In the battle of embodied intelligence, big tech companies are always on the scene.

The Bubble Fades, and Big Tech Enters the Scene

Rarely has a new technology track been as "hyped" as embodied intelligence. Players are flocking in one after another, and platforms for "creating brains" and "building bodies" are emerging in waves. The scene looks prosperous, but there is also an undercurrent of anxiety. Is it a long - term competition of capabilities or just another relay race of concepts?

On August 4th, China Hi - Tech Industry Guide interviewed several experts to discuss the investment motivation and bubble phenomenon in the field of embodied intelligence.

Among them, Li Junlan, a research manager at IDC China, believes that "there is indeed a certain structural 'bubble' in the field of embodied intelligence at present. Some projects may receive high valuations when the products are not yet mature and the application verification is limited. However, the overall industry is still in a stage of reasonable expansion driven by technology."

In addition, Lu Jiaqing, a senior partner at Guoke Jiahe, and Kang Lidong, the managing director of Wanchuang Capital, also said in the interview that there is a certain bubble in embodied intelligence currently.

When big tech companies come together to build platforms, complete the toolchain, connect the data closed - loop, and establish a cooperation ecosystem, they are essentially raising the industry threshold, squeezing out the speculative space, and creating more fertile ground for vertical players who can truly deliver results.

Yuan Foyu, the vice - president of Baidu Group, said in a speech on July 26th that based on the common industry needs and the challenges in implementation, Baidu Smart Cloud focuses on four major areas: embodied brain, motion control cerebellum, construction of embodied data sets, and R & D of the whole - machine body to assist enterprises in the R & D of key technologies and products in embodied intelligence.

Comprehensive information from Tianyancha media shows that the popularity of embodied intelligence is no longer just about showing off models at press conferences. It has entered a stage that truly tests the implementation ability, systematic ability, and organizational ability. Standardized, platform - based, and reusable technology bases for embodied intelligence are the key levers to suppress bubbles such as redundant development and performance - chasing, guiding capital and talents towards effective innovation.

As Li Junlan believes in the research "Embodied Intelligence 'Breaking the Circle' + The Rise of Chinese Power: The Robot Competition Enters the 'Chinese Time'", China has become an important force in the global robot market, with the industrial chain, technology, and application foundation to move towards the next stage of embodied intelligent robots, showing a booming development trend as a whole.

In her opinion, to become a leading enterprise in embodied intelligent robots in the end, it is not necessary to have the most cutting - edge technology. More importantly, it is to intersect or overlap with the technological trend and achieve successful commercial applications in the process. Only in this way can an enterprise have sustainable competitiveness and finally grow into an industry leader.

From this perspective, big tech players such as Baidu, Tencent, and Alibaba Ant have all shifted their focus to the underlying infrastructure and core links of embodied intelligence. They are no longer limited to competing in the size and speed of models, but are making differentiated investments and R & D around who can make the technology useful and who can run through the closed - loop.

"Creating Brains" Means Being Always Present

The essence of this "brain - creating" competition is not that big tech companies directly build robots themselves, but rather to empower the entire industrial chain with the reusable and deliverable underlying capabilities released after squeezing out the industry bubbles.

Comprehensive data from Tianyancha and IDC research point out that "Chinese manufacturers are accelerating the construction of a complete industrial chain for embodied intelligent robots. By being the first to implement applications in multiple industry scenarios, they are promoting the technology from verification to large - scale deployment. Diverse participants are jointly building an open ecosystem, which is accelerating the process of embodied intelligent robots moving out of the laboratory and into real - world production and living scenarios."

From being just able to calculate to being truly capable of doing things, it has completed a role transformation. The key point is straightforward: move the large model from the "cloud advisor" into the robot body to create a "portable brain". Perception, cognition, decision - making, and execution are integrated into a single chain, with the virtual and real worlds running in sync. It can both think and act.

Source: IDC

IDC predicts that by 2029, the global robot market will exceed $400 billion. Among them, the Chinese market will account for nearly half of the share and rank among the top in the world with a compound annual growth rate of nearly 15%, becoming the core engine driving the growth of the global robot industry.

However, what drives this leap is not only the robot industry but also the full - link collaboration of embodied intelligence, including the simultaneous advancement of algorithms, data, computing power, hardware, and application scenarios. Only when all links work together can the technology truly reach a new level. "Embodied intelligent robots are an important direction for the intelligent development of robots. Their realization depends on the integration of multiple key technologies and the integration of complex systems to meet the diverse and complex application scenario requirements," IDC research points out.

Today's embodied intelligent robots are no longer just the "tomorrow" in the laboratory. The so - called universality does not rely on just talking about semantics. It requires being able to start working as soon as it arrives at the scene.

Yuan Foyu disclosed in the speech that in terms of scene applications, Baidu Smart Cloud has accumulated experience in the implementation and replication of artificial intelligence in thousands of industries such as education, scientific research, healthcare, manufacturing, logistics, energy, commerce, and life sciences through years of practice.

Making this happen is not easy. Every aspect needs to be carefully considered, such as how to build the system, where to get the data, how to allocate computing power, how to run simulations, how to test real machines, and finally how to deliver in batches. In short, it tests the organizational ability to handle large - scale projects.

According to a report in Shanghai Securities News on July 27th, on that day, Tencent's Robotics X Laboratory and Futian Laboratory jointly released the embodied intelligence open platform Tairos, which is the first domestic software platform for embodied intelligence that provides large models, development tools, and data services in a modular way and is open to the robot industry through a plug - and - play approach.

Zhang Zhengyou, the chief scientist of Tencent, the director of Robotics X Laboratory, and the director of Futian Laboratory, said that the Tairos embodied intelligence platform is committed to complementing the key software capabilities for robot body and application developers, just like installing a "brain" for various robots.

In fact, embodied intelligence is not just a stage for big tech companies. Start - ups, solution providers, open - source communities, research institutions, and system integrators are all filling in the gaps in the industrial chain. Those who can refine the links and ensure stable delivery will have more say.

Of course, there are also thresholds. This is not the old path of light applications and rapid expansion. It is a competition of multi - terminal integration. It is necessary to shoulder the computing power and algorithms of large models and integrate the capabilities into various hardware, and adjust them repeatedly in real scenarios until they are useful, easy to use, and stable.

The Real World Is the Ultimate Test Ground

Taking a broader view, this "brain - creating" competition is not about making a single robot stand out, but about the entrance to the next - generation platform dividend. After the PC era came the mobile era, and after the mobile era came the cloud era. Now it's the turn of the operating system that embodied intelligence depends on.

It's not easy to understand human language, and it's even more difficult to understand and deal with the real world. Those who can make it to the second half may not be the ones with the loudest voices, but rather those who can align the brain, body, and scenarios, run through the full - link, and achieve long - term reuse and stable delivery.

In the specific route of "brain - creating", big tech companies have shown different strategic focuses. From Baidu's VLA adaptation to Tencent's Tairos platform, they are trying to find a connection point between technology generalization and the developer ecosystem. They are not competing with single products but are using the entire technology stack to integrate model training, verification, reasoning, and deployment, providing a capability base for the entire industry.

In this regard, cloud computing is not just a provider of computing power but also a "dispatching center" for model capabilities, determining who can quickly move from algorithm implementation to commercial transformation.

In the R & D of embodied brain models, relying on Baidu's Baige GPU cloud platform, Baidu Smart Cloud is the first cloud provider to fully adapt to the three major mainstream open - source embodied VLA models, RDT, π0, and GR00TN1.5, helping enterprise algorithm engineers quickly start exploring the VLA technology route.

Source: Baidu Smart Cloud

Tencent's Robotics X Laboratory announced on August 1st that Tairos is based on the technical capabilities accumulated by Tencent's Robotics X and Futian Laboratory in the robot field over the years and is widely combined with industrial - end requirements.

It includes large planning models, multi - modal perception models, and joint perception - action large models, and provides services externally through standardized interface calls and software development kits (SDKs), making it convenient for developers to access embodied intelligence models at a lower cost.

In addition, the platform supports one - click startup of a simulation environment on the cloud for the verification of planning and perception models, with multiple pre - set robots, scenarios, and tasks, allowing more developers to quickly verify and understand the model capabilities.

Source: Tencent Robotics X Laboratory

Compared with the strategies of Baidu and Tencent, an article in China Fund News on March 11th said that "Lingbo Technology, a subsidiary of Alibaba Group focusing on embodied intelligence, is committed to the service field, aiming to create industry - leading robot products and make robots enter people's lives and become everyone's good helper." As an embodied intelligence company under the Alibaba system, if it integrates the capabilities of Alibaba's Tongyi large model, it may have greater implementation capabilities.

The competition between platforms is essentially a competition of closed - loop speed, scenario migration ability, and developer penetration. This is not only a battle of large - model tools but also a game about who can define the next - generation operating system.

After all, the reason why big tech companies can rise to the top after the bubbles are squeezed out is not because they tell the most appealing stories, but because they have the systematic ability to deepen the technology and expand the ecosystem.

The essence of this ability is to be willing not to rush to outpace others but to take root in the industry and polish every seemingly insignificant link to make it sustainable in the long run.

Conclusion

The development of embodied intelligence no longer follows the familiar pattern. What the outside world sees is only the increase in the number of projects and the flow of funds, but it is difficult to touch the adjustments happening deep inside.

As more forces interact and correct each other behind the scenes, the original competition mode is being rewritten. The next stage may not continue along the current bustling path but will embark on a longer and narrower path, using time to filter out the forces that can leave value.

Which participants will make it to the end? Which ones will slow down at key nodes? When the noise fades, there is no clear answer, and there is no inevitable winner. But one thing is certain: those who can stay until the end will be the enterprises that make continuous investments and patient accumulations to make embodied intelligence a reality.

This article is from the WeChat official account “LingTai_LT”, written by Lin Feixue and edited by He Kun, and is published by 36Kr with authorization.