A Chinese vision for AI's development seems the smartest. Other countries should follow an example
China uses AI in the approach “here and now”, not “someday” – And they do it with full state support. American giants invest billions and gigawatts in models, closer to Aga, counting on strategic and scientific advantage. Chinese leadership talks about a strong orientation on applications and intentionally rewards implementation, which immediately increase the efficiency of the economy.
In China, in addition to the development of general purpose models, you can see primarily a rapid increase in applications in education, weather forecasting, public order, agriculture, as well as in numerous factories. A simple logic works in the background: if Agi turns out to be distant or economically elusive, The advantage will be won by the one who will squeeze the maximum from today's AI. This is exactly the calculation of Chinese government media and documents, as well as analyzes of Western commentators.
Check also: “Godfather” AI warns against artificial intelligence
“AI+” as China's strategy and goals for 2027–2030
The state gives this direction and pace. The Chinese government published “AI+” guidelines in August, which are to blend in with six pillars of development – from science and industry to public services and social well -being. The document assumes that by 2027 the indicators of the penetration of intelligent terminals and AI agents will exceed 70 percent, and until 2030, 90 percent, making an intelligent economy an important driving force. This is a plan for wide application and an anchor for regional investment and implementation programs.
In addition to regulation, money and infrastructure appear. In January 2025, a state industry fund worth approximately 60 billion yuan (approx. $ 8.2 billion) was launched for early AI projects, and a few months later the National Guidance Fund with a scale of 1 trillion of Juan was announced, to mobilize private capital for technology companies. These are mechanisms focused on hundreds of thousands of implementations and thousands of SMEsnot for individual training records.
As a result, smaller, closer to users Data Center “implementation” are more often created than giant campuses for training the most powerful models.
Read also: IT after new. AI, agents and automation will reformat the labor market. Check how to win it
Cheaper and open source. Usability before perfection
Open or with open scales have become a large “accelerator”. Deepseek-R1 drew the attention of the scientific community with the cost-possibilities in the tasks of reasoning. In addition It is easy to adapt and run with smaller budgetswhich clearly accelerated adoption.
At the same time, Alibaba aggressively develops the family of Qwen models, widely made available to communities and enterprises, which builds a critical mass of tools, integration and staff. In practice, this means that Tens of thousands of companies can fold their own AI “sandwiches” from Open Source blocks instead of waiting for the next closed breakthroughs.
This turn towards Open Source can also be seen in the debate in the USA. Leaders like Yann Lecun, vice president and chief scientist dealing with AI in Meta, argue that it is the open approaches that push the industry ahead the fastest.
At the level of Chinese sectors, the image is consistent. The University of Tsinghua inaugurates “AI agent hospital”, where Doctors are supported by virtual colleagues equipped with the latest medical knowledgewhich is to shorten the diagnosis time and unify the quality of clinical decisions. Local administration implements agents to service mass hotlines and classify residents' applications. In the production of work and vision systems, real time controls, a Agriculture gets consulting based on weather and soil data. Each of these examples does not require Aga – it only needs efficient agents, well -tagged data and sensible integration with the process.
The strategy focused on applications “here and now” maximizes the return on capital and energy. Instead of leading the costs of training to billions of parameters and gigawatts, China widespread smaller models and agents where What counts is the cost of the user and the time you need to get business value. This philosophy allows you to learn from leaders' mistakes, take over what works, and standardize the tools in supply chains faster.
At the same time, restrictions in access to the highest shelf of chip, which are the result of American export controls, somehow force efficiency – they stimulate work on lighter, domain models and optimization of infection, and not only on training records. From the macro side, this gives a real, gradual increase in TFP (complete productivity of the factors), instead of a binary plant for the uncertain breakthrough of Aga.
See also: Artificial intelligence is not a friend. When teenagers talk to bots, health and life can be at stake
Costs and risks
This business development model is not free from tension. However, if Aga arrives faster, the cognitive and patent gap can materialize almost overnight and then the Chinese ecosystem will have to urgently buy general competences.
What's more, export hardware barriers can limit the scale of infection and services in the cloud, which was already visible with rapid increases in demand on queries to API. The regulating regime in China, which combines development and safety, can also quickly tighten the screw – it is politically understandable, but creates implementation uncertainty for innovation at the risk limit. The balance for today, however, remains beneficial to the implementation close to the process, where the rate of return on each yuan and kilowatt hours is easy to measure.
So if a given company expects with AI specifics within 3-12 months, the Chinese thinking model is worth imitating. Instead of obsessed with the “biggest” model, the most important are integrations, the quality of operational data, Governance and control of the cost of the task. In practice, the portfolio of agents and models adapted to tasks, running close to data sources, and the ability to quickly measure ROI. Agi – if it comes – will not annul these foundations. For now, those who have carried out productivity here and now will win, as Beijing does as part of their plan “AI+”.
Author: Grzegorz Kubera, Business Insider Polska journalist
