The video highlights a massive scale-up in AI usage, but it also reveals a desperate scramble to bridge a critical computing power gap. It portrays a revolution driven more by brute-force infrastructure spending than by fundamental algorithmic breakthroughs.
Deep Dive
Prerequisite Knowledge
- No data available.
Where to go next
- No data available.
Deep Dive
The token surge: Powering China's AI industrial revolutionAdded:
Tokens are the fuel that powers artificial intelligence. For Li Jian Wei, an AI content creator, this fuel feeds his team of custom AI agents he playfully calls his lobster squad.
I have an info lobster for data research, PPT lobster for slide creation, and a local lobster for regional content tasks.
I can't afford a personal assistant, so I built my own AI lobster team to handle my schedule.
>> These AI agents handle nearly all his daily work, but his monthly token consumption has skyrocketed. Even more than the cost, Li struggles with a bigger problem. Tokens are often hard to buy.
It's like buying mobile data plans. The small and medium token packages are always sold out, leaving only the large, high-capacity option, costing over 400 yuan per month. As user demand for tokens surges, how have token supplies responded? At Moon Shot AI, business director Huang Chenxin is in the middle of an unprecedented computing power crunch.
We used to focus on promoting our models to enterprises. Now, I spend most of my time apologizing to clients. Every client needs massive new token volumes daily. Tokens have become an extremely scarce resource. Huang now coordinates computing power delivery down to the minute. He describes this as a good problem to have, one that arises for a paradigm shift in large model applications. Since this year, AI agents have boomed, upgrading large models from simple chat tools to digital employees that execute complex tasks. Processing long documents, writing code, analyzing data, and generating videos now consume dozens of times more tokens than basic chat. Enterprise-scale deployment has triggered an explosive rise in token usage.
Token usage per user is 10 times higher than last year. We've moved from casual chatting to work-critical AI applications. Moon Shot AI's paid user orders jumped nearly 100 times month-on-month in January alone. The 140 trillion daily token calls are reshaping more than just one company. What about the token factories, the computing centers?
The surge in token demand has created a computing power shortage. China will face a computing power gap of 200 EFLOPS in 2026. We've set a 25.5 billion yuan investment target for computing infrastructure this year. We also need to boost computing power efficiency.
Data centers run around the clock with chips and power systems operating at full capacity. China is accelerating the construction of national hub computing clusters to meet surging demand. The industrial chain transformation driven by the token boom is quietly becoming a new engine of the country's economic growth.
Related Videos
The #1 Reason Your Top People Keep Leaving (How to Fix It)
Entreleadership
470 views•2026-05-29
What Happens After A Motorcycle Dealership Shuts Down?
FastestWay.1
374 views•2026-05-29
The Evolution of DSP's Pokemon Unpack-ack-acking Grift
Toxicity_Unmasked
2K views•2026-05-29
Help re-structure my finances, I want to buy a house, save and invest
JennNxumalo
2K views•2026-05-29
Asian Paints Q4 Results: Revenue Beats Estimates, 5 Key Takeaways For Investors
NDTVProfitIndia
111 views•2026-05-29
Trying to Afford Vancouver on a Single Income | $2,550 Mortgage
chelseaspursuit
308 views•2026-05-28
Are you busy but still feeling broke?
TaraWagner
305 views•2026-06-01
7 Nigerian Stocks That Could Explode Because of Dangote Refinery IPO
femiakinwale9269
478 views•2026-05-29











