.

Research and HRI

With over 3,600 independent research studies citing NAO ¹¹, the robot's contribution to human-robot interaction (HRI) research is not in dispute. NAO became, for roughly a decade, the de facto standard platform for HRI experimentation — a status analogous to what the Baxter robot achieved in manipulation research or what the TurtleBot achieved in mobile robotics. This standardisation had genuine scientific value: it enabled cross-study comparisons, shared codebases, and a cumulative body of knowledge that would have been impossible if every research group used a different platform.

The research market, however, does not generate the revenue required to sustain a manufacturing company. Research grants fund robot purchases, but grant cycles are irregular, budgets are constrained, and universities rarely purchase more than one or two units at a time. The RoboCup Standard Platform League, which used NAO as its mandatory hardware from 2008 onwards, created a reliable if modest demand signal from robotics competition teams worldwide, but this too is a low-volume, price-sensitive segment.

The transition to ROS2 compatibility — documented in the Open Access NAO (OAN) framework paper ²¹ — extended NAO's research utility by decoupling it from Aldebaran's proprietary APIs, which was a double-edged development. On one hand, it made NAO more attractive to researchers who preferred open-source toolchains. On the other, it reduced dependency on Aldebaran's software ecosystem and therefore weakened the company's ability to monetise software services alongside hardware sales.

Healthcare and Social Care

Healthcare represented Aldebaran's most aspirational market claim and its most inconsistent commercial reality. The use cases were genuine: NAO and Pepper were deployed in paediatric wards, geriatric care facilities, and rehabilitation settings, with documented applications in cognitive stimulation for elderly patients, physical therapy guidance, and social engagement for isolated individuals ³⁴. These applications are not trivial — the evidence base for social robots in dementia care and ASD therapy is growing, and NAO's consistent behaviour made it a credible research instrument in these contexts.

The commercial problem was that healthcare procurement is slow, risk-averse, and heavily regulated. Hospitals and care homes are not early adopters. The robots' limitations — inability to climb stairs, open doors, or carry objects ³⁰; frequent hardware failures; cloud dependency for some AI functions ¹ — were disqualifying in any context requiring reliable, unsupervised operation. Healthcare deployments remained largely at the pilot and research stage rather than scaling to operational use. No verified evidence in the dossier confirms large-scale paid deployments in healthcare settings as opposed to research or pilot programmes.

Service and Retail

Pepper's primary commercial rationale was the service and retail sector: customer greeting, information provision, and ambient brand engagement in banks, retail stores, and public spaces. The headline deployments — HSBC branches, SoftBank stores in Japan, museums, and reception halls ³⁴ — generated substantial press coverage and positioned Pepper as the vanguard of a coming wave of service robots.

The commercial reality was considerably less impressive. Only 15% of Pepper customers planned to renew their contracts as early as 2015 ³⁴ — the year of the robot's commercial launch — a figure that, if accurate, represents a catastrophic net promoter signal. Pepper's production halt in approximately 2021 ¹⁸, combined with the company's subsequent financial deterioration, suggests that the service robot market never materialised at the scale required to justify the investment. The robot's wheeled base, chest tablet, and scripted interaction model were adequate for controlled, high-footfall environments with patient customers, but fell short in the messier reality of retail operations where staff time spent managing robot failures exceeded any efficiency gain.

The structural problem with service robotics at Pepper's capability level is that the robot occupied an awkward middle ground: too expensive and unreliable to replace human staff, but not sufficiently capable to justify the cost as a supplementary tool. At €17,000–€20,000 per unit ⁴ plus ongoing subscription costs, the return-on-investment case was difficult to make in any retail context where a tablet kiosk or a trained member of staff could perform the same function more reliably at lower cost.

Summary Assessment

The education and research verticals provided Aldebaran with a stable if insufficient revenue base. The healthcare and service verticals consumed significant sales and marketing resource while generating neither the revenue nor the renewal rates required to sustain the company. This misallocation of commercial effort, combined with the structural shift of the robotics market toward more capable platforms, is a proximate cause of the financial collapse documented in §7.

---

09Competitive Landscape

Aldebaran's competitive position must be assessed across two distinct time periods: the era in which NAO and Pepper were launched (2008–2018), when the company was a genuine pioneer, and the era of its decline (2018–2025), when the competitive landscape shifted dramatically and Aldebaran failed to keep pace.

The Pioneer Era (2008–2018)

When NAO was first sold commercially in 2008, the market for programmable humanoid robots was essentially non-existent. Boston Dynamics was a DARPA-funded research organisation, not a commercial entity. SoftBank's own Pepper did not exist. The academic robotics community had access to expensive custom platforms or primitive kits. NAO filled a genuine gap: a robust, programmable, commercially supported humanoid robot at a price point accessible to universities and research labs.

The Intel Capital funding round of $13 million ¹⁵, announced in 2010, validated this positioning. Aldebaran was, at that moment, a credible pioneer in a nascent market. The subsequent SoftBank acquisition in 2012/2015 ¹⁶¹⁷ provided the capital and distribution required to scale, and the combination of NAO (research/education) and Pepper (service/retail) represented a coherent two-product strategy for different segments of the social robotics market.

The Decline Era (2018–2025)

The competitive landscape that Aldebaran faced by 2020 bore little resemblance to the one it had entered in 2008. Several forces converged to erode its position.

Capability inflation. The broader robotics industry moved rapidly toward more capable platforms. Boston Dynamics' Spot (launched commercially 2020) demonstrated that a robot could be both commercially viable and genuinely capable in unstructured environments. Unitree's G1 and H1 platforms offered bipedal locomotion at price points that undercut Aldebaran's premium positioning. Against these benchmarks, NAO's 25-DOF, 58-cm form factor looked increasingly anachronistic.

Chinese hardware competition. Shenzhen-based manufacturers began producing NAO-comparable social and educational robots at significantly lower price points. Platforms such as Unitree's smaller robots, and a range of Chinese-manufactured educational humanoids, offered similar or superior hardware specifications at 30–50% of NAO's price. This price pressure was structurally difficult for a French manufacturer with European labour costs to absorb.

Software commoditisation. The emergence of large language models (LLMs) and open-source robotics middleware (ROS2) reduced the proprietary value of Aldebaran's software stack. A research group could, by 2023, deploy a NAO with GPT-based conversation ²¹ using open-source tools, without purchasing any Aldebaran software services. The OAN framework paper ²¹ explicitly documents this decoupling. When the software moat erodes, the hardware must justify its price on its own merits — and NAO's hardware, while reliable in its class, was not competitive on a cost-per-capability basis.

Tablet and screen alternatives. In the service and retail sector, the competitive set for Pepper was not other humanoid robots but tablet kiosks, digital signage, and conversational AI interfaces. These alternatives were cheaper, more reliable, and easier to maintain. The 15% contract renewal rate for Pepper ³⁴ reflects this substitution dynamic directly.

Direct Competitors by Segment

The table above is not exhaustive but illustrates the breadth of competitive pressure Aldebaran faced. In no segment was Aldebaran the lowest-cost option, the most capable option, or the best-supported option by 2023. Its competitive moat — the NAO research ecosystem and the 3,600+ published studies — was real but non-monetisable in the short term.

Maxvision's Competitive Rationale

The acquisition of Aldebaran's IP by Maxvision ⁷⁹ makes strategic sense from a Chinese industrial perspective. Maxvision acquires a globally recognised brand, a 20-year IP portfolio, an established research community relationship, and a distribution network — all at distressed-asset prices. The competitive question is whether NAO Robotics SA, operating under Chinese ownership with a 59-engineer team, can produce a NAO7 that is cost-competitive with Chinese-manufactured alternatives while retaining the European brand credibility that justified NAO's premium pricing. That question is unresolved.

Competitive comparison

Robot	Maker	Autonomy	Conf.
iRobot Roomba Combo 10 Max	iRobot	Autonomous	0.90
Mobile ALOHA (Stanford)	Stanford University	Teleoperated	0.90
1X NEO	1X Technologies	Remote-Assisted	0.90

---

10Geopolitical Context and Constraints

The Aldebaran story is inseparable from the geopolitical currents that shaped European technology investment, Chinese industrial strategy, and the global robotics supply chain over the past decade. The company's trajectory — French pioneer, Japanese-owned scale-up, German-funded turnaround attempt, Chinese acquisition — maps almost precisely onto the fault lines of contemporary technology geopolitics.

The SoftBank Chapter: Japanese Capital, French Engineering

SoftBank's acquisition of Aldebaran in 2012/2015 ¹⁶¹⁷ was part of a broader Japanese strategic interest in social robotics as a response to demographic pressure. Japan's ageing population and labour shortage created genuine demand for social and service robots, and SoftBank's deployment of Pepper in its own retail stores was both a commercial experiment and a public demonstration of commitment. The Foxconn manufacturing arrangement for Pepper ³⁴ — producing the robot in Taiwan — reflected the standard Japanese electronics industry model of design-in-Japan, manufacture-in-Taiwan.

This arrangement had structural implications for Aldebaran's European operations. The French engineering team retained design authority but had limited influence over manufacturing costs, supply chain decisions, or the pace of hardware iteration. When SoftBank's enthusiasm for Pepper waned — evidenced by the production halt in approximately 2021 ¹⁸ — the French entity was left with a discontinued flagship product and a parent company that had moved on to other priorities.

The URG Chapter: German Capital, Unrealistic Targets

The 2022 acquisition by URG (United Robotics Group), backed by RAG-Stiftung — the German foundation managing the legacy assets of the Ruhr coal industry — represented an attempt to reposition Aldebaran as a European robotics champion ³⁶. The strategic logic was defensible: European industrial policy increasingly prioritised technological sovereignty, and a European-owned social robotics company with a globally recognised brand and an established research ecosystem had genuine strategic value.

The execution was, by multiple independent accounts, deeply flawed. Employees cited unrealistic two-year profitability targets ³⁴ that bore no relationship to the company's cost structure, competitive position, or market dynamics. RAG-Stiftung's decision to stop funding in August 2024 ³⁶ — triggering the cascade toward bankruptcy — suggests that the foundation concluded the investment was unrecoverable. The net deficit of €156 million accumulated between 2019 and 2022 ¹², and the operating loss of €26 million in 2023 ¹², were not figures that a two-year turnaround plan could plausibly address.

The URG episode illustrates a recurring pattern in European deep-tech investment: strategic rationale is sound, but the capital commitment required to compete with American and Chinese players is systematically underestimated, and the timeline for profitability is systematically compressed by investors who do not understand the product development cycles involved.

The Maxvision Acquisition: Chinese Industrial Strategy

The acquisition of Aldebaran's core IP and engineering team by Maxvision, a Shenzhen-based company operating in biometrics, security, and smart transportation ⁷⁹, raises questions that the available evidence does not fully resolve.

Maxvision is not a robotics company by background. Its acquisition of Aldebaran's assets is consistent with a pattern of Chinese industrial acquisitions of distressed European technology assets — acquiring IP, brand, and talent at prices that reflect financial distress rather than strategic value. The planned NAO Robotics SA entity, with 59 engineers and a French R&D presence ⁹, is structured to preserve European regulatory compliance and brand credibility while transferring IP ownership to China.

The geopolitical implications are significant. NAO is deployed in thousands of universities, research labs, and schools across Europe and North America. The robot's cameras, microphones, and network connectivity create a data collection surface that, under Chinese ownership, is subject to Chinese data governance frameworks including the Data Security Law (2021) and the National Intelligence Law (2017), which require Chinese companies to cooperate with state intelligence activities on request. The cloud dependency noted in the safety and compliance assessment ¹ — where shutdown of cloud services could render robots non-functional — means that Maxvision has, in principle, the ability to affect the operation of robots deployed in sensitive research and educational environments.

This is not a claim that Maxvision intends to misuse this capability. It is an observation that the capability exists, and that European and North American institutions deploying NAO units under Maxvision ownership should conduct their own data governance assessments. The OAN framework ²¹, which enables ROS2-based operation independent of manufacturer APIs, partially mitigates this risk for technically sophisticated users.

European Regulatory and Policy Context

The European Union's AI Act, which entered into force in August 2024, classifies certain AI systems used in education and social care as high-risk applications requiring conformity assessments, transparency obligations, and human oversight mechanisms. NAO's use in autism therapy and educational settings may fall within this classification, depending on the specific application and the degree of autonomy involved. The compliance implications for NAO Robotics SA under Maxvision ownership — a Chinese-owned entity operating under EU law — are not yet resolved and represent a material regulatory risk for the new entity.

The broader European robotics policy context is one of increasing concern about technological dependency on non-European suppliers. The European Commission's robotics and AI strategy documents consistently emphasise the importance of European-owned platforms for sensitive applications. The loss of Aldebaran as a European-owned entity is, from this perspective, a policy failure as much as a commercial one.

---

11The Hype, the Real and the Ugly

Any honest assessment of Aldebaran's twenty-year history must separate three distinct categories of claim: what the company genuinely achieved, what it overstated, and what went demonstrably wrong. The evidence record is sufficiently detailed to make these distinctions with reasonable confidence.

What Was Real

NAO as a research platform. The 3,600+ independent research studies ¹¹ are not a marketing figure — they represent a genuine, cumulative scientific contribution. NAO became the standard platform for HRI research for approximately a decade, enabling cross-study comparisons and a shared experimental infrastructure that accelerated the field. This is a real and lasting contribution that survives the company's bankruptcy.

Educational deployment at scale. Approximately 20,000 NAO units sold across 70+ countries ⁷ is a verified deployment figure, not a pipeline or a letter of intent. Whatever the limitations of the product, it was purchased, deployed, and used by real institutions. The RobotLAB distribution network in North America ¹¹ represents a genuine commercial infrastructure, and RobotLAB's post-liquidation pledge to continue servicing existing units ¹¹ reflects the real value of the installed base.

Safety profile. NAO's 5 kg weight and low-power motors mean that the robot poses minimal physical injury risk in educational and healthcare settings ³⁰. This is a genuine design virtue, not a marketing claim, and it is one reason the robot was deployable in paediatric and geriatric contexts where heavier platforms would face insurmountable safety barriers.

ROS2 and open-source ecosystem. The OAN framework ²¹ and the broader ROS2 compatibility of NAO represent genuine technical contributions. The decoupling of NAO from proprietary APIs extends the platform's research utility and partially insulates the installed base from the consequences of manufacturer failure.

What Was Overstated

Autonomous capability. Company materials and some research papers describe NAO as capable of "autonomous sensorimotor context discovery" and "autonomous model learning without supervisory signal" ²⁰. These descriptions are accurate in the narrow context of controlled laboratory experiments but are deeply misleading as characterisations of the robot's practical capability. NAO cannot climb stairs, open doors, or carry objects ³⁰. It executes scripted or pre-programmed behaviours via Choreographe. Its "autonomy" is the autonomy of a well-programmed vending machine — it responds to inputs without real-time human control, but it does not generalise, adapt, or reason about novel situations in any meaningful sense.

Total units sold. The company's claim of "more than 40,000 robots sold" ¹ is not supported by independent evidence. The better-supported figure of approximately 37,000 (20,000 NAO + 17,000 Pepper) ⁷ is itself a combination of two products with very different commercial trajectories. The 40,000+ figure appears to be a rounded-up marketing claim that may include Plato units or other products.

Service robot market potential. The framing of Pepper as a transformative service robot — capable of revolutionising retail, banking, and hospitality — was never supported by the renewal rate evidence. A 15% contract renewal rate in 2015 ³⁴, the year of commercial launch, was a clear signal that the product was not delivering value in service contexts. The company continued to invest in and market Pepper for several years after this signal was available.

Profitability timeline. URG's two-year profitability target ³⁴ for a company with €156 million in accumulated losses ¹² and a declining product portfolio was not a credible business plan. It was either a failure of due diligence or a deliberate misrepresentation to secure RAG-Stiftung funding. The evidence does not permit a determination of which.

What Went Demonstrably Wrong

Hardware reliability. Ribbon harness failures are a known, recurring hardware defect in NAO units ³⁰. The difficulty of repair — noted consistently across independent community and commerce sources — is not a minor inconvenience but a structural problem for a robot deployed in educational settings where technical support is limited. A product that fails frequently and is hard to repair is not commercially viable at any price point.

Cloud dependency. The risk that cloud service shutdown could render deployed robots non-functional ¹ is a design choice that prioritised capability over resilience. For an educational or research robot deployed in institutions with multi-year planning horizons, this dependency was always a liability. The bankruptcy and liquidation of 2025 made this risk concrete: institutions with NAO units faced genuine uncertainty about whether their robots would continue to function after the company ceased operations.

Financial management. The accumulation of €156 million in net losses between 2019 and 2022 ¹², followed by a €26 million operating loss in 2023 ¹², represents a failure of financial governance that cannot be attributed solely to market conditions. The company was burning cash at a rate that its revenue base could not support, and neither SoftBank nor URG imposed the financial discipline required to bring costs into alignment with revenue.

Workforce management. The termination of 106 employees ³⁶ following liquidation, after a period in which the company had already cut nearly half its workforce ⁶¹⁰, represents a significant human cost. The employees who remained through the URG period did so under conditions of financial uncertainty that the company's leadership had an obligation to communicate honestly. The evidence record suggests this obligation was not fully met.

Claim tracker

Approximately 20,000 NAO and 17,000 Pepper units were sold across 70+ countries, totalling ~37,000 robots (vs. the company's own claim of 40,000+).Unknown

The ~37,000 figure (20,000 NAO + 17,000 Pepper) is cited consistently across multiple independent news and community sources [2][3][7][34], lending it credibility, but the underlying data originates from company-reported figures repeated by press — no independent audit or third-party sales verification is present in the dossier; the company's own higher claim of 40,000+ [1] is likely a rounded-up marketing figure that may include Plato units.

Pepper was a commercially viable service robot for retail and banking, with broad customer adoption (e.g., HSBC banks, SoftBank stores Japan).Not supported

Independent sources directly contradict commercial viability: only 15% of Pepper customers planned contract renewal as early as 2015 [34], production was halted in 2020–2021 [3][7][18], and the company accumulated net deficits exceeding €156M between 2019–2022 [12][13] — deployment at named retailers is confirmed but sustained commercial success is not, and the robot's limited robustness for commercial use is a consistent independent criticism [30][31][34].

NAO has been used in 3,600+ independent research studies, making it the dominant platform in HRI and educational robotics research.Supported

Multiple independent academic and news sources [21][31][32][33][34] corroborate the 3,600+ research studies figure and NAO's widespread use in HRI labs and universities globally; peer-reviewed arxiv papers [20][21][22][23] themselves constitute independent evidence of active research deployment, though the exact count of 3,600+ is a company-originated figure repeated by independents rather than an independently audited number.

NAO's hardware is robust enough for repeated use — its 5 kg weight and low-power motors mean falls cause no damage.Not supported

An independent Hacker News community source [30] explicitly states NAO can fall without damage due to its 5 kg mass, but this is directly contradicted by multiple independent sources [31][34] citing frequent hardware failures including ribbon harness tears as a well-known chronic problem, and the dossier notes NAO is described as hard to repair — indicating the robot is not reliably robust in practice.

Aldebaran's liquidation and Maxvision acquisition will result in continuity of NAO robot development, with NAO V7 in development and a new entity NAO Robotics SA planned with 59 engineers.Unknown

Multiple news sources [7][9][11][13] confirm the Maxvision acquisition of core IP (~July 19, 2025) and report plans for NAO Robotics SA and NAO V7, but these are forward-looking announcements from the acquirer with no independent verification of actual engineering progress, funding adequacy, or timeline — RobotLAB's pledge of service continuity [11] is a partner statement, not independent confirmation of product development.

Aldebaran's financial collapse was caused primarily by URG's underinvestment and unrealistic 2-year profitability targets, not fundamental product failure.Unknown

Independent news and community sources [6][10][34] cite employee accounts attributing the collapse to URG's unrealistic targets and withdrawal of RAG-Stiftung funding in August 2024, but the dossier also shows net deficits of €156M accumulated under SoftBank ownership (2019–2022) [12], suggesting structural product/market problems predating URG — the causal attribution to URG alone is not independently verified and is contested by the broader financial record.

NAO supports ROS2-based autonomous HRI capabilities independently of manufacturer APIs, including GPT-based conversation and advanced sensor fusion.Unknown

An independent peer-reviewed arxiv paper [21] specifically documents the Open Access NAO (OAN) ROS2 framework enabling HRI applications independent of manufacturer APIs, and [23] demonstrates text-driven motion generation via reinforcement learning on NAO — these are credible independent research demonstrations, but they represent experimental lab capabilities, not validated production-level autonomous performance; cloud/network dependency for GPT functions is an additional reliability caveat noted in the dossier.

---

12Future Scenarios

The acquisition of Aldebaran's core IP by Maxvision ⁷⁹ creates a branching set of possible futures for the NAO platform, the installed base, and the broader social robotics market. The following scenarios are assessed on the basis of available evidence; probabilities are editorial inferences, not forecasts.

Scenario A: Successful Continuity Under Maxvision (Probability: Low-Moderate)

In this scenario, NAO Robotics SA is established as a functioning entity with 59 engineers ⁹, NAO7 is developed and released within 18–24 months, and Maxvision's manufacturing capabilities reduce the cost of NAO hardware sufficiently to restore price competitiveness. The French R&D presence preserves European regulatory compliance, and the brand credibility of NAO in the research and education markets is maintained.

The conditions required for this scenario are demanding. Maxvision must demonstrate genuine commitment to the robotics business rather than treating the acquisition as an IP extraction exercise. The 59-engineer team must be retained and supplemented. NAO7 must offer meaningful capability improvements over NAO6 — particularly in hardware reliability and software openness — to justify the purchase price in a market where Chinese competitors offer comparable hardware at lower cost. European and North American institutions must be willing to purchase robots from a Chinese-owned entity, which is not guaranteed given the geopolitical context described in §10.

The evidence base for this scenario is thin. Maxvision's background in biometrics and security ⁷ does not suggest deep robotics expertise. The planned entity structure is known, but no verified evidence of actual engineering activity, product roadmap, or customer commitments has been published.

Scenario B: Brand Preservation, Capability Stagnation (Probability: Moderate)

In this scenario, Maxvision maintains the NAO brand and provides basic support for the existing installed base, but NAO7 is delayed or released as an incremental update rather than a meaningful capability advance. The research and education market continues to use NAO6 units until they fail, at which point institutions migrate to alternative platforms. NAO Robotics SA becomes a support and distribution entity rather than a genuine R&D organisation.

This scenario is consistent with the pattern of Chinese acquisitions of distressed European technology brands in other sectors, where brand value is preserved for export market credibility while R&D investment is redirected to domestic priorities. It is also consistent with the limited engineering team size (59 engineers) relative to the scope of work required to develop a new robot generation, maintain the existing platform, and service a global installed base.

Under this scenario, the NAO research ecosystem gradually migrates to alternative platforms — most likely ROS2-compatible open-source platforms or newer Chinese-manufactured educational robots — over a three-to-five-year horizon.

Scenario C: Fragmentation and Ecosystem Collapse (Probability: Low-Moderate)

In this scenario, Maxvision fails to establish NAO Robotics SA as a functioning entity, cloud services are discontinued, and the installed base of 20,000 NAO units becomes progressively non-functional. Universities and research institutions face the robot obsolescence crisis already documented in independent analyses ³¹³²³³. The NAO research ecosystem fragments as groups migrate to incompatible platforms, and the cumulative investment in NAO-specific software, curricula, and research infrastructure is written off.

The OAN framework ²¹ partially mitigates this scenario by enabling ROS2-based operation independent of manufacturer cloud services. Institutions that have adopted OAN or equivalent open-source frameworks are insulated from cloud dependency. Those that have not are vulnerable.

RobotLAB's post-liquidation pledge to continue servicing existing units ¹¹ provides some continuity for North American customers, but RobotLAB is a distributor, not a manufacturer, and cannot produce replacement parts or develop new software features independently.

Scenario D: NAO7 as a Competitive Platform (Probability: Very Low)

In this scenario, Maxvision invests substantially in NAO7 development, producing a robot with meaningfully improved locomotion, manipulation, and AI capabilities that competes with the new generation of Chinese and American educational robots. This would require not only the 59-engineer team but substantial additional investment in hardware development, AI integration, and manufacturing scale.

The evidence base for this scenario is essentially absent. The robotics market in 2025–2026 is moving toward platforms with significantly more capable locomotion (bipedal walking with dynamic stability), manipulation (dexterous hands), and AI integration (on-device LLM inference) than NAO6 offers. Closing this capability gap would require investment of a scale that Maxvision has not signalled and that the distressed-asset acquisition price does not suggest.

Scenario Summary

The most likely outcome, on current evidence, is a combination of Scenarios A and B: a functioning but under-resourced NAO Robotics SA that maintains the brand and provides basic support, releases an incremental NAO7 on a delayed timeline, and gradually loses market share to more capable and better-supported alternatives. This is not a catastrophic outcome for the installed base in the short term, but it is not the revival that the "Pepper and Nao saved" headline ⁹ implies.

---

13What to Watch: A Live Monitoring Checklist

The following indicators are the most informative signals for tracking Aldebaran's post-liquidation trajectory and the viability of the Maxvision acquisition. Analysts and institutions with NAO deployments should monitor these on a quarterly basis.

Corporate and Legal

• NAO Robotics SA registration. Has the new legal entity been formally registered in France? What is its registered capital, and who are the named directors? French company registry (Infogreffe) filings are publicly accessible and would confirm whether the planned entity has been constituted.
• Maxvision financial disclosures. Maxvision is a Shenzhen-based company; its financial statements and any regulatory filings related to the Aldebaran acquisition may be accessible through Chinese corporate registries or Hong Kong Stock Exchange disclosures if the company has listed subsidiaries.
• IP transfer confirmation. Has the transfer of Aldebaran's patent portfolio to Maxvision been recorded in the European Patent Office and USPTO databases? Patent assignment records are publicly searchable and would confirm the scope of the IP transfer.
• Employee retention. Are the 59 engineers named in the acquisition plan still employed by NAO Robotics SA six months post-acquisition? LinkedIn activity and French employment filings (DARES) may provide indirect evidence.

Product Development

• NAO7 announcement. Any official announcement of NAO7 specifications, pricing, or availability timeline. Distinguish between press releases (company claims) and verified product listings with confirmed delivery dates.
• Hardware reliability improvements. Any independent reports of ribbon harness or other known failure mode remediation in NAO7. Community forums (RoboCup mailing lists, ROS Discourse, academic HRI forums) are the most reliable early indicators.
• Software ecosystem updates. Choreographe version updates, SDK releases, and ROS2 compatibility patches. GitHub commit activity on Aldebaran-related repositories is a leading indicator of engineering activity.
• Cloud service continuity. Any announcements regarding the continuity or migration of cloud-dependent NAO and Pepper services. Institutions should test cloud-dependent functions periodically to detect service degradation.

Commercial

• New customer announcements. Any verified (not press-release-only) new customer deployments under Maxvision/NAO Robotics SA ownership. Distinguish between pilot programmes and paid commercial deployments.
• RobotLAB inventory and pricing. RobotLAB's product listings ⁵ are a real-time indicator of NAO availability and pricing in the North American market. Removal of NAO from active listings would signal supply chain disruption.
• RoboCup Standard Platform League. The RoboCup organisation's decision on whether to continue using NAO as the mandatory platform for the Standard Platform League is a significant demand signal. Any announcement of a platform change would accelerate the research ecosystem migration.
• Academic citation trends. Annual counts of new peer-reviewed publications citing NAO as a platform (trackable via Google Scholar or Semantic Scholar) are a lagging but reliable indicator of research community engagement.

Geopolitical and Regulatory

• EU AI Act compliance. Any regulatory guidance or enforcement action related to NAO's use in high-risk AI Act categories (education, social care). The European AI Office's published guidance on high-risk AI systems is the relevant primary source.
• Data governance assessments. Any published institutional assessments of data governance risks associated with NAO under Maxvision ownership. Academic institutions in the EU and UK are increasingly required to conduct data protection impact assessments (DPIAs) for AI systems; published DPIAs for NAO would be informative.
• Export control developments. Any EU or US export control measures affecting Chinese-owned robotics companies operating in sensitive research environments. The US Bureau of Industry and Security (BIS) Entity List and EU dual-use export control updates are the relevant sources.
• Competitive platform announcements. New educational or social robot platforms announced by European or North American manufacturers that could serve as NAO alternatives for institutions concerned about Chinese ownership. This is a substitution risk indicator for Maxvision's market position.

---

14Sources and Methodology

Methodology

This report was produced using a structured evidence-grading framework applied to a research dossier of 35 numbered sources gathered as of 21 June 2026. Sources were classified into four categories — official (company-published), commerce (distributor/reseller), research (peer-reviewed or primary academic), news (independent journalism), video (audiovisual documentation), and community (forums, podcasts, social media) — and weighted accordingly.

All factual claims in the report are graded using the following evidence labels:

Choreographed demonstration videos are not treated as proof of autonomous capability. Partnership announcements