Top Ten Critical Moments in Chinese Business in 2025: Reconstruction, Awakening, and Competition-Collaboration
In 2025, the hustle and bustle is a collective search for certainty. Amid the dissolution of old rules and the gestation of a new order, enterprises are seeking a sustainable balance between technological ideals and business realities, scale expansion and in - depth value exploration, as well as the speed of innovation and the safety baseline.
As the technological singularity draws nearer, consumer cognition iterates gradually, and the cost of traffic continues to rise, the basic rules of business competition are being systematically rewritten.
Looking back at 2025, it is not just an ordinary year but a concentrated turning - point after long - term accumulation. In the past few years, technological reserves, consumer habits, and policy frameworks have been silently gathering strength. This year, the potential energy formed by the three has begun to be fully released, driving one industrial structural adjustment after another.
In this process, enterprises have to engage in a high - intensity "arms race" in fields such as AI, robotics, and intelligent driving to keep up with technological advancements. On the other hand, they must also explore sustainable business models more pragmatically and meticulously to find certain growth amid uncertainties.
The emergence of each key variable, whether it is DeepSeek reshaping the AI competition logic through open - source, humanoid robots moving from demonstrations to practical use, or the trust discussions triggered by pre - prepared meals, is not just an isolated event but a manifestation of the evolution of the underlying logic of the industry.
By sorting out the top ten key changes of the year, we attempt to restore the evolution context of the business world in this year. Why did the changes occur? Who is driving the rule reconstruction? And beneath the surface hustle and bustle, which trends have quietly settled, pointing to a visible future?
01
The Arrival of the DeepSeek Moment: The Shift of the "Hundred - Model War"
Dark Horse
Perhaps no one expected that during the Spring Festival in 2025, the focus of public discussion shifted from the traditional red envelopes to the AI large - scale models, which were originally high - threshold technologies. The core cause of this shift was that DeepSeek broke the long - standing deadlock in the domestic AI large - scale model track with its open - source model.
Before that, the so - called "hundred - model war" had fallen into an obvious scale involution. Manufacturers were competing to invest huge resources in comparing the number of parameters and data scale, but no substantial breakthroughs in user experience or business had been made. High R & D investment and unclear profit prospects made the entire industry gradually return to rationality from the early enthusiasm. The "appearance" of DeepSeek was like a reset of the existing game rules.
The industry's reaction was rapid and real. Alibaba organized a team to cancel the Spring Festival holiday and launched an emergency attack, releasing the flagship version of Tongyi Qianwen, Qwen2.5 - Max, and claiming that this model comprehensively surpassed top - tier models such as GPT - 4o, DeepSeek - V3, and Llama - 3.1 in multiple tests. The market responded strongly, and Alibaba's stock price soared.
DeepSeek became so popular that even the elderly dancing square - dance could talk about it. Tech giants like Tencent, Baidu, Alibaba, and ByteDance, which have self - developed large - scale models, surprisingly reached a consensus and chose to connect to DeepSeek, forming a rare pattern of intertwined competition and cooperation.
The arrival of the DeepSeek moment not only activated the innovation imagination of AI applications but also touched two fundamental aspects of the industry.
Firstly, it is a reflection on the industry's cost structure and business model.
The open - source path fundamentally shakes the traditional model of building moats with closed - source technologies and profiting through API calls. It forces the entire industry to re - evaluate one thing: when the core model can be obtained at a low cost, where should the real value be created? Is it the optimization ability for specific scenarios, the in - depth understanding of segmented needs, or the applications and services built on the model? This drives the industry to shift from an arms race to a value reconstruction.
Secondly, it is the migration of the traffic distribution logic.
Large - scale models are gradually evolving from a tool to a new entry point. When users are used to "asking the large - scale model when in doubt", the central position of traditional search engines begins to loosen. This is not only a challenge for Baidu but also means that the underlying paradigm of Internet traffic distribution is changing, from active keyword searches to interactive intelligent Q&A. Traffic will flow with the ability of dialogue and services, rather than simply relying on information indexing.
DeepSeek's breakthrough is not so much a victory for a single company as a validation of a development model. It marks that the domestic AI large - scale model industry is starting to bid farewell to the homogeneous resource - consuming war and enter a new stage based on open - source and centered on ecological construction and real - world applications.
02
Humanoid Robots: From Show - off to Profit - Making
Actor
During the Spring Festival Gala in 2025, humanoid robots became the focus under the spotlight. Sixteen H1 robots from Unitree Technology danced yangge and twirled handkerchiefs on stage with human dancers. Their precise posture control and smooth coordination made them an "instant tech sensation" overnight, and Unitree Technology also became the focus pursued by the investment circle.
However, the glorious moment on the stage does not directly illuminate the path to commercialization. Stepping off the dazzling stage, humanoid robots face a more complex and cruel business world, that is, what exactly can they be used for?
Throughout 2025, we saw humanoid robots appear intensively in various scenarios. They participated in running races, competing with humans in speed and endurance. The Zhiyuan Expedition A2 walked 106.286 kilometers and won the Guinness certification. The XPeng IRON robot walked the catwalk and was even suspected of being a human in disguise due to its extremely human - like posture. Six Unitree Technology G1 humanoid robots performed as backup dancers at singer Wang Leehom's concert, effortlessly performing actions such as waving arms, kicking legs, turning around, and even doing flips.
This series of high - profile exposures can be better understood as a collective release of anxiety in the industry. After investing a huge amount of R & D resources, the industry urgently needs to prove to the market and investors that this technology is not only cool but also useful.
Technological feasibility does not equal commercial feasibility. This is the most realistic portrayal of the humanoid robot industry at present, which has always been struggling between technological excitement and business rationality.
After years of development, the industry has made quantifiable progress in core technologies such as motion control and environmental perception and can already complete a series of complex actions. However, there is a huge gap between "being able to do" and "being worth large - scale purchase and use".
The transition from the stage to the real world is destined to be difficult. Humanoid robots have proven their technological potential, and the next step is to prove their commercial value.
This is no longer just a laboratory project that only needs to overcome technological difficulties but a systematic project that requires in - depth understanding of the industrial chain, cost constraints, and real - world needs.
03
The "Three - Kingdoms Battle" in the Food Delivery War: A Long - Awaited Change in the Landscape
Drama
JD.com, which was deeply trapped in the bottleneck of retail growth, entered the food delivery market in February under the banner of high - quality food delivery. Subsequently, Meituan launched a defensive counter - attack, and Alibaba upgraded Ele.me to Taobao Flash Delivery and fully participated in the war, triggering an unprecedented food delivery battle.
The most direct benefit of this battle was enjoyed by consumers. The platforms launched intensive 10 - billion subsidy programs, making "a cup of milk tea for a few cents" and even "free purchases" a national consumption phenomenon.
However, this round of competition had a far - reaching impact because it quickly went beyond the simple dimension of price war and pointed directly at the deep - seated structural problems that had long existed in the industry.
The focus of competition shifted from C - end traffic to the competition for B - end (merchants) and P - end (riders) ecosystems.
JD.com's entry strategy was targeted. On the one hand, it attracted small and medium - sized merchants with the policy of "commission - free for the whole year". On the other hand, it addressed the pain point of rider social security in the industry by "covering the five social insurances and one housing fund for full - time riders". The essence of this combination of strategies was to use better ecological terms as a tool for supply - side reform, attempting to shake the operational foundation of its opponents.
This forced Meituan and Ele.me to follow up and adjust their relevant strategies, which generally improved the operating pressure on the merchant side and the discussion space for rider rights and interests protection.
According to a report by J.P. Morgan, Meituan, relying on its extensive merchant coverage and delivery network, maintained a 50% market share; Alibaba, by integrating Ele.me and Taobao traffic, occupied 42%; and as a new entrant, JD.com obtained 8% of the market share.
The cost of this competition was huge. In the third quarter of 2025 alone, the incremental marketing expenses related to food delivery of Meituan, Alibaba, and JD.com reached 61.4 billion yuan, directly impacting their profits.
This unsustainable consumption war eventually attracted regulatory intervention. After several rounds of interviews, the platforms promised to stop malicious competition, and the market gradually returned to rationality.
The real legacy of this battle is not the change in market share but the renewal of industry consensus. The competition in the food delivery market has entered a new stage. The era of simply relying on subsidies to obtain traffic is over. In the future, the competition will be a comprehensive value - chain ecological competition covering merchant empowerment, rider system protection, user experience improvement, and compliant operation.
04
Instant Retail: A Lucrative and Certain Market
Incremental
The real strategic value of the food delivery war lies in paving the way for the broader instant retail market. This war verified the efficiency of the urban instant fulfillment network and the user mindset, and naturally extended the battle to a more certain trillion - level market, an upgrade from "delivering food" to "delivering everything".
If the previous food delivery war was a crude competition of "burning money for traffic", the competition in the instant retail track has entered a high - level stage where each player shows its unique skills. The three giants, Alibaba, Meituan, and JD.com, have put forward differentiated strategic chips based on their core genes.
Alibaba's trump card is its comprehensive ecosystem. It quickly completed strategic integration, merging Ele.me and Fliggy into the Alibaba E - commerce Business Group, and deeply binding Taobao's overall traffic with Ele.me's instant fulfillment ability. It horizontally expanded to all - category scenarios such as supermarkets, fresh produce, beauty products, and digital products, with the advantage of wide - range category coverage and a rich merchant ecosystem.
Meituan began to shrink non - core businesses and focus on the main line. It included Flash Delivery as a first - level entry on the home page and consolidated its basic market with a high - density fulfillment network, expanding in - depth. Its core competitiveness is the high - density fulfillment network that covers the whole country and penetrates into every corner.
In fact, Meituan's strategy is more like the "principle of pressure concentration". It deepens and refines its operations in the local - life field where it has an advantage, efficiently replicating the user habits, rider capacity, and merchant relationships accumulated in food delivery to the retail category, pursuing vertical penetration and operational efficiency.
JD.com, relying on its long - established warehousing and logistics system and supply - chain management ability for standard products such as 3C and large household appliances, positions itself with the differentiation of high - quality instant and precise fulfillment, serving orders from customers who have higher requirements for product reliability and delivery professionalism, thus building a barrier in the segmented market.
Instant retail is a rare incremental market in the second half of the e - commerce era. The core of future competition is ecological competition based on efficiency. The key to victory no longer lies in who offers more subsidies but in who can provide more effective digital empowerment for a wider range of offline merchants; who can operate a more stable fulfillment network with a better cost structure; and who can have a more intelligent technology platform to predict demand, match inventory, and plan routes.
Behind this is a comprehensive and arduous test of the internal and external collaboration ability of the organization, the in - depth understanding of offline business, and the efficiency of technological empowerment.
05
Putting on AI Glasses: Not a New World Yet
Proliferation
2025 is known as the first year of AI glasses. Some say there are about 30 - 40 entrants, while some institutions have counted more than a hundred.
In addition to traditional AR hardware entrepreneurs, search platforms, social platforms, e - commerce platforms, mobile phone manufacturers, computer manufacturers, telecom operators, and car manufacturers... all have entered the field of AI glasses production.
Behind the diverse market, there are only a few products that can make it to the mainstream. Products such as Thunderbird Innovation X3 Pro, Xiaomi AI Glasses, China Telecom Tianyi AI Smart Glasses, Rokid Glasses, Xiaodu AI Glasses Pro, Quark AI Glasses S1, and Li Auto AI Glasses Livis have formed a preliminary market segmentation.
Currently, mainstream products can be roughly divided into three categories: AI audio type (focusing on voice interaction and auditory enhancement), virtual display type (providing floating screen information prompts), and AR immersive type (achieving deep integration of virtual and real worlds).
This differentiation is not accidental. It reflects the pragmatic exploration of manufacturers between technological boundaries and market acceptance. Instead of pursuing an immature all - around experience, it is better to make specific functions usable and user - friendly.
Different from previous VR devices, this generation of AI glasses generally takes wearing comfort, daily - use design, and clear scenario value as core selling points, trying to shed the label of a geek toy.
Price reduction is a key step in accelerating market education. Manufacturers such as Xiaomi and Li Auto have set the starting price of their products below 2,000 yuan. After the national subsidy, the price is close to that of ordinary high - end glasses, which undoubtedly lowers the threshold for consumers to try new products and aims to quickly expand the user base.
However, the fundamental challenge in the industry remains clear. AI glasses have not yet found a rigid scenario that makes them indispensable to ordinary users. Currently, they are more like auxiliary peripherals of mobile phones rather than substitutes. AI glasses have not yet reached a perfect balance in terms of weight, battery life, and truly practical AI capabilities that consumers can accept.
For the entire AI hardware industry, AI glasses represent a path full of hope but requiring patience. It is no longer just about showing off technology but answering the most practical question: in users' real lives, what pain points can I solve that other devices cannot solve better?
06
Pre - prepared Meals in the "War of Words": Old Problems, New Controversies
The War of Words
A public debate initiated by Jia Guolong, the founder of Xibei, and Luo Yonghao pushed pre - prepared meals to the center of public opinion. This seemingly sudden "verbal fight" actually revealed a long - standing deep - seated contradiction in the industry. Today, when pre - prepared meals have highly penetrated the catering system, there is still a significant information gap and cognitive misalignment between consumers and the industry.
Consumers' attitudes have shifted from early resistance to conditional acceptance, but the premise has become clear: the price must truly reflect the value. When consumers pay a premium for "freshly - cooked" food but may actually taste the standardized products from central kitchens, the perceived unfairness can turn into a trust crisis.
This verbal fight triggered a struggle for the definition right of pre - prepared meals, industry transparency, and consumer right to know. Some netizens reported that many restaurants, such as Green Tea Restaurant and Xijiade Dumplings, removed the "freshly - made" label. Jia Guolong changed from being tough to admitting his mistake, and consumers temporarily gained the upper hand.
In the long run, the development path of pre - prepared meals should not be to "replace freshly - cooked food" but to become a transparent, compliant, and valuable