China has planted itself in the CES with a single objective: to gain muscle. The humanoid robotics is one of the pillars of Chinese technological development and at the fair held in Las Vegas they made their ambition clear. The other speech with which they have gone to the United States is with that of the open source AI. It is something that clashes with the American approach and there is one name that stands out: Alibaba and, specifically, its Qwen series.
The reason is that it is the spark of life for China’s artificial intelligence strategy.
In short. Xinhua is the official news agency of the Chinese government and has recently outstanding that the homeland technology companies have attended the international stage to demonstrate how collaboration between industries and, above all, models Open source are the key to the future technological network.
The case of Qwen stands out. This is a series of AI models developed by Alibaba (one of the giants of e-commerce, online payment and cloud storage services). Alibaba’s strategy with its model is that it be openand the objective is that whoever wants can base themselves on it to create their variants. This is important because Alibaba has the muscle to create such a model, and other companies that wish to do so can take advantage of it to modify it and adapt it to their needs.
Sovereignty (word of the year). The objective is to create a network, an ecosystem with accessible tools and, as Xinhua points out, close the digital divide. Because Qwen has a lot of models depending on what they are needed for. It has Coder for programming, Image Edit for image generation, VL for visual recognition either even older to compete against Claude and GPT.
According to According to Chinese media, international developers have already created more than 100,000 variations based on Qwen series models and more than 700 million downloads. And Alibaba’s is not the only one. The R3 from DeepSeek is another that operates under an open license. Both have a final objective: to promote technological sovereignty.
Driving physical AI. Developing artificial intelligence models is… expensive. We have seen it thoroughly in recent months, with big technology companies burning money to create huge data centers powered by very expensive graphics cards used in AI training, a RAM that has become prohibitive and astronomical energy needs (soon, literally this “astronomical).
Therefore, AI being open source means that large companies can create their model so that others can then bring their technology to life using a modified version of a previously trained model. Thanks to AIs from DeepSeek and Alibaba, Chinese robotics unicorns like Unitree or Agibot are developing their products, which will be the “Physical AI” by being able to interact with the “real world”.
Other industries. But it’s not just about robots. Fields in which AI is very beneficial, such as research or medicinecan take advantage of that open-source philosophy. For example, a previously under-resourced medical institution can access highly capable models that would otherwise be limited to major hospitals and research centers.
And the others? The summary is that China’s vision is that AI models respond to a strategic, but also global, interest. Open source technology can fuel other projects that require AI, but AI is not the ultimate goal. And here the big question is what is being done outside of China. The model in Silicon Valley has been different.
It is evident that this innovation exists and Big Tech is the engine of AI worldwidebut this software is more closed and controlled. Curiously, who had a more open approach was Meta with LLaMAalthough if the plans for 2026 are fulfilled, It will also become a more closed model. Because, in the end, in China this opening is internal policy, while in the US there is pressure from investors who protect those proprietary assets.
Now, not all.
NVIDIA in the game, of course. Jensen Huang is one of the standout names so far this year. The CEO of NVIDIA has sent a conciliatory message in the trade and technology war between China and the United States and has also appeased the search for a new TSMC. Furthermore, it is clear that the advancement of open models is something that will ensure let no one be left behind. Google with Gemma either OpenAI with GPT-OSS They are already offering semi-open models.
An example is the collaboration between Nvidia and Siemens with the aim of creating a kind of “AI-based operating system” for industrial segments. Meanwhile, although in Europe there is reputation for regulating a lot and inventing littlesteps are being taken to promote that open source model that boosts European competitiveness in the sector. MistralFor example, it is the great European reference and has open versions.
Beyond the data highlighted by Xinhua, which logically sweeps home those more than 100,000 versions derived from Qwen, what stands out is what seems to be a trend: less protectionism and more collaboration with the idea that, as Huang points out, no one is left behind on the AI train.
Images | Nick Wood (edited), Xataka with Mockuuups Studio
GIPHY App Key not set. Please check settings