After raising RMB 700 million in Series A financing, where does Habitat want to bring the concept of "home" to?

On August 20, Habitat, a smart residential brand, announced the completion of 700 million yuan Series A financing, which was led by the Science and Innovation Fund of Xinbei District, Changzhou. This is also the largest financing deal in the whole home industry on China's primary market in the past three years. Prior to that, Habitat completed its first round of financing of 653 million yuan in 2024.

Almost at the same time, the 2026 World Humanoid Robot Games provided an observation perspective: the organizing committee added a home scenario competition, which required a sufficiently realistic home environment that could provide unified conditions for participating teams from different cities. After several twists and turns, the Beijing Experience Center of Habitat was finally selected as the exclusive venue provider for the home scenario.

The venue fully adopts the product standards that Habitat delivers daily in many cities across the country. Shen Yanan, founder of Habitat, described it in this way, "Buy the venue, move in and you can live directly". This brings up a more noteworthy question — When AI starts to move from screens to the physical world, will the "home" we are familiar with also change accordingly?

When AI Enters Home, the Home Needs to Evolve First

The problem reflected by the robot games venue lies not only in the capabilities of the robots themselves, but also in the home environment that carries them.

Home is a highly non-standard working environment with different apartment layouts, devices and interaction modes, and the algorithm iteration of embodied intelligence will not automatically eliminate these differences.

Take turning on the light as an example. In a traditional residence, the robot needs to identify the position and function of the switch before completing the mechanical operation. When moving to another home, the shape, position and control relationship of the switches may all change. Especially when facing more complex devices such as knobs and touch panels, the adaptation cost will only rise further.

While smart homes provide another path. Habitat has connected fixed devices such as temperature regulation, fresh air system and lighting into a unified system, and built in the spatial agent "Xiaoqi". When the robot needs to turn on the light, it can directly interact with the residential system, and "Xiaoqi" will call the lighting system.

Home spatial agent Xiaoqi (the picture shows one of its embodied forms)

Shen Yanan calls this relationship "two-way mutual advancement" — The more fully the residential infrastructure is prepared, the less repeated adaptation work the robot needs for different home environments, and the more opportunities it will get to take on truly valuable work earlier.

Robot is just one aspect. No matter it is embodied devices, or spatial agents that run 24 hours a day, the premise for AI to work really in the home is that the residence first becomes a whole that can be perceived, called and continuously upgraded uniformly.

This is of more universal significance than renovating the residence specifically for a certain type of robot, and it brings the real problem back to the residence itself. Shen Yanan is used to using the concepts of "DIY assembled PC (buying parts to assemble a computer)" and "laptop" to intuitively show the difference between the traditional home furnishing mode and Habitat's mode.

Under the traditional home furnishing mode, consumers need to find design services, materials, woodwork, equipment and construction teams respectively, and then combine these links to form a home. Devices of different brands operate independently, the result is highly dependent on personal experience, and there is no unified interface. Such a space is difficult to be called uniformly, let alone continuously upgraded.

Habitat's solution is to first make the residence a finished product.

However, turning a service that has long relied on the experience of designers, foremen and construction workers into a product that can be continuously researched, manufactured, delivered and iterated is itself a systematic project. It requires enterprises to re-complete product definition, and establish a research and development, supply chain and delivery system around this set of definitions.

The first problem to be solved is how a residence is defined in the first place. Habitat breaks down its products into three levels: smart home chassis, living functions and appearance suites: capabilities such as air, water, light and intelligence are deposited into the chassis, living functions are combined in a modular way, and colors, materials and other options are provided through different suites.

The core of its product definition is to reorganize the capabilities that were previously scattered in design, materials, equipment and construction into a set of product architectures that can be repeatedly researched and iterated. It is worth mentioning that based on this chassis, the residence begins to have unified perception and calling capabilities. "Xiaoqi" can read the environment, call devices, and actively adjust environmental parameters such as light, temperature, and air quality.

Secondly, after the product definition is completed, this set of standards needs to be fully implemented in the supply chain and construction sites. Before the construction of each Habitat residence, a digital twin will be formed first. The design effect, cost and materials will be confirmed in the digital space first, and the same set of data will run through subsequent links such as supply chain, woodwork manufacturing, construction drawing drawing and work order generation. The residence is fully defined in the digital world first, and then reproduced on the physical site.

This forms Habitat's starting point in the AI era: first turn the residential delivery work that highly relies on manual experience into a definable, researchable, replicable and continuously iterable product, and then let AI further become the organization center of the space.

The Evolution Path of Home

"Build homes in the way of manufacturing cars" is the most easily understood expression of the outside world for Habitat, which explains what Habitat has done in the past three years, but it is only the first step.

Build homes in the way of manufacturing cars

Through standardized and product-oriented processes, Habitat's previous generation of products endow the residence with a unified software and hardware base, and access spatial agents such as "Xiaoqi".

Shen Yanan defines this as the L1 stage of smart home evolution. In his vision, smart homes will continue to evolve along the L1-L4 path. Each upgrade means that the residence gains more capabilities of understanding, judgment and action. Up to now, L1 has been realized: AI has entered the residence, but the underlying architecture still follows the logic of adding intelligent agents after the residence is completed.

When it comes to the L2 stage, AI will change from an additional capability after the product is completed to the organization center of the space, which requires re-designing the residence starting from AI, and the software and hardware architecture, communication mode and intelligent hub will be adjusted accordingly.

Then in the L3 stage, AI will further gain the ability to act in the physical world. Embodied intelligence will be connected to the residence, so that the spatial agent will have "hands and feet" that can move and operate. According to Shen Yanan, the company has begun to consider introducing embodied intelligent products into its product system, and related products may be included in the existing delivery system as early as next year.

In Shen Yanan's view, middle and high-income families will generally have embodied intelligent devices in five years, "the work they undertake will not be limited to cleaning, and they will be more like a physical agent, a family housekeeper that can intervene in the physical world."

L4 is a longer-term vision. Shen Yanan calls it a "spatial intelligent terminal". The home intelligent agent will no longer be limited to a fixed residence. It can continue to accompany users when they go out, on business trips or stay in hotels, and preserve the relationships, habits and memories formed by the whole family for a long time. He calls this form "silicon-based family members".

Shen Yanan uses Tesla as an analogy for this evolution path: L1 corresponds to the integration of software and hardware, L2 corresponds to the AI-native architecture, and L3 corresponds to embodied intelligence. From his perspective, FSD for Tesla is similar to the spatial agent for Habitat: the intelligent agent running in the space is the core that can continue to add value.

A detail that is easy to ignore is that Habitat has built a home storage and computing center in each set of residences since its first generation of products. Shen Yanan explained, "You can hardly imagine that the soul of your most important silicon-based family member is placed on a public server. The optimal and only optional place to store it is your own home."

The boundary Habitat has set for itself has gone beyond residential delivery.

A complete smart home is the most important product form at the present stage, but productization solves the problem of whether the residence can be defined and called uniformly, and AI-native solves the problem of who organizes the space.

Moving forward from L1 to L4, the center of value will gradually shift to the space itself as a terminal. What Habitat is betting on is the redefinition of living space as an intelligent terminal in the AI era.

What is the capital betting on after the 700 million yuan financing?

To make judgments, capital needs not only future narratives, but also facts that have already happened. The 700 million yuan Series A financing, the number itself represents a judgment from the capital side.

Looking back on the evolution path of smart homes, the present is exactly the node where L1 moves to L2. What the capital is betting on is the technical dividend and market space that have not been fully released between the two stages.

On the one hand, in the view of capital, L1-level smart homes have obtained clear market verification. According to the disclosure of Habitat, its current annual contracted scale exceeds 1 billion yuan, and its revenue and order volume have doubled for three consecutive years. Since the first generation of new products was released on February 28, 2025, the number of users has maintained a quarterly growth rate of about 50%, and its direct sales network has covered 7 cities including Beijing, Shanghai, Hangzhou, Shenzhen, Guangzhou, Chengdu and Wuhan.

The access threshold for AI models and single-point technologies is decreasing, but the underlying architecture of residences, product definition, supply chain organization, digital twin and cross-city delivery capabilities require long-term and continuous accumulation. Turning complex technologies into stable products and delivering them in different residences one by one is itself a commercial barrier.

These results show that the market has begun to accept the way of purchasing complete smart home products; at the same time, Habitat has been able to deliver L1-level smart home products stably. According to the data of Habitat, the delivery NPS has increased from 27.8 in the early stage of entrepreneurship to 51.1, which is at the leading level of the industry and is still steadily improving.

On the other hand, this round of financing also represents the capital's recognition of the evolution path of smart homes. According to 36Kr's understanding, Habitat will hold an autumn technology conference at the Modern Shanghai Design Week on September 7, officially releasing AI-native smart homes and underlying architectures, marking that the smart home industry has officially entered the L2 era.

The concept of AI-native has been popular for a long time. Manufacturers of mobile phones, PCs and even automobiles have all put forward this concept and taken a lot of actions: mobile phones add end-side computing power and personalized operating systems; PCs reorganize experiences around chips, systems and local intelligent agents; automobiles take technologies such as intelligent driving and intelligent cockpit as the measurement standards of technical reserves.

But the core of these different industries is essentially the same — reorganize computing power, data, operating systems, sensors and control interfaces around AI. What mobile phones, PCs and automobiles are experiencing is the process that AI moves from the application layer into the system architecture and becomes the organization center.

Residences go further than other terminals. It not only carries the data and relationships formed by the family for a long time, but also connects air, light, temperature, energy, home appliances and embodied devices. For AI to obtain complete capabilities of perception, judgment and action, the residence will naturally become a part of the extension of AI-native terminals to the physical world.

In this regard, Habitat itself is also laying infrastructure for the next stage. According to the financing announcement, the funds from this round will be mainly used for the previously announced global headquarters and R&D center of smart homes, as well as the embodied intelligence laboratory and production-level AI design and delivery system.

The 700 million yuan Series A financing can be regarded as the comprehensive pricing of the capital for Habitat's current ecological position in the industry. Habitat, which takes the lead in moving from L1 to L2, has initially mastered the right to define the industry, and has obtained the future ticket for the right to speak in the relationship between human beings and space.

The relationship between people and home has long been one-way.

People are responsible for designing, arranging, using and maintaining the space, and the home passively carries the life that happens in it. In the process where AI and embodied intelligence gradually go deep into the family, this relationship begins to become two-way. The unified device system, local computing power, data storage and software system enable the home to continuously learn and iterate, and grow continuously according to the changes of family members and lifestyles.

It is conceivable that when spatial agents like "Xiaoqi" really become "silicon-based family members", the residential space can take on more judgment, coordination and daily affairs, and people can return to life itself, and devote time to companionship, rest, interests and creation again.

At that time, the living space will become the most important intelligent terminal for human life, and will continuously understand and adapt to people in the long-term co-living process. This may be the next-generation living mode that Habitat is trying to reach.