China bypasses US sanctions and creates its own analogues of AI chips

Post Thumbnail

Financial Times journalists discovered on satellite images 3 previously unknown factories located near the Huawei company base. According to insiders, these enterprises produce the latest chips for artificial intelligence. This is AI Ascend 910D. Interestingly, the construction of these facilities began immediately after the US imposed the first technological sanctions against China.

And rumors are spreading on social networks that Huawei’s new chips may surpass the H100 model from Nvidia in performance. This is the current standard for training large language models. Moreover, there is information that the Chinese DeepSeek-R2 model is already being trained on AI Ascend cards.

Financial Times sources claim that all 3 discovered factories are built in the same architectural style and receive subsidies from the Chinese government. And supposedly this indicates that the development of proprietary chips for artificial intelligence has become a national priority for China. Probably the next shocking discovery will be that all Chinese government institutions use the same flag.

The situation is especially important against the background of current restrictions. Currently, the import of top-tier Blackwell series graphics cards to China is prohibited. Which creates a serious shortage of computing power for Chinese technology companies and startups working in the field of artificial intelligence. If AI Ascend chips can indeed replace H100 graphics cards in terms of characteristics, this could radically change the balance of power in the global artificial intelligence race. Chinese companies will gain access to computing power comparable to that used by their Western competitors. But without dependence on American manufacturers.

Experts note that such technological import substitution is a long-term trend in the Chinese economy. Which began long before the current deterioration of relations with the US. However, it is now that this strategy is beginning to bring concrete results in the most sensitive areas of high technology. Well, the question remains whether the characteristics of Chinese chips actually correspond to the declared parameters. An objective assessment can only be given after the appearance of independent tests and real examples of using this equipment in complex artificial intelligence tasks.

Почитать из последнего
UBTech will send Walker S2 robots to serve on China's border for $37 million
Chinese company UBTech won a contract for $37 million. And will send humanoid robots Walker S2 to serve on China's border with Vietnam. South China Morning Post reports that the robots will interact with tourists and staff, perform logistics operations, inspect cargo and patrol the area. And characteristically — they can independently change their battery.
Anthropic accidentally revealed an internal document about Claude's "soul"
Anthropic accidentally revealed the "soul" of artificial intelligence to a user. And this is not a metaphor. This is a quite specific internal document.
Jensen Huang ordered Nvidia employees to use AI everywhere
Jensen Huang announced total mobilization under the banner of artificial intelligence inside Nvidia. And this is no longer a recommendation. This is a requirement.
AI chatbots generate content that exacerbates eating disorders
A joint study by Stanford University and the Center for Democracy and Technology showed a disturbing picture. Chatbots with artificial intelligence pose a serious risk to people with eating disorders. Scientists warn that neural networks hand out harmful advice about diets. They suggest ways to hide the disorder and generate "inspiring weight loss content" that worsens the problem.
OpenAGI released the Lux model that overtakes Google and OpenAI
Startup OpenAGI released the Lux model for computer control and claims this is a breakthrough. According to benchmarks, the model overtakes analogues from Google, OpenAI and Anthropic by a whole generation. Moreover, it works faster. About 1 second per step instead of 3 seconds for competitors. And 10 times cheaper in cost per processing 1 token.