China plans a $295 billion AI infrastructure investment, but commentator Teortaxes argues capital cannot solve its chip constraints
Story Overview
Beijing's draft five-year blueprint earmarks roughly 2 trillion yuan for a web of interconnected data centers meant to power AI across healthcare, transport, and urban systems, with state operators like China Mobile expected to run most of the show and local suppliers targeted for at least 80 percent of the hardware.
Money meets the real choke point
Commentator Teortaxes notes that power grids and land are already handled separately and that earlier huge chip outlays produced setbacks, leaving datacenter construction itself as the lesser obstacle.
Early talks leave room for shifts
The plan remains in initial discussions with no fixed allocations or milestones yet confirmed, and any integration with the power grid could push the total spend toward 5 trillion yuan.
Many users praise China's $295B AI data center plan for its sustainable cost-efficient focus on practical productivity gains versus US overspending, while some dismiss the total as too low.
Most Activity
but what would they even spend $1.5T on? Grid and power are being taken care of separately. Land is not an issue. Chips… last time China tried to solve chip problems with a lot of money, they got a big fiasco. In any case, their bottleneck is not datacenter construction.
Need to be 5x higher
Nah... China probably does NOT need that 5x spending...
Let the U.S. spend whatever it wants on AI infrastructure... China has never tried to win by matching dollar 4 dollar spending...
The U.S. model is often: throw $5 at the problem and hope it works... well, it does solve the problem...
However, Chinese model usually is: spend $1, optimize it, improve utilization... shorten the supply chain, drive down hardware costs, and figure out how to get a similar outcome...
People said China could NOT build a high-speed rail network without spending like the West...
People said China could NOT deploy renewable energy at scale...
Now China needs 5× more AI infrastructure investment?
Maybe... Or maybe the real question is NOT who spends the most...
It is who gets the most compute, the lowest cost per token, the highest utilization rate, and the fastest iteration cycle...
In technology, the winner is NOT always the one who spends the most money...
Lmao, quite often, it is the one who converts money into capability most efficiently...
@teortaxesTex This allocation of resources has nothing to do with ai. Xi’s digging a system of tunnels for the regime to migrate once d-day 2 commences on tianjin
@zephyr_z9 Details are as follows 👇👇
@zephyr_z9 Details are as follows 👇
@zephyr_z9 Money already goes 5x further in China, if their military budget is any indication
@zephyr_z9 Honestly, this is about 1/4th of the Belt and Road Initiative cost.
@zephyr_z9 This news was already published on the official website of the Chinese government in 2025，Bloomberg posted it again today with ulterior motives.…
@zephyr_z9 🍹🤎
🌦️
🫰
@zephyr_z9 My strategy plan.
🔻↩️↩️
@zephyr_z9 obviously bbg want to pump, original news in Jan. 2025 (2tn RMB, equiv. 300bn USD) https://www.tjftz.gov.cn/contents/7236/371633.html
@zephyr_z9 梭哈China AI
@zephyr_z9 that’s definitely an ambitious target, how are you handling the latency?
@zephyr_z9 My strategy plan. .
🔻↩️↩️
@zephyr_z9 This is public sector I think, private sector would ramp it up 5x?
@teortaxesTex Not to mention that’s converted to dollars. In PPP, China will go much further than what $295 billion would achieve in the U.S. in construction of databases etc.
@zephyr_z9 With the cost advantages of China, it's easily equivalent to serveral trillion USD spent inside America. It's more than it looks like.
@zephyr_z9 So higher for $DRAM . Cxmt has no capacity to export
@zephyr_z9 A billion in China goes much further than US
@zephyr_z9 We'd likely include energy generation etc in this fund in the west. They've already got that covered in other funds. They'll stand up a f* ton of GPUs with "just" this amount.