Tech Explained: 2026 Outlook: Capital Speeds Up AI Application Implementation, Tech Giants Cease “Skill in Simple Terms

In 2025, the artificial intelligence industry accelerated its evolution amidst the intertwining of technological breakthroughs, application implementation, and capital waves. This year, with the emergence of phenomenon – level applications like DeepSeek and the emergence of humanoid robots, technology has integrated into reality in diverse forms. The core logic of industrial competition has also undergone fundamental changes: from the competition of parameter scale, it has comprehensively shifted to the competition of the in – depth implementation capabilities of enterprises such as Alibaba, Ant Group, ByteDance, Tencent, and Baidu in life and production scenarios.

Capital is the accelerator and touchstone of this process. From the concentrated listing of domestic GPU enterprises to the sprint of Zhipu and MiniMax for the “first share of large models”, the industry has galloped forward with the help of capital. However, beneath the feast, the harsh realities of “high hallucination, high power consumption, high cost” and “low user retention”, as well as the incomplete commercial closed – loop, are continuously questioning the sustainability of each track.

Facing the challenges, “collaboration and co – creation” has changed from an option to a necessity. When the advantage of a single technology is no longer sufficient, achieving system synergy through open – source and ecological co – construction has become the common choice of leading enterprises. This structural transformation will deepen in 2026 and become the key driving force for promoting the large – scale and industrial implementation of the AI industry.

Large models shift from competing on parameters to competing on applications

From completing payments with a single voice command, to getting professional advice from an AI health butler, to robots autonomously collaborating to complete complex industrial operations… These increasingly popular scenarios clearly outline the industrial landscape of the “highlight explosion” and “embodied implementation” of artificial intelligence in 2025. If 2023 is regarded as the “enlightenment year” of generative AI and 2024 as the “application exploration period”, then 2025 is undoubtedly the key leap for AI technology to comprehensively move from the laboratory to the industry and achieve penetration from single – scenario to full – scenario.

This year, artificial intelligence achieved fundamental qualitative changes in multiple dimensions. The large language model evolved from a mechanical word – patcher to a “thinker” with logical reasoning ability; the video generation model is no longer satisfied with “looking like”, and has begun to build a “world simulator” that understands physical laws; more breakthroughly, AI has truly “grown legs” and entered factories and families in the form of humanoid robots, achieving a leap from the digital world to the physical world.

Along with technological breakthroughs, the focus of industrial competition has shifted from parameter scale to the breadth and depth of scenario implementation. ByteDance’s “Doubao” has penetrated into high – frequency scenarios such as short – video creation and intelligent customer service, and achieved cross – application task collaboration through system – level cooperation with mobile phone manufacturers; Tencent’s “Yuanbao” has become an all – weather “personal companion” and deeply integrated into the ecosystems of Video Accounts and Enterprise WeChat; Alibaba Group has successively launched several AI – native applications such as Tongyi Qianwen APP, Ant Lingguang, and Ant Afu. Among them, “Afu” uses the “doctor AI avatar” technology to make health services more inclusive; “Lingguang” allows users to quickly generate interactive lightweight applications through natural language, and the number of “flash applications” created by users has exceeded 12 million within just one month of its launch.

The continuous expansion of application scenarios has further catalyzed innovation in niche tracks such as intelligent hardware. With the deep integration of large models and spatial computing, the smart glasses field has rapidly evolved from a niche market dominated by start – ups to a “battle of a hundred glasses” where technology giants gather. The entry of enterprises such as Xiaomi, Lenovo, Baidu, and Alibaba is accelerating the industrialization process in this field.

While the application layer is iterating rapidly, the competition in underlying model capabilities remains fierce. Alibaba’s Tongyi Qianwen has topped the global open – source model list, with a cumulative download volume of over 600 million; the daily call volume of the Doubao large model has exceeded 50 trillion Tokens, and its multimodal intelligent agent capabilities have been applied in scenarios such as education and industrial quality inspection; Tencent’s Hunyuan large model has released over 30 new models throughout the year; Baidu’s Wenxin large model has also been upgraded to version 5.0.

Qi Yuan, the dean of the Shanghai Institute for Science and Intelligence, said that looking back at 2025, the competition of enterprises in the field of artificial intelligence has gone beyond the comparison of computing power and parameters and entered the stage of in – depth competition in value – creation capabilities. Especially in high – demand fields such as finance and healthcare, technology must achieve three leaps from “usable” to “dare to use” and then to “easy to use”. This requires enterprises not only to have strong technological capabilities but also to deeply understand the industry context, rules, and core pain points to truly form a differentiated competitive advantage.

Liu Xingliang, the dean of the DCCI Internet Research Institute, pointed out that the core theme of the evolution of AI applications in 2025 is the systematic upgrade of Agents. Agents have changed from early auxiliary tools that execute single instructions to organic systems capable of autonomous planning and collaborative execution of complex tasks, thus deeply reconstructing the workflow and decision – making mechanism of enterprises. In his view, in the next three to six years, expert – type agents targeting vertical scenarios and possessing in – depth industry knowledge will enter the stage of large – scale implementation and become an important factor in promoting the leap of industrial efficiency and decision – making quality.

Capital helps AI develop at an accelerated pace

In addition to the in – depth iteration of technology and applications, the collective rush of AI enterprises to the capital market in 2025 has become a prominent trend throughout the year. With the successive listings of companies such as Moore Threads and MXIC Semiconductor, technology companies represented by large models and embodied intelligence have set off a wave of listings and financing.

According to incomplete statistics, as of the end of 2025, about 215 new companies were listed. Among them, the number of companies with AI business has rapidly increased from 21 last year to 51, an increase of 143%.

Looking at the specific process, in July, Unitree Technology started the IPO guidance on the STAR Market and completed it in November; in August, Coohom submitted a listing application to the Hong Kong Stock Exchange, with J.P. Morgan and CCB International as joint sponsors; in December, two large – model companies, Zhipu and MiniMax, passed the hearing of the Hong Kong Stock Exchange and are competing for the “first share of large models”; in the same month, the official website of the China Securities Regulatory Commission showed that Deep Robotics started the listing guidance on the STAR Market, with CITIC Construction Securities as the guidance institution.

These enterprises generally have the common characteristics of large R & D investment, long cycle, and significant early – stage losses, and are highly dependent on capital market support to build technological barriers. Financial data also confirms this: Zhipu’s net loss from 2022 to 2024 has expanded from 144 million yuan to 2.958 billion yuan, and the loss is expected to continue to expand; MiniMax’s loss as of the third quarter of this year is about 3.61 billion yuan, and the cumulative loss in the past three years has exceeded 800 million US dollars. Both companies stated in their documents that the losses are mainly due to the continuous investment in large – model R & D and computing power infrastructure, and the possibility of profit and dividend in the short term is low.

In this context, the screening logic of capital has become clearer: it is necessary to evaluate both the hardcore degree and core barriers of technology and the continuous operation ability of enterprises. This means that the AI enterprises that can ultimately break through in the competition rely not only on technological strength but also on financing ability and capital operation level.

The support of the capital market has also further promoted enterprises to make strategic investments. Alibaba said that it is promoting the construction of a 380 – billion – yuan AI infrastructure and plans to increase investment; ByteDance initially plans to invest 160 billion yuan in AI development in 2026; Tencent and Baidu said that they will upgrade their R & D architecture and establish independent departments to strengthen the R & D system.

At the policy level, the “Artificial Intelligence +” initiative has been continuously deepened. Data from the National Internet Information Office shows that as of early November, 611 generative AI services have completed the filing nationwide, and the average monthly filing number has increased by 55% compared with last year, indicating that the industry has entered the fast – lane of standardized development.

Zhang Xiaorong, the dean of the Deep Technology Research Institute, pointed out that different from the expression of “main business + AI” in the prospectus last year, AI has become a real source of income and infrastructure for enterprises this year. As these companies move from the primary market to the secondary market, it is expected that the “AI content” of listed companies will continue to increase in 2026, and the integration of the capital market and the AI industry will enter a new stage.

Collaboration is the key to lowering the threshold of AI applications

Although capital continues to pour in and the market remains highly popular, for the AI end – side industry to mature from a concept, it still needs to overcome multiple tests such as technology, ecosystem, and cost. Making the model operate reliably, compliantly, and profitably in the real world has become the core issue of the industry’s evolution in 2026.

An unnamed person related to the AI industry said that currently, there are a large number of products in the market, but few have become phenomenon – level applications. There are multiple constraints behind this: due to the limitations of computing power and power consumption of terminal devices, it is difficult to ensure smooth and stable user experience; the application ecosystem is still in its early stage, and most functions have not yet met the high – frequency and essential needs of users’ daily lives; the model capabilities are also diversified. General models need to be improved in logical understanding, task execution, and cross – modal coordination, while vertical models need to be further optimized in professional depth, operational reliability, and scenario adaptability.

Xu Siyan, a senior researcher at the Tencent Research Institute, added from the perspective of the implementation of embodied intelligence that although robots have gradually penetrated into scenarios such as industry, logistics, and services and undertake high – risk tasks, the industry still faces key bottlenecks such as high costs, insufficient technological maturity, and lack of real interaction data.

Facing the challenges, technology companies are exploring different ecological breakthrough paths. The ByteDance ecosystem deeply integrates AI into the content ecosystem through “traffic aggregation and distribution” and focuses on entertainment and information portals; the Alibaba ecosystem positions AI as a productivity and life portal around people’s livelihood service infrastructures such as e – commerce, finance, and local life, thus constructing a commercial link of “tool generation – service implementation – payment closed – loop”. At the same time, manufacturers such as Tencent and Baidu are also accelerating the embedding of generative capabilities into high – frequency scenarios such as payment, social networking, and maps, and building user stickiness and competitive barriers through ecological synergy.

When it comes to the core logic of future ecological competition, in the view of a relevant person from Ant Group, the competitive barrier in the AI era has evolved into an intelligent flywheel of “data – model – ecosystem”: rich scenarios and massive user data train more accurate models, intelligent models improve user experience and attract more users, and continuous user interaction feeds back data and model iteration. When this flywheel is combined with the platform’s commercial performance ability (such as payment and offline services), the moat formed will far exceed the ability boundary of pure technology companies.

Technological innovation is generally regarded as the key to solving the cost dilemma and promoting ecological expansion. A relevant person from Volcengine pointed out that in the future, the core task of the large – model industry is not internal competition but to jointly expand the market. Reducing costs through continuous technological optimization, lowering the threshold of AI applications, and promoting the inclusive development of AI are the basic prerequisites for ecological synergy.

From a longer – term industrial perspective, a relevant person from Alibaba believes that in the transformation from AGI (Artificial General Intelligence) to ASI (Artificial Superintelligence), large models will gradually become the next – generation operating system, and AI cloud will constitute the next – generation computer. Based on this judgment, Alibaba has fully open – sourced Tongyi Qianwen with the goal of creating the “Android system” in the AI era to accelerate the synergy of the entire industry through an open ecosystem.

Looking forward to 2026, Zhang Xiaorong pointed out that ecological synergy will develop further: leading platforms will continue to integrate full – scenario resources and promote the in – depth penetration of generative capabilities in the entire ecological chain; at the same time, the division of labor within the ecosystem will be clearer. Platform enterprises with traffic and technological advantages will form a closer collaborative relationship with real – economy industries with vertical data and offline scenarios to jointly define the new business rules of the AI era.

This article is from the WeChat official account “Finance” (ID: mycaijing), author: Shu Zhijuan, editor: Gao Suying, published by 36Kr with authorization.