Aug 27,2026, (This is an article by AI, but after several rounds of discussion) Nvidia’s latest earnings report was not just about another record quarter. It offered an important clue about where the AI infrastructure boom is actually happening.
Nvidia reported quarterly revenue of $96.2 billion, up 106%, with Data Center revenue reaching $89 billion, up 117%. For the next quarter, Nvidia expects revenue of about $108 billion. Even more striking, CEO Jensen Huang said the company could still grow revenue by roughly 70% in fiscal 2028.
In other words, Nvidia sees no sign that the AI infrastructure boom is running out of steam.
But there is another number worth watching: China accounted for less than 1% of Nvidia’s Data Center revenue from Hopper products in the latest quarter, and Nvidia’s next-quarter guidance assumes no Data Center computing revenue from China.
That is interesting when viewed alongside China’s optical industry.
Eoptolink, one of China’s leading optical-transceiver companies, generates roughly 98% of its revenue overseas. Much of the global AI infrastructure demand it serves is coming from the U.S. market.
This raises a simple question:
If China is becoming a major AI-model power, where is the infrastructure behind those models?
China certainly has enormous data-center capacity. It is investing heavily in computing centers across Inner Mongolia, Ningxia and other regions. But much of this infrastructure is described simply as “computing centers” rather than dedicated AI data centers.
The contrast with the U.S. is striking.
In America, the AI infrastructure chain is increasingly visible:
Nvidia GPUs → hyperscalers → AI data centers → networking → optical modules → power infrastructure.
Microsoft, Google, Amazon and Meta are all spending enormous amounts of money building AI capacity. Nvidia is at the center of this ecosystem, and its latest results suggest that spending is still accelerating.
China’s AI ecosystem looks different.
DeepSeek, Doubao and Qwen are clearly major AI models, but it is much harder to identify exactly where their computing capacity sits, how much is dedicated to AI, how many accelerators are deployed and how much capital is being invested.
Ulanqab is one of the clearest examples. DeepSeek has been linked to plans for a large AI computing project there. But beyond projects like this, the scale of China’s dedicated AI data-center buildout is much harder to see.
This does not mean China does not need AI infrastructure. It does.
The more interesting question is how much infrastructure is required to produce the same amount of AI computing.
The U.S. model is heavily based on Nvidia GPUs. As GPU clusters become larger, the amount of data moving between chips explodes. That drives demand for high-speed networking, 800G and 1.6T optical modules, and eventually co-packaged optics.
China is developing a more domestic ecosystem based on Huawei and other Chinese accelerators, together with model optimization and shared computing infrastructure.
Perhaps this means China can develop powerful AI models without creating the same level of demand for Nvidia GPUs and high-end optical infrastructure.
We do not yet have enough data to prove that.
But Nvidia’s earnings make one thing clear:
The AI infrastructure race is still accelerating—and so far, the most visible battlefield is the United States.
China may be winning important battles in AI models. But the bigger question for investors is whether it is building physical AI infrastructure at the same scale.
For companies such as Eoptolink, that distinction matters.
The next phase of the AI boom may not be determined only by who builds the best models, but by who builds the most computing power, networking capacity and optical infrastructure behind them.