4-19. Reading NEURA Robotics' Physical AI Strategy
Is NEURA Robotics the leading contender for European humanoids? 4NE1, cognitive robots, and German manufacturing AI
Western Physical AI 2026 Complete Edition Part 19
Series: Western Physical AI 2026 Complete Edition
Subtitle: 'Top 20 Western AI Companies + Extras' as seen through humanoid robots, autonomous driving, industrial robots, and robot foundation models
Evaluation criteria: AI division fame / Capital strength / Technical capability / Profitability / AI division value
In this report, we distinguish between company announcements, third-party reports, future plans, and external estimates. Content claimed by the company itself is clearly marked as 'according to the company' or 'in company announcements,' and figures reported by news organizations are noted as 'based on reports.' Plans that have not yet been realized are written in the future tense to avoid confusion with actual achievements. Monetary amounts are roughly converted at 1 USD = 160 JPY and 1 EUR = 170 JPY (as of August 2026).
Before reading | What makes NEURA different from other humanoid companies is that it was generating revenue from industrial robots before releasing a humanoid.
NEURA Robotics was founded in 2019 in Metzingen, Germany. It is a regional city with a population of just over 20,000.
The company is known for its 4NE1 humanoid, but before that, it had a product lineup consisting of the MAiRA and LARA collaborative robots and the MAV AMR, claiming an order backlog of 1 billion euros.
In June 2026, it announced a Series C round led by Tether, with Qualcomm, Amazon, NVIDIA, Bosch, Schaeffler, and the European Investment Bank participating.
First, the conclusion: The lineup of investors tells the story of the European humanoid strategy itself.
Reading the shareholder list provides more information than the monetary amount. AI chips, two of the world's largest industrial parts manufacturers, a logistics giant, public finance, and a cryptocurrency company.
The supply chain necessary for the humanoid business to succeed has become the investor base itself.
However, the full amount is conditional on achieving performance milestones, and the definition of the 1 billion euro order backlog has not been disclosed. The figure of 5 million units by 2030 is a goal, not a plan.
Table of Contents
Chapter 1: From the town of Hugo Boss to a 7 billion dollar valuation
Chapter 2: 5-axis evaluation—Capital is top-tier, but the reality of revenue is opaque
Chapter 3: The founder, David Reger
Chapter 4: The substance of the term 'cognitive robotics'
Chapter 5: MAiRA—The world's first commercial cognitive robot
Chapter 6: LARA, MAV, MiPA—An Overview of the Product Portfolio
Chapter 7: NEURA Omnisensor—Technology to Eliminate Fences
Chapter 8: 4NE1—The Humanoid Named "For Anyone"
Chapter 9: 4NE1 Gen 3.5 and 4NE1 Mini—Pricing and Shipping Dates
Chapter 10: CES 2026—Porsche Design and Quadruped Robots
Chapter 11: Neuraverse—An App Store for Robots
Chapter 12: NEURA Gym—Where to Create Training Data
Chapter 13: Funding History—From Buying Back Chinese Capital to $1.4 Billion
Chapter 14: What the List of Investors Tells Us
Chapter 15: From China to Germany—The 2024 Production Transfer
Chapter 16: €1 Billion in Orders and the Figure of 5 Million Units by 2030
Chapter 17: Competitive Landscape—US Players, European Players, and the Connection to Chapter 20
Chapter 18: Conclusion—Is NEURA Robotics the Leading Contender for European Humanoids?
Chapter 1: From the Town of Hugo Boss to a $7 Billion Valuation
In the southwestern German state of Baden-Württemberg, there is a town called Metzingen. It has a population of just over 20,000. It is a provincial city that is otherwise unremarkable, known for being the headquarters of the fashion brand Hugo Boss and for its outlet mall.
It is about a 40-minute drive from Stuttgart. Scattered around the area are Bosch, Mercedes-Benz, Porsche, and countless parts manufacturers. It is fair to call this region the heart of German manufacturing.
In this town, there is a company that, as of 2026, is valued at **$7 billion (approximately 1.12 trillion yen)** (based on reports; the company has declined to comment on its valuation).
NEURA Robotics. Founded in 2019. It is a company that makes the humanoid robot "4NE1".
On June 10, 2026, the company announced a Series C round of up to $1.4 billion (approximately 224 billion yen). The lead investor is Tether, a major stablecoin issuer. Alongside them are Qualcomm, Amazon, NVIDIA, Bosch, Schaeffler, and the European Investment Bank (EIB).
This list of investors is more eloquent than the amount itself. The largest AI chip maker, two of the world's largest industrial parts manufacturers, a hyperscale logistics operator, and a cryptocurrency company. The supply chain components necessary for a humanoid business to exist have become the investors themselves.
Seven years since its founding. This level of capital has been gathered from a regional city in Europe.
David Reger stated the following at the announcement: Many people believed that globally significant AI infrastructure companies could only be born in Silicon Valley. However, I believe that the next generation of AI leaders can emerge from anywhere in the world, provided there is sufficient vision, engineering talent, and execution speed (summary).
This is a clear response from Europe to the US players we have looked at so far in this series—Figure AI, Apptronik, 1X, and Agility.
Chapter 2: 5-Axis Evaluation—Top-tier Capital, Opaque Revenue Reality
【AI Division Fame】 8.7 / 10
【Capital Strength】 9.4 / 10
【Technical Capability】 8.9 / 10
【Profitability】 7.5 / 10
【AI Division Value】 8.8 / 10
【Overall Evaluation】 8.7 / 10
I ranked capital strength highly because the Series C in June 2026 was up to $1.4 billion, which is outstanding for a European robotics company. However, the full amount is conditional on achieving performance milestones and is not a lump-sum payment.
Profitability remains in the middle because the company is private and sales/profits are not disclosed. While there was an announcement in January 2025 that 'sales increased 10-fold in the last year,' the absolute amount is unknown. The figure of 1 billion euros in order backlog also lacks a breakdown of whether these are confirmed contracts or include expressions of interest.
Regarding technical capability and AI division value, I evaluate the identification of people and objects using the patented NEURA Omnisensor, as well as the ecosystem concept called Neuraverse. On the other hand, the incentive for other companies' hardware to join Neuraverse is weak, and there are limits to its expansion as a platform.
Fame has risen significantly due to the unveiling of the Studio F.A. Porsche design at CES 2026.
Chapter 3: The Founder, David Reger
NEURA Robotics was founded on March 26, 2019, by David Reger. The names of Mavarick H and Sugeeth Gopinathan (currently Vice President of Project & Product Management) are recorded as co-founders.
Reger comes from the robotics industry. He is often described as someone who grew up in the German machinery industry and saw the limitations of robots on the front lines.
His argument is consistent: Current robots have neither eyes nor ears.
Industrial robots move exactly as they are taught. They do not know what is around them. That is why they must be isolated from humans and enclosed by safety fences. This has driven up the cost of automation and stripped away flexibility.
Reger stated the following at the Series B announcement in January 2025: We expect cognitive robotics to become bigger than smartphones. I am proud that NEURA is the first to ship commercially viable cognitive cobots and is one of the most important players in humanoid robotics (summary).
'Bigger than smartphones.' This type of statement is common in the robotics industry, but in Reger's case, what makes him different is that he already has products with sales. NEURA is a company that does business with industrial cobots before humanoids.
And there is one more characteristic decision by Reger: he bought back Chinese capital himself. This will be detailed in Chapter 13.
Chapter 4: The Content of the Term 'Cognitive Robotics'
NEURA positions itself as a 'pioneer of cognitive robotics' and calls itself the 'creator of the first cognitive robot.'
This term was effectively defined by the company and is not an industry-standard term. Therefore, it is necessary to confirm its content.
According to the company's explanation, a cognitive robot is a machine that does the following.
See: Identifies and tracks objects in real-time using advanced machine vision.
Hear: Processes voice commands from any direction using a 360-degree microphone array.
Feel: Detects contact through force/torque sensing and artificial skin.
Learn: Learns from observation and experience to update its behavior.
And most importantly, it interacts with humans without coding. You can recognize hand movements and teach it without any programming experience. You can give instructions by voice. The company claims that this method can reduce training time by up to 80% (company claim).
There is also a description that it processes multilingual instructions with over 95% accuracy (company claim). In global manufacturing sites, it is not uncommon for workers to have different native languages. This point is a practical selling point.
What is important here is that NEURA's "cognition" is not LLM-based general-purpose reasoning. At least in terms of its technical lineage since its founding, it centers on real-time environmental adaptation through sensor fusion and machine learning.
While Skild AI from the 6th installment of this series and Physical Intelligence from the 7th are attempting to create "intelligence independent of the body" from the research side, NEURA has taken the approach of first building a body equipped with sensors and then building intelligence on top of it.
This difference in order is also reflected in the design of the Neuraverse, which will be described later.
Chapter 5 MAiRA—The World's First Commercial Cognitive Robot
At the center of NEURA's product lineage is **MAiRA (Multi-sensing Intelligent Robotic Assistant)**. The company positions this as the "world's first commercially available cognitive robot."
MAiRA takes the form of a collaborative robot arm. However, what makes it different from regular cobots is that the sensors are integrated into the main unit.
360-degree vision sensors
Spatial audio processing via AI-driven microphone array
Real-time environmental adaptation via on-device neural networks
Pressure and contact detection
According to the company's explanation, MAiRA can be trained with a step-by-step wizard to learn to recognize and identify objects. With repeated training, it becomes able to autonomously pick up different objects. It is intended for a wide range of applications such as laboratories and logistics.
And when MAiRA is mounted on the AMR "MAV" described later, it becomes a mobile manipulator. Applications such as warehouse organization and supermarket shelf replenishment are cited in the company's explanation.
The applications the company lists as expected include gluing and dosing, machine tending, palletizing, sanding and polishing, quality inspection, and welding. Industries include food and beverage, logistics, medical, metal processing, and transportation equipment.
Before talking about humanoids, I want to establish the existence of this product lineup. NEURA is not a humanoid company that creates buzz with videos. It is a company that sells robot arms used in factories, and it is building humanoids as an extension of that.
Chapter 6: LARA, MAV, MiPA—An Overview of the Product Portfolio
As of 2026, NEURA's product line consists of the following.
LARA (Lightweight Agile Robotic Assistant)
A lightweight collaborative robot arm launched in November 2020. Its maximum payload is 18kg. It is used in applications such as automotive, aerospace, and e-mobility, and allows for intuitive programming via hand-guiding. The company states that it can reduce setup time by up to 80% and ensures collision-free interaction through patented non-contact human detection. It is intended for manufacturing, healthcare, logistics, and the service industry.
MAV (Multi-sensing Autonomous Vehicle)
An autonomous mobile robot (AMR). It is designed with the premise of working alongside humans. It has a maximum payload capacity of 1.5 tons and is intended for use in areas such as automotive manufacturing. By combining it with LARA or MAiRA, it transforms fixed workstations into mobile workflows.
MiPA (My Intelligent Personal Assistant)
A cognitive home and service robot. It takes the form of a mobile manipulator. Its market launch was announced at Automatica 2025 in June 2025. While not a humanoid, it is a product designed to work in living spaces.
LiWA (Lightweight Robotic Arm) / The Ai Scan Hub
Listed as a product on the company's website.
4NE1 / 4NE1 Mini
Humanoids. Detailed in Chapter 8 and beyond.
NEURA Quadruped Robot
A quadruped robot. Unveiled at CES 2026. Covered in Chapter 10.
Add-ons
NEURA OmniSensor, NEURA SenseKit, NEURA Teach, and NEURA Touch.
This breadth is something not found in startups specializing solely in humanoids. At the same time, it raises the question of whether development resources are being spread too thin. This point needs to be evaluated in conjunction with the scale of funding mentioned later.
Furthermore, the product lineup has won various design awards. These include the Red Dot Design Award, German Design Award, German Innovation Award, UX Design Award, and European Design Award (according to company announcements). True to German manufacturing, they value both design and user experience.
Chapter 7: NEURA Omnisensor—Technology for Eliminating Safety Fences
The technology that the company repeatedly cites as its technical core is the patented NEURA Omnisensor.
In a nutshell, the function of this sensor is to distinguish between humans and objects.
Under industrial robot safety standards, if there is a possibility of a person entering the robot's operating area, light curtains, safety fences, or force/speed limitations are required. Many collaborative robots ensure safety by using a "stop upon contact" method, but this means significantly reducing operating speed.
What the Omnisensor aims for is the step before that. Before contact, it determines whether what is approaching is a person or a workpiece. If it is a person, it slows down; if it is an object, it maintains normal speed.
The company explains that with this technology, 4NE1 can collaborate directly with humans without safety fences. It reliably distinguishes humans from objects, recognizes them, and adapts its behavior accordingly.
The sensor configuration includes 360-degree 3D vision, depth cameras, LiDAR, force/torque sensing, artificial skin, and non-contact human detection. It is designed to ensure safe interaction through sensor skin that detects before contact and force feedback.
I believe this is the substance of what NEURA means by the word "cognition." It is not about abstract intelligence, but rather perception for removing safety fences.
And whether or not safety fences can be removed determines the factory layout itself. If fences are not needed, floor space is freed up. Humans and robots can share the same traffic lines. It can be retrofitted into existing production lines. The hurdles to implementation are physically lowered.
This is precisely what the German manufacturing industry is looking for.
Chapter 8: 4NE1—A Humanoid Named "For Anyone"
4NE1 stands for "For Anyone." It is written with the number 4 and the letters NE1, and officially does not include a hyphen. While early reports sometimes referred to it as "4NE-1," the company's current notation is "4NE1."
The 3rd generation was unveiled to the world in June 2025 at Automatica 2025 held in Munich. According to the company's explanation, 4NE1 is a humanoid that can work autonomously and safely with humans in real-world environments.
The main specifications (based on company announcements and reports) are as follows.
Height: 180 cm
Payload: 100 kg. The company positions this as the highest level among general-purpose humanoids.
Dual battery system designed for 24-hour continuous operation. Based on reports, it supports hot-swapping for 6–8 hours of operation.
Learning-based hand capable of fine motor movements
The computing platform is based on the NVIDIA Jetson Thor (T5000) series, with a configuration that combines it with Isaac GR00T.
Operation without protective fences
Identification of humans and objects via Omnisensor
The 100 kg payload figure is outstanding for a humanoid. Many of the US companies we have looked at in this series list a payload of around 20–25 kg as their specification. This difference reflects a difference in design philosophy.
Many US companies are optimizing in the direction of "substituting for tasks humans can do," such as moving tote boxes in warehouses or handing over parts. NEURA's 100 kg is closer to the idea of using it in factories for tasks that humans cannot do.
The applications the company envisions are said to extend beyond complex industrial tasks to activities in the home and service industries.
Chapter 9: 4NE1 Gen 3.5 and 4NE1 Mini—Price and Shipping Timing
As of 2026, 4NE1 is in the pre-order stage. We have organized the information that can be confirmed from the company's reservation page.
4NE1 Gen 3.5
Expected availability is end of 2026
The reservation fee is 100 euros (approx. 17,000 yen). It is fully refundable and will be applied toward the final purchase price.
The shipping unit includes an integrated high-dexterity hand, a high-capacity battery, and a dedicated charging station.
Specialized end-effectors or third-party end-effectors are not included.
Reservations are for individuals. Industrial and commercial quantities are handled via separate inquiries.
You can request a full refund of the reservation fee at any time before signing the final purchase agreement.
The company does not specify the unit price on the reservation page. According to reports, it is 98,000 euros (approx. 16.66 million yen) for a single unit, and 60,000 euros (approx. 10.2 million yen) for bulk orders of 20 or more units. Since this figure has not been verified firsthand, it should be clearly stated as being based on reports when used in an article.
4NE1 Mini
Height: 132 cm
Two configurations: Standard and Pro
25 degrees of freedom (based on reports)
Expected availability is during 2026 (shipping in spring/April according to reports)
The price based on reports is 19,999 euros (approx. 3.4 million yen)
Intended for classrooms, laboratories, and prototyping environments.
Equipped with the same AI and comprehensive perception as the foundation model, opening up applications for research, education, and entertainment (company description).
Stair climbing is currently under development. Operation on flat and uneven terrain is the current scope.
This statement that 'stair climbing is under development' is written by the company on its official page. As a public relations move for a humanoid company, it is worth noting that they are clearly stating what they cannot yet do.
Furthermore, the company also states that it may make improvements to the hardware and design to ensure performance and ease of use ahead of the 2026 release. Specifications are subject to change.
Chapter 10: CES 2026—Porsche Design and Quadruped Robots
On January 5, 2026, at CES 2026 held in Las Vegas, NEURA showcased three things.
First, the next-generation 4NE1 machine. The design is a collaboration with Studio F.A. Porsche—the design studio that worked on the Porsche 911. Visitors were able to see it up close, take selfies with it, and interact with it.
Employing a Porsche design studio for a humanoid. This decision carries more meaning than mere differentiation. If humanoids are to enter not just factories but living spaces, design becomes a condition for acceptance. It is as if a German company has brought in its country's strongest design asset.
Second, the 4NE1 Mini.
Third, the NEURA Quadruped Robot—a four-legged robot.
This third one is the most important in the context of this series.
The company explained that these new robots represent a major leap in real-world robotic capabilities and compete directly with existing Asian products. And it positions NEURA as the first supplier in the West in this field.
It is clear what "existing Asian products" refers to. It is the group of four-legged robots led by China's Unitree.
And at the same time, this is also the market for Switzerland's ANYbotics, which we will cover in the next installment, the 20th. ANYbotics was born out of ETH Zurich in 2016 and has entered oil and gas, mining, power generation, and metal sites with its industrial inspection quadruped robot, ANYmal.
The leading European humanoid has descended into the market of the leading European quadruped inspection robot. This composition serves as the best connection point for reading through the 19th and 20th installments.
Spencer Huang of NVIDIA and David Kehr of Schaeffler are listed as guest speakers. These two are also representatives of the companies that later became Series C investors.
Reger stated the following regarding CES 2026: The world is preparing to welcome robots, but the robots themselves also need a world that is prepared to welcome them. Through Neuraverse, we are creating an "invisible engine" that connects, empowers, and scales intelligence across all devices and environments (summary).
Chapter 11 Neuraverse—The App Store for Robots
Neuraverse is the robotics ecosystem that NEURA launched at Automatica 2025 in June 2025.
The design philosophy is clear. It is designed like an app store, allowing learning to be shared in a networked manner between different types of robots (company description).
Breaking this down, the concept is as follows:
Partners develop applications (skills) for NEURA robots
Development can take place in industrial, service, or home sectors
Developed skills are reused across different models
What one robot learns is transferred to other robots
The company explains that its platform integrates all relevant sensors and components with artificial intelligence into a single device. Partners build applications on top of that device.
Here, a comparison with other installments of this series is effective.
Skild AI (Part 6): Creating a 'robot brain' that fits any body as a foundation model
Physical Intelligence (Part 7): Aiming for a general-purpose robot OS with the π model
Intrinsic / Google (Part 18): Flowstate, a hardware-agnostic development environment, and perception foundation models
NVIDIA (Part 1): Building from the ground up with Isaac / GR00T / Omniverse
NEURA: Starting from in-house hardware to build an ecosystem on top of it
Only NEURA calls itself a platform while manufacturing its own hardware. This is both a strength and a weakness.
The strength is the ability to design consistently from sensors to software. Sensor fusion, the core of cognitive robotics, cannot be fully controlled without owning the hardware.
The weakness is that there is little incentive for other hardware manufacturers to join Neuraverse. To use the Android analogy, Neuraverse is in a position similar to 'Samsung, which makes Android.' Other device manufacturers do not want to adopt a competitor's OS.
However, in 2024, NEURA joined NVIDIA's Humanoid Robot Developer Program and announced that it would use NVIDIA Isaac in combination with Neuraverse. It is not designed to be completely closed.
And there is another, less flashy but important participation. In 2025, the company joined the PROFIBUS & PROFINET International (PI) SRCI Working Group. SRCI is an initiative to define a standardized interface between robots and corresponding PLCs.
PROFINET is an industrial communication standard that has developed primarily around Siemens, which we covered in the previous Part 16. In other words, while NEURA promotes its own ecosystem, it is simultaneously taking a seat on the side of existing factory standards.
As we saw in Part 16, for a robot to work in a factory, it must be able to communicate with existing PLCs. No matter how smart a robot is, if it cannot communicate, it will not be adopted. Not many companies understand this point.
Chapter 12: NEURA Gym—Where to Create Learning Data
Announced at the same time as Neuraverse was the concept of NEURA Gym.
This is a physical training center where data is generated from actual application scenarios (company description).
The biggest bottleneck in physical AI is data. While LLMs could learn from text on the internet, robot motion data is nowhere to be found. Someone has to actually move a robot to collect it.
Each company's answer to this problem is different.
Teleoperation: Humans operate the robot and record its trajectory
Simulation: Generating massive amounts in a virtual environment (NVIDIA Isaac, Cosmos series)
Observation of human movement: Learning from videos of human work
Collection from actual deployment: Gathering data while operating at customer sites
NEURA Gym sits between the first and fourth methods.A dedicated facility is built to replicate actual application scenarios, where data is then collected.
The advantage of this method is that the quality of the data can be controlled. The disadvantage is that the scale is limited by the number of facilities. It is not designed to win on volume against US and Chinese competitors who earn their data through the number of actual deployments.
As for specific facilities, the TUM RoboGym in collaboration with the Technical University of Munich (TUM) can be cited. Established at Munich Airport with a scale of approximately 2,300 square meters, it is a concept where universities and companies operate robots in the same space to collect data.
Universities, a vast real-world space like an airport, and companies. This form of collaboration between these three parties is also a typical German approach to industry-academia cooperation.
Furthermore, the company states that as of 2026, it has opened offices in Hangzhou and Zurich, beginning its global expansion.
Hangzhou is a hub for the Chinese robotics industry, and Zurich is the center of European robotics, home to ETH Zurich.
Partnerships are also cross-industry. They have formed strategic alliances with home appliance maker Vorwerk in the smart home sector, as well as with major South Korean shipbuilder HD Hyundai Samho. In August 2025, they also held an announcement in Taipei. Factories, homes, shipyards. The scope of application the company envisions is intentionally broad.
Chapter 13: History of Fundraising—From Buying Back Chinese Capital to $1.4 Billion
The history of NEURA's fundraising well illustrates the character of this company. Let's follow it in order.
Early Startup Phase: Chinese Capital
NEURA raised approximately $80 million (about 12.8 billion yen) in its early stages, all of which came from a strategic supporter, the Chinese conglomerate Han's Group. While known for real estate development, the company also operates in hotel management, equipment manufacturing, and healthcare.
July 2023: $55 million, and a buyout. An additional $16 million in the autumn of the same year
Raised $55 million (about 8.8 billion yen) from Lingotto (an investment management firm under Exor N.V.), Vsquared Ventures, Primepulse, and HV Capital. At this time, NEURA bought out its previous supporters, cleaned up its cap table, and paved the way for pure financial investors.
Furthermore, within less than three months, they raised an additional $16 million (about 2.5 billion yen) from the US private equity firm InterAlpen Partners. The purpose was to enter the US market.
Mr. Reger explained at the time: In today's de-globalized world, we determined that NEURA has much greater potential globally as an independent company. A partnership form is more beneficial for both sides (summary).
Looking back from 2026, this decision was decisive. If Chinese capital had remained a major shareholder, it is doubtful whether NVIDIA, Qualcomm, or Amazon could have invested under US export controls and investment regulations. As the technological divide between the US and China deepens, the restructuring of the capital composition at this point has extremely significant meaning.
January 2025: Series B 120 million Euros
Led by Lingotto Investment Management. Participating investors include BlueCrest Capital Management, Volvo Cars Tech Fund, InterAlpen Partners, Vsquared Ventures, HV Capital, Delta Electronics, C4 Ventures, L-Bank, and founder David Reger himself.
At this point, the company announced that it had doubled its workforce to over 300 people and increased its sales tenfold in the last year.
March–June 2026: Series C
Since reports vary here, I will record this carefully.
One report states that approximately 1 billion euros (about $1.2 billion) closed in March 2026, with a valuation of approximately 4 billion euros.
Multiple reports on June 10, 2026, state a Series C of up to $1.4 billion, with a valuation of approximately $7 billion. The company declined to comment on the valuation.
The lead investor is Tether Holdings. Participants include Qualcomm Technologies, Amazon, NVIDIA, Bosch, Schaeffler, the European Investment Bank (EIB), imec.xpand, Lingotto Horizon, InterAlpen Partners, and others.
Nafea Bshara, Vice President and Distinguished Engineer at Amazon, stated regarding this investment that it is a natural extension of the existing partnership between the company and NEURA. In other words, this is not Amazon's first involvement as an investor.
There is an important caveat. The full funding is contingent upon achieving certain milestones based on company performance (as reported by multiple sources familiar with the matter). In short, $1.4 billion is the ceiling, not a lump sum payment.
According to the corporate information service Tracxn, the cumulative funding from 5 rounds and 17 investors is $1.68 billion, and the number of employees is 567 as of May 2026 (third-party data).
However, in the company's announcement at the time of Series C, it is stated that they employ over 1,400 people from more than 45 countries. The company's figures are more recent, and it is reasonable to use them as the benchmark. Tracxn's figure of 567 is likely outdated.
The company explains that it in-houses core technologies such as perception systems, motion control, materials, and embedded intelligence. The scale of 1,400 employees is a reflection of that in-house policy.
The company positions this Series C as the largest funding round in history for a full-stack robotics company, making it the most well-funded humanoid robot company in Europe.
Chapter 14: What the Investor Lineup Tells Us
There is more information to be gained from reading the composition of the investors than from the amount of money.
NVIDIA — Computing infrastructure and robot foundation models. NEURA already uses NVIDIA Isaac and participates in the Humanoid Robot Developer Program. As seen in the first installment, NVIDIA invests widely in leading physical AI companies, and the investment in NEURA is part of that effort.
Qualcomm (Qualcomm Ventures) — Chips and edge AI. Humanoids cannot rely on the cloud for all inference. Low-latency processing on the machine is essential, and that is Qualcomm's domain.
Amazon — Logistics and distribution. Amazon Robotics, which we looked at in the 12th installment of this series, has its own fleet of robots for its warehouses. The significance of Amazon investing in a European humanoid company cannot be explained by simple financial investment alone.
Bosch — The world's largest automotive parts manufacturer and a giant in sensors and actuators. This company can manufacture many of the parts that NEURA needs.
Schaeffler — Bearings and precision mechanisms. Humanoid joints are bundles of reducers and bearings. The bottleneck for mass production is not AI, but the supply capacity of these mechanical parts—a point repeatedly confirmed when discussing the humanoid industry.
Furthermore, in January 2026, a strategic partnership for the development and mass production of humanoid robots for industrial production was also announced with Bosch. Investment and business collaboration are proceeding simultaneously.
European Investment Bank (EIB) — The EU's policy financing institution. This is a sign that Europe positions robotics as a strategic industry.
Tether — Stablecoin issuer. They are directing the massive profits gained from cryptocurrency into real-economy assets. The reason for taking the lead is difficult to explain beyond investment returns.
Volvo Cars Tech Fund, Delta Electronics (Taiwan) — Continuing from Series B.
L-Bank — State bank of Baden-Württemberg. Local public financing is involved.
To summarize this composition in one phrase, it has everything: parts, chips, logistics, and public funding. While US humanoid companies have capital structures centered on VCs and tech giants, NEURA has secured the physical supply chain of the manufacturing industry as shareholders.
This is different from the capital structure of ANYbotics seen in Chapter 20—where the company is funded by its customers in the oil and gas industry—and it represents another model within Europe as well.
Furthermore, according to Dealroom data, the capital raised by robotics companies since the beginning of 2026 has reached $55.8 billion, nearly doubling the record from the previous year. While the majority of this comes from companies in the U.S. and China, new robotics companies are also emerging in Europe, such as the SoftBank-backed German firm Agile Robots and the UK-based Humanoid.
The total funding for European robotics companies was 1.6 billion euros in 2025, an increase of 110% from 761 million euros in 2024.
Chapter 15: From China to Germany—The 2024 Production Relocation
Alongside the restructuring of its capital, a key decision for NEURA was the relocation of its production.
In February 2024, the company moved its production from China to Germany. According to reports by Sifted, part of the reason for moving the majority of production back to Germany was to demonstrate confidence in Germany as a business location and to prove that Europe can produce leading robotics companies.
It is difficult to explain this decision based solely on economic rationality. It is cheaper to manufacture in China. The supply chain is also more robust. Reducers, motors, and sensors are all easier to procure in China.
Even so, there are three conceivable reasons for the return to Germany.
First, customer requirements. For German and European manufacturing customers, the production location can be a procurement requirement. In the automotive industry in particular, there is a trend toward requiring geographical visibility of the supply chain.
Second, geopolitical risk. As the technological decoupling between the U.S. and China progresses, bringing products manufactured in China into the U.S. market carries risks related to tariffs and regulations. Moves like the FCC's Covered List, which we have covered in this series, target Chinese robotic equipment itself.
Third, political messaging. As a company that accepts European public funding (EIB, L-Bank), production within Europe is part of its accountability.
Regarding production capacity, Mr. Reger told the Financial Times that a short-term goal is to increase robot production capacity from 6,000 units this year to tens of thousands next year (statement as of June 2026).
Chapter 16: 1 Billion Euros in Orders and the 5 Million Unit Figure
There are two figures that NEURA repeatedly cites. They need to be treated with caution.
1 billion euros (approx. 170 billion yen) in order backlog
At the time of its Series B announcement in January 2025, the company stated it had already secured a 1 billion euro order backlog. At the time of its Series C in June 2026, it was explained that the total of the order backlog and deployment pipeline exceeds $1 billion. Other reports have described it as a "multi-billion euro order pipeline," and both the units and definitions are inconsistent.
The customer list includes the name **Kawasaki (Kawasaki Heavy Industries)**. The fact that a major Japanese industrial conglomerate is included as a customer is seen as an indication that the commercial traction extends beyond Europe.
However, the content of the term "order backlog" varies by company. Is it a firm contract, or does it include letters of intent? When are the delivery dates? The company has not disclosed this breakdown. The 1 billion euro order figure is not an audited financial number.
Mass production of "multi-million" units by 2030
This is a goal the company set forth in its Series C announcement. While a specific figure of 5 million units has appeared in media reports, the company's official expression is "multi-million," and it has not promised a specific number of units. To grasp the scale of this figure, I will provide a comparison.
According to International Federation of Robotics statistics, the annual global installation of industrial robots has generally hovered around 500,000 units in recent years. In other words, NEURA is saying that it will single-handedly produce ten times the annual global installation of industrial robots.
Given that their current production capacity is 6,000 units per year and their goal for next year is tens of thousands, they need to increase this by several hundred times over the four years leading up to 2030.
This is a goal, not a plan. When covering this in an article, that distinction should be clearly stated.
Chapter 17: Competitive Landscape—US Players, European Players, and the Connection to Part 20
I will organize NEURA's position in relation to other parts of this series.
US Humanoid Players
: Figure AI (Part 5), Apptronik (Part 9), Agility Robotics (Part 8), 1X (Part 10), and Sanctuary AI (Part 11). What makes NEURA stand out among these is the fact that it is already generating revenue with industrial cobots. While Agility is advancing real-world deployments in warehouses and Apptronik in automotive factories, NEURA already has established commercial channels with other products.
Boston Dynamics (Part 4)
: Atlas's industrial deployment and partnership with Google DeepMind. While they lead in technical maturity, their track record in mass production is yet to come.
Agile Robots (Germany, backed by SoftBank)
: A competitor from the same country, Germany. Although not covered in the main body of this series, it is a name that should be listed when discussing European humanoids.
Humanoid (UK)
: Another emerging force in Europe.
Hexagon Robotics (AEON)
: It has been reported that BMW adopted the AEON unit from Hexagon Robotics for the deployment of humanoid robots at its Leipzig plant starting in the summer of 2026. This should be recorded as a case where a German automaker chose someone other than Germany's own NEURA. As seen in Part 5, BMW is also conducting trials with Figure AI at its Spartanburg plant.
Universal Robots / Teradyne (Part 17)
: They compete head-on in the collaborative robot market. The UR+ ecosystem built by UR and NEURA's Neuraverse are both aiming to be the same kind of "robot app store." However, there is a difference: UR uses a method of connecting third-party peripherals to its own arms, while NEURA uses a sensor-integrated approach.
ABB (Part 15), Siemens (Part 16)
: Giants of European factory automation. If NEURA is to enter factories, the question will be how it reconciles with the commercial channels of these two companies.
And ANYbotics (Part 20)
: With the NEURA Quadruped Robot unveiled at CES 2026, the two companies are now in direct competition in quadruped robots. ANYbotics has focused on specific applications like industrial inspection for nearly a decade, building an entry barrier known as Ex certification (explosion-proof). Whether NEURA can stand on the same playing field will be one of the subjects of the next installment.
Chapter 18: Conclusion—Is NEURA Robotics the Leading Candidate for European Humanoids?
First, Japanese companies are already partnering with NEURA.
NEURA is collaborating with Kawasaki Robotics (USA) on the CL series of cobots, and in May 2024, at Automate 2024, it announced a strategic partnership with OMRON aimed at innovating manufacturing efficiency through AI-driven cognitive automation technology. Kawasaki is also listed as a customer.
In other words, major Japanese robot manufacturers are treating NEURA's technology not just as a "competitor" but also as a "source of procurement." How this relationship evolves is a concrete issue for the Japanese robotics industry.
Second, there is the treatment of the word "cognition."
Japanese industrial robots have conquered the world through precision, reliability, and maintainability. Micron-level repeatability, durability that lasts for 10 years, and service networks spread across the globe. These are assets that cannot be imitated.
What NEURA has brought is a different axis from that one. Seeing, hearing, feeling, learning. And by doing so, removing the safety fences.
When competition arises along this axis, how will Japanese manufacturers respond? In sensor technology, Keyence, Omron, and Sony are at a world-class level. In components, Harmonic Drive and Nabtesco still maintain a strong hold on reducers. The elements are all there.
What is missing is an entity that integrates these and presents them as a platform. I wrote the same thing in the 18th installment about Intrinsic. Japan lacks the vertical integration of industrial software seen in companies like Siemens.
Third, there is the lesson of capital.
If NEURA had not bought back its Chinese capital in 2023, the 2026 funding from NVIDIA, Qualcomm, and Amazon would likely not have materialized. In a deglobalized world, where you receive capital from determines where you can sell.
This is a question that applies equally to Japanese robot startups.
And finally.
There is a company that reached a valuation of 1 trillion yen in seven years, starting from a town of 20,000 people called Metzingen. Bosch and Schaeffler have invested in it, and the state bank and the European Investment Bank are listed alongside them. Local manufacturing, public finance, and Silicon Valley capital are all on the same shareholder register.
This is the answer that Europe has provided in 2026. Whether it will succeed or not is still unknown. However, they are further ahead than regions that did not provide an answer.
Confirmation as of August 14, 2026
Confirmed facts (consistent across company announcements and multiple reports)
Founded in 2019, headquarters in Metzingen, Germany. Founder and CEO is David Reger
Initially raised approximately $80 million from China's Han's Group, bought it back in 2023 to clean up the cap table
July 2023, raised $55 million from Lingotto and others
February 2024, production moved from China to Germany
October 2024, joined the NVIDIA Humanoid Robot Developer Program
January 2025, Series B of 120 million euros. Led by Lingotto Investment Management
June 24, 2025, announced 4NE1 3rd generation, MiPA, and Neuraverse at Automatica 2025
January 5, 2026, unveiled 4NE1 next-generation machine (Studio F.A. Porsche design), 4NE1 Mini, and NEURA Quadruped Robot at CES 2026
June 10, 2026, announced Series C. Led by Tether, with participation from Qualcomm, Amazon, NVIDIA, Bosch, Schaeffler, EIB, etc.
Product lines include MAiRA, LARA, MAV, MiPA, LiWA, 4NE1, 4NE1 Mini, and quadruped robot
Collaborated with Kawasaki Robotics (USA) on the CL series cobots and partnered with Omron Robotics and Safety Technologies (2024).
Opened offices in Hangzhou and Zurich.
Based on reports and third-party data (not verified by primary sources).
Series C scale and valuation: Reports suggest approximately 1 billion euros with a valuation of about 4 billion euros in a March closing, while other reports from a June announcement suggest up to 1.4 billion dollars with a valuation of about 7 billion dollars. The company has not commented on the valuation.
Full funding is contingent upon achieving performance milestones (according to sources familiar with the matter).
Over 1,400 employees from more than 45 countries (company announcement at the time of Series C). Tracxn's figure of 567 (May 2026) appears to be outdated.
Founded on March 26, 2019.
Following 55 million dollars in July 2023, an additional 16 million dollars was raised from InterAlpen Partners in the fall of the same year (aimed at US expansion).
Series C participants include imec.xpand, Lingotto Horizon, and InterAlpen Partners. Amazon's Nafea Bshara commented that it is a 'natural extension of existing partnerships.'
In January 2026, announced a strategic partnership with Bosch for the development and mass production of humanoids for industrial manufacturing.
In 2025, joined the SRCI working group of PROFIBUS & PROFINET International (PI) (standardizing the interface between robots and PLCs).
Strategic partnerships with Vorwerk (smart home) and HD Hyundai Samho (shipbuilding). Announced in Taipei in August 2025.
In May 2024, announced a strategic partnership with OMRON at Automate 2024.
Explains that perception systems, motion control, materials, and embedded intelligence are developed in-house.
The 2030 goal is stated as 'multi-million' units in company announcements. The 5 million unit figure is based on reports.
Order backlog and deployment pipeline exceeded 1 billion dollars at the time of Series C (company announcement). Other reports cite a multi-billion euro scale, indicating inconsistency in units and definitions.
Information suggests the 4NE1 computing platform is based on the NVIDIA Jetson Thor (T5000) series + Isaac GR00T.
4NE1 Mini has 25 degrees of freedom (based on reports).
TUM RoboGym with the Technical University of Munich (Munich Airport, approximately 2,300 square meters).
4NE1 price: 98,000 euros per unit, 60,000 euros for 20 units or more.
4NE1 Mini price: 19,999 euros
4NE1 operating time: 6-8 hours, supports hot-swapping
Production capacity: 6,000 units in 2026, tens of thousands in 2027 (Reger's statement to the FT)
Total funding for robotics companies in 2026: $55.8 billion (Dealroom)
European robotics funding: 1.6 billion euros in 2025, 761 million euros in 2024
Company claims, unverified figures
Order backlog of 1 billion euros (breakdown and definition not disclosed)
Goal of manufacturing 5 million units by 2030
80% reduction in training time, over 95% accuracy in multilingual processing
Positioned as the "world's first commercial cognitive robot" and "first Western quadruped supplier"
Unconfirmed items
Revenue, operating profit/loss (not disclosed as it is a private company; only an announcement of "10x growth" as of January 2025 exists)
Number of 4NE1 units actually deployed, customer sites currently in operation
Specifications, price, and shipping date of the NEURA Quadruped Robot
Number of development partners participating in Neuraverse
Number and locations of NEURA Gym installations
Regarding notation
The official notation is 4NE1 (no hyphen). "4NE-1" is the notation used in initial reports.
Next Preview: Reading ANYbotics
In the 20th installment, we will look at the Zurich-based quadruped robot company ANYbotics, the world's first explosion-proof legged inspection robot ANYmal X, and the European-style capital structure where "customers invest." This is the final installment of the main TOP 20 series.
Key Reference Materials and Official Links
Reserve 4NE1: The Cognitive Humanoid Robot (NEURA Official - Reservation Page)
Reserve 4NE1 Mini: Compact Humanoid Robot (NEURA Official - Reservation Page)
NEURA Robotics at CES 2026 (NEURA Official, January 5, 2026)
NEURA Robotics Secures €120 Million in Series B Funding (NEURA Official, January 2025)
Humanoid robotics company Neura Robotics backed by Amazon, Nvidia (CNBC, June 10, 2026)
Neura Robotics raises $1.4bn backed by Amazon, Nvidia and Qualcomm (Sifted, June 10, 2026)
Neura Robotics Raises Up to $1.4 Billion (TechTimes, June 11, 2026)
Neura adds additional $55M to expand product line (The Robot Report, July 2023)
Neura Robotics picks up $55M to ramp up in cognitive robotics (TechCrunch, July 19, 2023)
Source Correspondence Table
| Description | Source Classification |
| Founded 2019, Metzingen, founder David Reger | Consistent across company announcements and multiple reports |
| Initial funding of approx. $80M from Han's Group and buyback | Based on reports (TechCrunch, Reger's comments) |
| $55M funding in July 2023 | Company announcement/reports |
| Production transfer (China to Germany) in February 2024 | Company announcement/reports |
| Series B €120M (January 2025) and investor composition | Company announcement |
| Over 300 employees, 10x revenue growth (as of Jan 2025) | Company announcement |
| €1B order backlog | Company claim (breakdown not disclosed) |
| Series C scale and valuation | Based on reports (multiple theories on amount and valuation; company did not comment on valuation) |
| Subject to milestone conditions | Multiple reports citing insiders |
| Cumulative funding $1.68B, 567 employees | Third-party data (Tracxn, May 2026) |
| Over 1,400 employees, 45+ countries | Company announcement (at time of Series C; newer than Tracxn's 567) |
| Founding date March 26, 2019 | Established fact |
| $16M from InterAlpen Partners (Fall 2023) | Company announcement |
| Comments from Nafea Bshara (Amazon) | Company announcement |
| Strategic partnership with Bosch (January 2026) | Company announcement/reports |
| Participation in SRCI working group (2025) | Publicly known fact |
| Partnerships with Vorwerk, HD Hyundai Samho | Company announcement |
| Expression 'multi-million by 2030' | Company announcement (5 million units is report-based) |
| Order backlog/pipeline over $1B | Company announcement (other reports mention billions of euros) |
| Jetson Thor (T5000) + Isaac GR00T configuration | Based on reports |
| 25 degrees of freedom for 4NE1 Mini | Based on reports |
| TUM RoboGym (Munich Airport, approx. 2,300 sqm) | Company announcement/reports |
| 4NE1 height 180cm, payload 100kg, dual battery | Company announcement |
| 4NE1 operation 6-8 hours, hot-swappable | Based on reports |
| 4NE1 price (€98k/€60k), Mini €19,999 | Based on reports (official page does not disclose prices) |
| Reservation fee €100, fully refundable, Gen 3.5 expected end of 2026 | Company announcement (official reservation page) |
| 4NE1 Mini stair climbing under development | Company announcement |
| NEURA Omnisensor patent acquisition and person/object identification | Company announcement |
| 80% reduction in training time, 95%+ accuracy in multiple languages | Company claim |
| Unveiling of quadruped robot at CES 2026 and 'first Western supplier' | Company announcement |
| Neuraverse, NEURA Gym, MiPA (Automatica 2025) | Company announcement |
| Participation in NVIDIA Humanoid Robot Developer Program | Company announcement |
| Collaboration with Kawasaki, Omron | Based on reports/company announcement |
| Goal of 5 million units by 2030 | Company goal (not a plan) |
| Production capacity 6,000 units -> tens of thousands | Reger's statement to FT (based on reports) |
| Total robotics funding $55.8B, Europe €1.6B | Third-party data (Dealroom) |
| Adoption of Hexagon Robotics AEON at BMW Leipzig | Based on reports |
| Implications for Japanese industry, lessons on capital structure | Author's view |
Final Confirmation Reflection | NEURA is shifting from a robot company to an AI platform company
The value of NEURA Robotics is not limited to the 4NE1 hardware.
By integrating cognitive robots, Neuraverse, learning environments, and industrial applications, they are aiming to become a robot-version ecosystem company.
While Figure AI and Tesla compete on mass production capabilities, NEURA is leveraging its connectivity with European manufacturing as its weapon.
Hashtags
#WesternPhysicalAI2026
#NEURARobotics
#4NE1
#HumanoidRobot
#CognitiveRobotics
#Neuraverse
#PhysicalAI
#PhysicalAI
#GermanManufacturing
#DavidReger
#CollaborativeRobot
#MAiRA
#Tether
#NVIDIA
#Bosch
#Schaeffler
#EuropeanInvestmentBank
#QuadrupedRobot
#IndustrialRobot
#Metzingen

