A new forecast from the China Telecom Research Institute points to a sharp shift inside China’s AI industry: by 2029, inference is expected to make up 80% of the country’s computing-power demand, as the market moves from training large models toward deploying and selling AI agents. The same report says Chinese technology companies will spend close to 600 billion yuan, about $89 billion, on AI in 2025, equal to more than a tenth of all investment in the country.
Training is the one-time cost of building a model. Inference is what every user request costs to run. That makes inference an operating expense, and it is also where the margin now sits, since companies are increasingly optimising existing models rather than building new ones.
What the China Telecom Research Institute report says
The report, carried by state broadcaster CCTV on Saturday, frames a fast pivot. Chinese technology companies are expected to spend close to 600 billion yuan, about $89 billion, on AI this year. The report puts that at more than a tenth of all investment in the country.
It expects agents to drive close to tenfold annual growth in computing demand over the next two to three years, with inference overtaking training as the dominant workload by 2029.
Why inference is overtaking training
Training builds a model once. Inference is what it costs every time somebody uses one, which makes it an operating expense rather than a capital one. As more products ship AI features to real users, the steady per-request cost of running a model adds up faster than the upfront cost of building it.
The report treats that crossover, where inference demand exceeds training demand, as the defining feature of the next phase of China’s AI build-out.
Europe is building for the same demand on a different clock
The European Commission opened bidding in July for up to seven AI gigafactories, a EUR 30 billion programme with about EUR 10 billion of public money and EUR 20 billion hoped for from private investors. Roughly EUR 1 billion of it is actually committed today.
Applications close on 12 November. Awards are expected in early 2027, construction starts that year, and the machines are due to run by mid-2028. On paper that arrives a year before China’s projected crossover.
What has already slipped
The European timeline has already moved. Bidding was pushed from May to July. The evaluation criteria were delayed more than once. Interest from companies has narrowed from about 70 interested parties to roughly ten expected bidders.
Most of the public half of the EUR 30 billion programme depends on a budget for 2028 to 2035 that member states have not agreed. Until that lands, only about EUR 1 billion is actually committed.
Where the two regions stand
Chinese technology companies are on course to spend about $89 billion on AI in 2025. Europe’s entire gigafactory programme, even at full EUR 30 billion, works out to roughly the same spend every five days inside China this year.
China’s report is a forecast about running models. Europe is still deciding where to put the buildings.
FAQ
What does the China Telecom Research Institute forecast for AI inference?
It expects inference to account for 80% of China’s computing-power market by 2029, overtaking demand from training.
How much are Chinese tech companies spending on AI in 2025?
They are expected to spend close to 600 billion yuan, about $89 billion, on AI this year, which the report puts at more than a tenth of all investment in the country.
What is the status of Europe’s AI gigafactory plan?
The Commission opened bidding in July for up to seven gigafactories on a EUR 30 billion budget, with applications closing on 12 November, awards expected in early 2027, construction starting that year, and machines due to run by mid-2028. Roughly EUR 1 billion is committed so far, and most of the public funding depends on a 2028 to 2035 budget member states have not agreed.
This article summarizes reporting from thenextweb.com. See our editorial disclaimer for how our articles are produced.
Run a free scan to see your AI Visibility Score, SEO rating, and local citation accuracy.