By Lao Yu'er, Edited by Yang Xuran
In June 2026, Dreame completed a far-reaching "strategic correction": it reorganized its 200 business units and ultimately locked in four core tracks: smart home, outdoor courtyard, smart mobility, and embodied intelligence.
During this process, businesses in the exploration stage were adjusted, and all funding, talent, and R&D resources were concentrated on core technologies and products related to Physical AI.
In fact, such refocusing is no longer an isolated case today. In recent years, a growing number of Chinese companies have been re-evaluating their business boundaries and unanimously choosing to "do subtraction": divesting non-core assets, terminating marginal explorations, and tightening diversified layouts, while heavily betting the freed-up capital, talent, and management attention on Artificial Intelligence (AI).
The "2025 Annual Guide to Enterprise Digital Intelligence" released by Sequoia China at the end of 2025 corroborates this trend.
Based on in-depth research of 239 enterprises, the report shows that up to 85% of the surveyed enterprises have incorporated AI into their strategic planning horizon. Among them, 27% have explicitly listed AI as a core strategy, doubling year-on-year; another 43% regard it as an important strategy.
The era of Chinese enterprises seeking incremental growth through diversified expansion is drawing to a close. A new stage has begun, with AI as the main thread, focusing on core capabilities to build technological barriers, which will quietly reshape the competitive landscape of Chinese business.
This article is an in-depth value piece from the content team of "Juchao WAVE". Welcome to follow us on multiple platforms.
Streamlining the Ranks
Diversified expansion was once the standard growth path for Chinese enterprises over the past two decades.
When the main business hit a ceiling, extending upstream and downstream, expanding into adjacent tracks, or even crossing over into entirely new fields were seen as inevitable choices for finding a second growth curve.
Given the advent of the AI wave, this logic is being subverted: more and more enterprises realize that instead of dispersing resources across multiple tracks, it is better to concentrate firepower to seize a leading position in the AI era.
Alibaba Group Holding Limited (Alibaba) is perhaps the most typical representative of the most drastic and resolute actions in this "slimming" campaign.
In December 2024, Alibaba sold its equity in Intime Department Store for a consideration of approximately RMB 7.4 billion. Just one month later, in early 2025, Alibaba sold a 78.7% stake in Sun Art Retail for a maximum price of approximately HKD 13.138 billion, completely exiting its position as the controlling shareholder of the parent company of RT-Mart.
This operation marks that Alibaba has basically bid farewell to its layout in the offline retail sector.
In May 2025, Alibaba sold an 85% equity stake in Trendyol GO, a Turkish instant delivery platform, to Uber for approximately USD 700 million, further divesting overseas non-core assets.
Asset disposal not only recycles funds, but more importantly, its strategic significance lies in the reshaping of the formation. Financial reports show that after excluding RT-Mart and Intime, Alibaba's overall revenue increased by 10% year-on-year, with a clearer business structure, basically forming a development formation of "AI + Cloud, E-commerce, and Other Internet Platform Businesses".
Tencent Holdings Limited (Tencent)'s operations, on the other hand, carry more of the meaning of structural upgrading by replacing the old with the new.
On July 6, 2026, Tencent reduced its holdings in Kuaishou by 272.9 million shares through an off-market block trade, officially exiting the ranks of major shareholders of Kuaishou. Just four days before the announcement of the share reduction, on July 2, Kuaishou's AI video generation large model "Kling" had just announced the completion of its first round of independent financing of nearly USD 3 billion, with Tencent being a key investor, contributing approximately RMB 1.363 billion.
Between this reduction and increase, Tencent has completed a structural shift from holding "traffic platforms" to holding AI productivity tools.
The choice of the gaming company Kunlun Tech is even more extreme. On August 10, 2026, Kunlun Tech issued an announcement proposing to transfer a 99% equity stake in its subsidiary Xianlai Interactive Entertainment for RMB 750 million. This company, which focuses on local card and board games, was once Kunlun Tech's most stable "cash cow", still contributing RMB 316 million in revenue and RMB 72.2 million in net profit in 2025.
Selling the cash cow against the backdrop of tightening regulation on card and board games and peaking growth, Kunlun Tech's goal is very clear: to replenish ammunition for the All-in AGI and AIGC strategy, and to comprehensively transform from a gaming company to an AI company.
Such strategic adjustments are not rare. According to the Shanghai Securities News, from October to mid-November 2025 alone, nearly 250 companies issued announcements related to asset sales, doubling year-on-year. As of April 27 this year, 91 listed companies have also completed asset sales.
This change is not hard to understand. The essential purpose of diversification is to find new increments. As the prospects of AI become increasingly clear, going all-in on AI has become the biggest and most certain main thread for numerous enterprises.
Addition
Strategic adjustment is "doing subtraction outward", while technological investment is "doing addition inward".
Dreame significantly adjusted its business landscape this year and chose to bet on Physical AI represented by robots, robot vacuums, and courtyard robots. Its long-term technological layout in the AI field is the core reason for this decision:
In 2021, the robot vacuum business line officially established an AI algorithm team, beginning to systematically explore how to make cleaning equipment move from "executing commands" to "autonomous perception and decision-making";
In 2023, Dreame took the lead in introducing reinforcement learning into the R&D and application of robot vacuums, achieving significant technological advantages. Currently, new products equipped with the RLPP all-scenario reinforcement learning engine have achieved ultra-fast obstacle avoidance within 0.2 seconds, improving navigation and obstacle avoidance efficiency by 50%, and can perceive in real-time and make instant decisions in dynamic scenarios such as pets running;
In March 2026, the industry's first active binocular vision solution was implemented. By simulating the principle of human eye parallax through dual RGB AI cameras and combining 3D depth perception, it achieves more three-dimensional environmental perception.
On this basis, Dreame further introduced cloud large model capabilities to the product end. Through the deep integration of the DeepSeek large model and the self-developed DreameGPT, the robots can not only understand vague natural language instructions such as "the milk is spilled" and "the living room is a bit dirty", but also combine the whole-house map and user usage habits to upgrade from "understanding instructions" to "proactive service".
Coupled with its core competency in vertical manufacturing, Dreame has actually formed a relatively complete industrial chain from upstream core components to downstream terminal products, and from hardware to AI capabilities.
Tencent's R&D investment reached RMB 85.75 billion in 2025, and capital expenditures reached RMB 79.2 billion, both hitting record highs. Among them, the investment in new AI products such as the Hunyuan large model alone was approximately RMB 18 billion. The company's management also explicitly stated that this investment is expected to "more than double" in 2026.
Alibaba is equally aggressive. CEO Eddie Wu announced in February 2025 that over RMB 380 billion will be invested in AI and cloud computing infrastructure construction over the next three years, an amount exceeding the total of the past decade.
Recently, Alibaba announced a rights issue financing of RMB 80 billion to further invest these capitals in the AI infrastructure field. At the same time, leaders such as Joe Tsai are increasing their holdings in Alibaba's stock. This is almost an all-or-nothing full-scale investment, which is extremely rare in the development history of Alibaba and even Chinese technology enterprises.
Promising Prospects
Technological investment must ultimately answer the question of commercial value.
Developed to this day, the AI competition has moved from the "technology show" stage of competing in parameter scale to the deep-water zone of commercialization and implementation. The AI commercialization of the giants is actually accelerating its realization.
As of the end of last year, Alibaba's AI-related product revenue achieved triple-digit year-on-year growth for the tenth consecutive quarter, with a growth rate far exceeding the overall business. During the earnings call for the third quarter of fiscal year 2026, CEO Eddie Wu explicitly proposed commercialization goals: in the next five years, the annual revenue of cloud and AI businesses must exceed USD 100 billion.
Tencent's AI is also rapidly integrating into the core business fabric. Based on the Hunyuan large model, Tencent's AI product line is rapidly expanding, with monthly active users of C-end AI products breaking 100 million one after another. On the B-end, driven by the explosive demand for AI, Tencent Cloud achieved its first full-year scaled profitability in 2025, and the enterprise service revenue growth in a single quarter reached 22%.
Compared to the platformization path of Internet giants, Dreame, as a hardware company, has a more direct commercialization path: deeply injecting AI capabilities into every hardware product, making consumers pay for "smarter and more user-friendly" products.
The core main business of robot vacuums remains the most mature category for the implementation of Dreame's AI capabilities.
By recognizing over 280 types of household objects through AI vision, combining reinforcement learning to achieve ultra-fast obstacle avoidance, and adding bionic mechanical arms to solve edge-cleaning blind spots, the upgrade in product experience has been directly translated into an increase in market share.
According to the "Global Home Smart Robot Vacuum Market Quarterly Tracking Report" released by IDC, in the first quarter of 2026, Dreame robot vacuums ranked first globally with sales of 1.555 million units and sales revenue of USD 924 million, and ranked first in market share in 30 countries worldwide. Among them, the market share in countries such as Belgium and New Zealand exceeded 60%, and the market share in 10 countries exceeded 50%.
The floor washer track has also staged a story of technology-driven growth. As of May 2026, the global cumulative shipments of Dreame floor washers exceeded 10 million units, taking only about 5 years from entering the market to reaching a scale of tens of millions. It ranks first in market share in 17 countries and regions, with the market share exceeding 60% in some countries.
The outdoor courtyard track, which is rarely involved by other enterprises, is also a core business that Dreame actively retains.
According to current public information, Dreame took the lead in introducing industrial-grade LiDAR into consumer-level mowing scenarios, integrating high-precision 3D LiDAR with a binocular AI vision solution. This enables its products to accurately distinguish the boundaries between lawns and non-lawns, and recognize over 300 common courtyard obstacles—automatically cutting close to the edge when seeing flower beds, and automatically detouring and planning paths when seeing pets.
AI has transformed the lawn mower from a semi-automatic device that "follows the boundary wire" into a more autonomously operating courtyard robot.
According to an independent research report by Frost & Sullivan, as of May 2026, the global shipments of MOVA brand lawn mowing robots exceeded 500,000 units, a year-on-year growth of over 400%, securing a leading position in the global market for this segmented market of courtyard robots.
It can be seen that from indoors to outdoors, from cleaning to washing and care, AI will redefine every traditional home appliance category. From January to May 2026, Dreame's main business increased by 80% year-on-year, continuing the high-speed growth trend for 9 consecutive years. Against the general backdrop of weak overall demand in consumer electronics, such a growth rate means that on the large board of Physical AI, various products still have broad room for growth.
Final Thoughts
Looking back at this round of strategic shift of Chinese enterprises, its essence is a gear-shift development after the traditional industries have completely peaked. A large number of enterprises have realized that the AI industrial chain is the most core incremental market space.
The enterprises mentioned above have already been able to enter the core field of AI through various means. However, more enterprises actually have few choices when transforming. The technology-intensive and capital-intensive attributes of AI determine that only a few enterprises can ultimately succeed in this process.
For example, according to statistics from Nanfang Wealth Network at the end of last year, there are already about 300 listed companies in the A-share "computing power concept", including 97 on the main board, and around 200 combined on the SME board, ChiNext, and STAR Market. These companies are basically traditional industry companies that have transformed and crossed over, and the low probability of their successful transformation is conceivable.
But even so, enterprises have few choices. In the next three to five years, AI transformation will still be a mandatory option for a large number of enterprises, changing from an icing-on-the-cake bonus to a vital basic capability.