National supercomputing internet surpasses 1 million users: the country's largest computing resource pool takes shape

2026-01-04 07:40:09

Abstract generation in progress

【CryptoWorld】The National Supercomputing Internet Platform has just announced a milestone figure—over 1 million users. What does this mean? Just look at the hardware investments behind it: currently, it has connected more than 30 national supercomputing centers and intelligent computing centers from 14 provinces and cities, forming a massive heterogeneous computing resource pool with a total scale of over 150,000 accelerators and 2 million CPU cores.

From another perspective, this platform can now provide nearly 70 different specifications of computing power configurations, covering a wide range of scenarios from general-purpose computing to specialized acceleration. For Web3, AI training, scientific computing, and other compute-intensive applications, such infrastructure upgrades mean cheaper and easier access to powerful computing resources.

View Original

This page may contain third-party content, which is provided for information purposes only (not representations/warranties) and should not be considered as an endorsement of its views by Gate, nor as financial or professional advice. See Disclaimer for details.

22 Likes

Reward
22
10
Repost
Share

Comment

0/400

PaperHandsCriminal

· 01-07 02:17

1 million users? Haha, now the computing power thing is really going to be democratized, no longer just a toy exclusive to big players. --- Speaking of 2 million cores sounds impressive, but I guess only those who can really use them, retail investors like us will still have to wait in line. --- 70 configurations? Honestly, that's a bit exaggerated; in the end, it's still the 5 types that are used. --- Cheap? I think it's relatively cheap, but it still hurts the wallet. --- Does Web3 training models really require that much computing power? Feels a bit like overbuilding.

View OriginalReply0

MEVictim

· 01-07 02:16

1 million users? This mining pool is really strong now, AI training costs are plummeting. When computing power becomes cheap, Web3 won't be far behind. This is getting interesting. 150,000 acceleration cards, 2 million cores... Just thinking about it is exciting. We finally don't have to be squeezed by cloud providers. This wave of national infrastructure policies was definitely worth the investment. Building the ecosystem is the real profit. But it still needs some time before it becomes truly affordable. Middlemen will still make a profit from the price difference, haha.

View OriginalReply0

WagmiOrRekt

· 01-05 06:58

1 million users have broken through, but the real game is just beginning Is the computing power cost going to collapse? The miners are about to cry ngl this infrastructure is indeed impressive, just not sure how cheap it can get

View OriginalReply0

EyeOfTheTokenStorm

· 01-04 08:10

From a quantitative perspective, surpassing the 1 million user mark itself is a bottoming signal. 150,000+ acceleration cards, 2 million CPU cores... this data structure is very similar to the eve of the cloud computing boom in 2017, but the key question is—now that policies are actively intervening, could there be an invisible liquidity lock? The democratization of computing power sounds good, but who benefits cheaply? Web3 has always been a concentration of risk assets. Once cheap computing power is supported, could there be another round of speculative cycles? Historical data shows that within 6 months after infrastructure upgrades, there are often intense emotional fluctuations... That said, this indeed changes the local technical supply and demand relationship. Friends involved in T trading can observe the trends of related concept stocks, but I still recommend prioritizing risk.

View OriginalReply0

HashBard

· 01-04 08:04

Ngl, 1 million users sounds impressive, but the real narrative arc is here—150,000 GPU acceleration cards mark the beginning of the democratization of computing power. The days when only miners and AI labs dominated are over; now ordinary people can join... This is a signal of a sentiment shift. --- So this is a national-level infrastructure play? It’s a bit like watching dragons be tamed in Game of Thrones... Why hasn't the market reacted yet? --- 70 specifications, 2 million cores... I’m actually wondering, will such pooling eventually turn into another form of centralization? The decentralization dream might be slapped in the face before it even wakes up. --- Calm analysis: what the Web3 community currently lacks is exactly this. Cheap computing power means more people can run nodes, more people can participate. Discourse will change. --- Wait, what’s the funding logic behind this? The government invests so much in hardware, who will cover the final ROI model... It’s a bit too romantic.

View OriginalReply0

PancakeFlippa

· 01-04 08:02

One million users, what the heck? Can this mining pool really become cheaper? I just want to know. --- 150,000 yuan acceleration card sounds pretty intimidating, but the final price will probably be marked up by middlemen. --- Web3 training models finally have affordable computing power, now no need to be exploited by cloud providers. --- 70 configurations, sounds fancy but how many are actually practical? 🤔 --- National-level infrastructure, I always feel efficiency might be hampered. Hope it doesn't disappoint.

View OriginalReply0

AlwaysMissingTops

· 01-04 08:00

1 million users surpassed, sounds impressive, but this wave of infrastructure is really coming AI training costs can be reduced, and the Web3 ecosystem is also saved Over 30 national supercomputing centers collaborating, this scale is indeed quite impressive But when will it truly become affordable for us retail investors to use... 2 million CPU cores, just numbers on paper, can it truly benefit us retail investors? Democratization of computing power, I've heard this a hundred times, is this really happening or just another hype? 70 different configurations, so many options, experiencing choice paralysis Spending 700,000 yuan on a model, changing to 70,000 yuan, is this called progress? When the government steps in, it's different, but I just want to know when the prices will be made public

View OriginalReply0

ZkProofPudding

· 01-04 07:55

Wow, 2 million CPU cores? The cost of AI training must plummet now, finally no more being gouged by cloud providers.

View OriginalReply0

AirdropHunterWang

· 01-04 07:49

Brothers, now the computing power is no longer a concern. The AI projects before kept shouting that the computing power was too expensive, but this time there's real hope. The concept of 2 million cores... feels like we need to redefine the cost structure. For us Web3 project developers, the biggest fear is infrastructure bottlenecks. This time, it looks reliable, with 14 provinces and cities connected, and the scale is there. But how cheap it really is depends on how they set the prices. Anyway, it's another option, which is better than only being able to rely on cloud service providers before. If this thing really takes off, the miners will probably need to reshuffle again.

View OriginalReply0

GasWaster

· 01-04 07:41

Computing power costs are dropping linearly, which is really good news for on-chain computation. 150,000 accelerators stacked together? Wow, that scale is indeed outrageous. Web3 infrastructure is becoming more competitive, and the national teams are taking action—it's a different level. Wait, will this cheap computing power be monopolized again by big players... 100 million users sounds like a lot, but how many can actually use it? This will significantly reduce the costs of training AI models, for sure. Guo Chao is still quietly making a fortune; why didn't I pay attention before? By the way, what does 2 million cores mean? How many chain nodes can it run? Damn, a computing power price war is coming, and small miners are about to face another bloodshed.

View OriginalReply0