Last year, the Trump administration launched a new AI initiative, Pax Silica, "to secure supply chains for AI models, semiconductors, and critical minerals," according to CNBC.
China countered in July with the “World Artificial Intelligence Cooperation Organization," which promotes the country's "open-weight" technology. When several Pax Silica nations joined the Chinese group as well, Washington realized it was going to be an "us vs. them" world on AI and decided to put its foot down.
A draft letter prepared by the State Department and addressed to the 35 signatories of a U.S. “AI Opportunity Statement,” states that countries must choose a side; they cannot belong to both Pax Silica and the Chinese organization.
“To be part of everything is to be part of nothing. Signature of the Pax Silica Declaration is not merely a membership subscription, but a commitment,” the letter says, urging countries to “choose deliberately” on AI.
“It cannot be held alongside membership in duplicative initiatives whose expectations conflict with our own,” the letter said, without specifically mentioning China.
Here's the biggest problem for the U.S.: China's "open-weight" technology is much cheaper than the proprietary models owned by Anthropic, OpenAI, and other U.S. companies.
Open-weight technology refers to artificial intelligence models where the underlying trained parameters, known as weights, are made publicly accessible to developers and researchers to download, modify, and run locally. The pre-trained numerical values (the learned knowledge and patterns resulting from training) are downloadable, allowing users to fine-tune or run the model on their own hardware.
It's a much less expensive way to adopt AI, though open-weight models may not include the full training datasets, source code, or training pipelines needed to recreate the model from scratch. For many business applications, it's a much cheaper alternative.
China is making a massive bet that open-weight models will beat Silicon Valley's models that are "predicated on investing billions and getting users paying top dollar for the most powerful, proprietary AI technology," according to Bloomberg.
Export controls imposed by Washington restrict China’s access to US-designed chips that are around 20% faster and consume as much as 30% less power than their Chinese competitors as they squeeze more transistors onto each sheet of silicon. China’s tech community has sought to overcome this handicap by developing AI software that is more “compressed.” This means fewer steps of calculation are required to achieve outputs comparable to those of the best-performing US AI models.
One way this is achieved is by employing a technique known as mixture of experts. When a user prompts a Chinese chatbot such as DeepSeek or Alibaba’s Qwen, the model doesn’t need to mobilize its entire “neural network” to generate a response (the process known as inference). Instead, it activates specialized subnetworks known as experts that engage only a fraction of the software’s available computational capacity.
"The models of U.S. labs such as OpenAI and Anthropic often have the edge when it comes to the sheer volume of information they can handle in a single interaction and the sophistication of their responses," states Bloomberg. "But to achieve this, they activate far more of a model’s available “neurons” and thereby consume a lot more computing power."
According to LiveBench’s large language model rankings, the OpenAI, Anthropic, and Google models are still the top-performing AI models. But three Chinese models have moved into the top ten, challenging the U.S. for dominance.
The question facing CEOs is: how exactly are they going to adopt AI, what tasks do they want it to perform, and how much are they willing to spend?
On raw cost, China has a significant edge now and will have an even bigger one in the near future. According to The Deep View, "even as unit economics improve, overall AI costs continue to rise without a clear or predictable path to matching value."
"The current environment has created a top-down fervor, in fact a mandate, for virtually all enterprises to aggressively adopt AI en masse," Scott Bickley, Advisory Fellow at Info-Tech Research Group, told The Deep View. "This blind foray into the AI abyss often lacks the in-depth understanding of the total cost of ownership, can ignore the culture of technology adoption within a given enterprise, and makes it difficult for one to advocate for anything but an 'innovator/early adopter' position."
The U.S. is right to make countries choose between America and China. It's very hard to see at this point whether China's AI models will surpass the U.S. proprietary models in more than just cost. U.S. models may offer more value in performance than China's cost efficiencies.
Who wins and who loses will determine the future for both countries.
