Content/Sophisticated

Edited by/Yonge

Proofread by/Mangfu

On the cusp of 2026, robots are poised to usher in the 'era of anthropomorphism.'

In November, the Zhiyuan Expedition A2 robot achieved a Guinness World Record by walking 106.286 kilometers, marking it as the first humanoid robot to complete an 'endurance run.' Simultaneously, the Xiaopeng IRON robot strutted down the catwalk with postures that closely mimicked those of a real human. These robots are no longer mere assemblies of mechanical arms and sensors; they are increasingly resembling humans in their movements and behavioral patterns.

The expansion of technological boundaries has ignited collective enthusiasm across the industrial chain. However, amidst a surge of cross-border entrants and soaring financing, mass production remains delayed, and technological implementation lacks momentum. Is this robot wave, fueled by both capital and technology, the driving force behind the next industrial revolution, or merely another speculative bubble destined to burst?

Part.1

The Third Pole of Industrial Products Enters Phase 2.0

Elon Musk once asserted in a conversation with Jensen Huang, 'AI will keep you busier, and humanoid robots will become the largest industry ever.' Despite the absence of truly versatile and mature products, this has not dampened the world's richest man's optimism about the robotics industry's prospects.

Indeed, since the Turing test in the 1950s and 1960s, which posed the question of whether robots could think, robots have gradually transitioned from the realm of science fiction movies to become the third standardized industrial product with trillion-dollar potential, following smartphones and automobiles.

An industry practitioner with extensive experience in the consumer electronics sector told TrueView that the uniqueness of robots lies in their ability to bridge the gap between C-end consumption and B-end industry, enabling bidirectional empowerment and value creation for both sectors.

Simply put, unlike smartphones, which primarily focus on personal communication, and smart cars, which emphasize travel scenarios, robots can serve as both living assistants in households and production laborers in factories. This dual-track attribute allows them to meet the dual demands of consumption upgrades and industrial upgrades.

The practitioner further pointed out that it is precisely this dual-track attribute, combined with the breakthrough explosion of AI technology around 2020, that has propelled the robotics industry from the conceptual exploration phase of 1.0 to the rapid entry into the scaled development phase of 2.0. The industrial landscape of Phase 2.0 is characterized by three significant features: rapid market growth, deep integration into scenario practices, and the coexistence of enthusiasm and bubbles.

In terms of industry scale, the growth potential of the robotics track is evident. A report released by the Development Research Center of the State Council predicts that China's embodied intelligence market (with robots as its main industrial carrier) is expected to reach 400 billion yuan by 2030 and exceed 1 trillion yuan by 2035. Counterpoint Research believes that robot shipments will maintain a compound annual growth rate of over 20% in the next five years, with revenue expected to reach nearly 128 billion US dollars by 2029.

Among them, industrial robots have a relatively broader market. The International Federation of Robotics' recently released 'World Robotics Report 2025' shows that in 2024, 542,000 new industrial robots were installed globally, more than doubling from a decade ago. China installed 295,000 new units, accounting for more than half, with an average annual growth rate of at least 20% expected in the future.

Entering Phase 2.0, the core hallmark is the shift from single-function implementation to deep adaptation across multiple scenarios. Early robots were mostly 'functional tools'; industrial robots could only perform simple repetitive welding and handling tasks, while household robots could only complete basic operations like sweeping and mopping, with extremely poor scenario adaptability.

Nowadays, with the iteration of AI large models and sensor technologies, robots are becoming smarter and more flexible. Midea's Kuka collaborative robots can accurately recognize worker movements, enabling human-machine collaborative operations with an error margin controlled within 0.1 millimeters. In medical scenarios, surgical robots can perform minimally invasive surgeries with precision far exceeding manual operations.

Goldman Sachs previously predicted that humanoid robots would be first applied in factories from 2024 to 2027 and in consumer markets from 2028 to 2031. From the current pace of industrial advancement, Goldman Sachs' predictions appear overly conservative.

The transition from being usable to being useful not only amplifies the industrial value of robots but also allows more players to see opportunities, setting the stage for track congestion.

Part.2

The Dual Faces of the Industry: Enthusiasm Amidst a Bubble

The other side of the robotics 2.0 era is the 'coexistence of fire and ice' in the industry. On one hand, there is a fervent celebration of capital and technology; on the other, there is the cold water of bubble controversies. This contrast constitutes a typical feature of the development of emerging industries.

The enthusiastic side is manifested in three aspects: financing, players, and technological iteration.

In most cases, capital serves as a precursor to trending industries, with its profit-driven nature giving it an innate sense of smell. Public data shows that according to IT Juzi, in the first three quarters of this year alone, the financing volume in the domestic robotics sector has reached 38.624 billion yuan, 1.8 times the total financing volume of 21.254 billion yuan for the entire year of 2024.

With 'embodied intelligence' being included in the national strategic emerging industries, a new wave of investment is on the way. Meanwhile, unlike the angel investment approach of previous years, investors may now focus more on the commercialization efficiency and monetization speed of manufacturers.

The deployment of capital directly fuels another aspect of the enthusiasm: track congestion, especially evident in the terminal product interface downstream of the robotics industrial chain.

Automakers, smartphone manufacturers, internet companies, and home appliance giants are all crossing over into the field. Robots seem to have become an industry where 'anyone can give it a try,' a commercial landscape rarely seen in most emerging industries before.

A senior executive from Zhiyuan once pointed out, 'We expect it to undertake repetitive and boring tasks such as delivery, sorting, and reception in scenarios like retail, catering, corporate front desks, and even family services, deeply integrating into our daily lives as trustworthy partners for humans.'

Looking back, from the perspective of the layout paths of relevant companies, the market has shifted from short-term thinking of chasing trends to deep cultivation in niche areas. For example, traditional manufacturing enterprises leverage technological homogeneity to extend their reach across borders, while technology companies focus on breakthroughs in niche scenarios. This differentiated layout reduces industrial trial-and-error costs, accelerates collaboration across the entire industrial chain, makes industry development more resilient, and forms a dual-wheel-drive pattern of core component research and scenario application implementation.

This has also driven the continuous acceleration of technological iteration. For instance, the integration of AI large models with robots is becoming increasingly close. Natural language processing and computer vision technologies make robot interactions smoother, while the precision of core components such as servo motors and reducers continues to improve.

However, the flip side of the influx of hot money is the breeding of bubbles and the accumulation of risks.

Goldman Sachs clearly pointed out in its '2025 Global Robotics Industry Outlook' that there is currently a significant overvaluation phenomenon in the robotics track. Globally, about 60% of robotics companies have valuations exceeding 100 times their revenue, and the global robotics industry may face an overcapacity rate of 25% by 2025.

A large number of companies lack core technologies and rely solely on assembly and branding to ride the hype. To some extent, the current bustling scene of the robotics track is quite similar to the O2O (Online to Offline) trend in 2015. It may look lively, but many companies are engaged in pseudo-innovation without truly addressing scenario pain points.

Over time, this has led robots into a dilemma. On one hand, there is a significant deviation between market expectations and actual development; on the other hand, the real difficulty they face in completing open tasks such as fine hand operations is widely underestimated, with the challenges of technological breakthroughs far exceeding existing perceptions. Additionally, referring to the new energy vehicle industry, robots face internal competition and homogenization while also grappling with prominent issues such as insufficient national standards, vague safety boundaries, and ethical controversies.

As a result, since 2025, the embodied robotics track has simultaneously entered a period of retreat, with multiple companies successively announcing shutdowns. For example, K-Scale Labs, a humanoid robot startup established just a year ago, declared closure; Aldebaran, a company with 20 years of industry experience and the creator of the Pepper robot, could not escape bankruptcy; Embodied, a U.S. AI companion robot company, officially terminated operations. In October of the same year, domestic embodied intelligence startups also suffered losses, with OneStar Robot (OneStar) reported to have disbanded.

Even CloudMinds, which once had a valuation of up to 3 billion US dollars, found itself in a predicament of a broken capital chain and had to initiate a contraction strategy to save itself.

Part.3

Mass Production and Technology: Which Is the Core Constraint?

When the tide recedes, the core contradiction of corporate shutdowns points to the commercialization bottleneck. Which is the key constraint hindering industrial progress and corporate competitiveness: mass production or technology?

Logically, mass production can drive price reductions and market penetration. However, the premise of mass production is not a lack of manufacturing capabilities or untested business models; the fundamental bottleneck still lies in immature technology.

In other words, robots may still lack a 'dedicated large model' distinct from general-purpose large models. Especially in key areas such as motion control and autonomous navigation, the technological chain relying on intelligent perception, decision-making, and execution is not yet proficient. This is also the basis for Elon Musk's emphasis that robots are not yet fully practical.

A dedicated large model can enable robots to receive end-to-end instructions and better execute tasks. Wang Xingxing, the founder of Unitree Technology, has repeatedly emphasized that the bottleneck for robot scaling is the insufficiency of AI models and has called on the industry to accelerate the arrival of the 'ChatGPT moment' for embodied intelligence.

The next question is, how to break through at the model level?

Some industry insiders emphasize that this essentially tests the R&D capabilities of all players in the embodied intelligence track, with 'R&D resource redundancy' being the core factor affecting the efficiency of developing dedicated large models for robots.

Currently, entrants can be roughly divided into two categories: cross-border players and startups focused on the track. The former, while supported by their main businesses, often struggle to allocate R&D resources to dedicated large models for robots; the latter, while focused, are often constrained by funding shortages and unable to keep up with the iteration pace of large models.

In contrast, teams with sufficient R&D resource reserves can plan the R&D path for dedicated large models for robots in advance, simultaneously determine product layouts and R&D progress, allocate core human resources and sufficient R&D funds ahead of time, and seize technological opportunities. They can take the lead in model advancements in core scenarios such as motion control and autonomous navigation. As for supply chain construction, material reserves, and marketing, these can be gradually advanced as second-priority items.

Therefore, as the ebb and flow of capital become more rapid, we may need to re-examine the essence of the robotics track. This is not a bubble game of rapid birth and death but a long-term test of industrial patience and strategic resilience.

The current 'congestion' is precisely a typical characteristic of the pre-explosion phase of the industry. It does not indicate a wrong direction but rather announces the end of simple models. The era of gaining valuations through assembly, branding, and concept hype is coming to an end. The real competition has just begun. The ticket to the second half lies in the 'brain' of the robots.

The future industry landscape will also be divided accordingly. Teams with abundant resources capable of diligently tackling the 'perception-decision-execution' closed loop will define the 'intelligence' of machines; while players merely mimicking forms and lacking core AI capabilities may run aground in the technological deep waters.

Therefore, for the robotics industry, we should have more long-term patience and less short-term frenzy for chasing trends. Its future does not depend on the number game of financing volumes but on whether intelligence can emerge at the critical model layer.

Whoever can first enable robots to truly understand the world and autonomously respond to complex environments will truly hold the key to unlocking the next era.