China’s $295 Billion AI Buildout Would Need 80% Home-Made Chips
China has a huge new plan. It wants to spend $295 billion on AI. AI means artificial intelligence, which is computer software that can learn and think a bit like people. A “buildout” just means building all the computers and special buildings that run AI.
The big surprise is one rule. At least 80% of the technology, including the AI chips, must come from Chinese companies. That one rule could lock out famous American chip makers like Nvidia and AMD. This story was reported by The Decoder.
In short, China wants to build its AI future mostly with its own parts. Below we explain what the plan is, who is behind it, and why it matters for India, business founders, and the wider tech world.
What is the plan?
The plan is to spend about $295 billion over five years. That is about 2 trillion yuan. Yuan is China’s money, just like the rupee is India’s money.
This money would build a big network of data centers. A data center is a building full of strong computers. These computers run AI and store huge amounts of information.
China wants this network finished by 2028. But the cost goes higher when you add power plants and electricity systems. AI computers use a lot of electricity. So China will also need to spend on power. With power added, the total reaches at least 5 trillion yuan.
The 80% home-made chips rule
The main rule is simple but strong. At least 80% of the technology, including the AI chips, must come from “domestic” suppliers. Domestic just means companies based inside China.
An AI chip is a small but powerful computer part. It does the heavy math that AI needs, and it does it very fast. For many years, the best AI chips came from American firms like Nvidia and AMD. This new 80% rule would keep those two out of China’s big plan.
China is not starting from zero. Its government has checked and approved several home-made chips. The report says nine Chinese AI chips passed a government safety check. Approved suppliers include Huawei, Alibaba, and Shanghai Biren Technology, and a few others.
Who is behind the buildout?
The plan is being written by China’s National Development and Reform Commission. Think of it as a top government group that plans the country’s economy.
This buildout is part of a bigger program called “Six Networks.” That program was announced earlier in 2026.
State-owned companies are leading the work. State-owned means the government owns them. Two big phone companies, China Mobile and China Telecom, would run most of the data centers. They already run large networks all across the country.
How will it be paid for?
A project this big needs money from many places. The report says the money will come from several sources mixed together:
- Ultra-long-term government bonds. A bond is a loan. The government borrows money now and pays it back slowly over many years.
- State investment funds. These are big pools of government money.
- Bank loans.
- Private capital. This is money from private companies and investors.
The goal is bigger than just the data centers. China wants its whole AI industry to grow past 10 trillion yuan in total value.
Key facts
| Item | Reported figure |
|---|---|
| Data center investment | ~$295 billion (about 2 trillion yuan) |
| Time period | Over five years |
| Network completion target | 2028 |
| Total spend (with power) | At least 5 trillion yuan |
| Domestic technology required | At least 80% (including AI chips) |
| Chinese AI chips cleared | Nine passed security review |
| AI industry value target | More than 10 trillion yuan |
What it means for US chip suppliers
For American chip makers, this is bad news. China has been one of the biggest buyers of AI chips. But the 80% rule would shut the door on Nvidia and AMD for this huge plan. That means they could lose a lot of sales and a lot of money.
There is also a Taiwan angle. Taiwan is a key place where the most advanced chips are made. The report says Taiwan is thinking about making AI chip smuggling into China a crime for the first time. (Smuggling means moving goods in secret and against the rules.) Taiwan is also looking at limiting sales to buyers who want very powerful chips. In short, the chip world is splitting apart, and the walls between China and others are getting higher.
FAQ
How much is China planning to spend?
About $295 billion (around 2 trillion yuan) on the data center network over five years. With power systems added, the total reaches at least 5 trillion yuan.
What does the 80% rule mean?
At least 80% of the technology, including AI chips, must be made by Chinese companies. This basically keeps Nvidia and AMD out of the plan.
Which Chinese companies will supply the chips?
Approved suppliers include Huawei, Alibaba, and Shanghai Biren Technology, and a few others. Nine Chinese AI chips have passed a government safety check.
When is it meant to be ready?
The linked data center network is meant to be finished by 2028.
Why it matters (especially for India / founders)
This story shows a clear trend. Countries want to build their own AI base. They do not want to depend on other countries. China is spending big on local chips and data centers. That way, it will not get stuck if other countries cut off supplies. India is thinking in the same way, with its own push to build data centers and chips.
For founders (people who start companies) and students, there are two big lessons. First, AI is now a top goal for whole countries. So skills in AI, chips, and data centers will be in high demand. Second, supply chains are splitting along borders. A supply chain is the path that parts and goods take from the maker to the buyer. A startup that depends on only one country’s chips faces real risk. Having backups and local options is now smart business, not just a “nice to have.”
The bottom line: China’s $295 billion AI plan, with its 80% home-made chip rule, is one of the boldest bets yet on building AI on your own. It could change who sells chips, where money flows, and how fast the global AI race speeds up. Keep watching, because the effects will reach far beyond China.
Source: The Decoder
