Dear Readers,

Six of the top six AI models on OpenRouter by weekly token consumption are Chinese, Alibaba's Qwen has spawned over 100,000 derivatives, and DeepSeek is preparing to run its next frontier model entirely on Huawei chips, no NVIDIA silicon required. If that doesn't make you pause, it should.

In today's deep dive, we unpack China's new Five-Year Plan and what it reveals about the most ambitious national AI strategy ever committed to paper, one that treats artificial intelligence not as a sector but as the operating system for an entire economy.

We trace the numbers behind the open-source explosion, examine why DeepSeek V4 on Huawei's Ascend chips could mark a genuine inflection point in semiconductor independence, and ask the uncomfortable question Western policymakers would rather defer: are export controls still buying time, or have they already become yesterday's tool? Whether you work in AI, invest in it, or simply want to understand the forces reshaping global technology power, this issue lays out the evidence, and it demands your attention.

All the best,

Kim Isenberg

In early March 2026, Beijing unveiled its 15th Five-Year Plan, the policy blueprint that will steer the country through 2030. On paper, it reads like any other sprawling government document. In practice, it might be the most ambitious technology strategy any nation has ever put to writing. The plan does not merely mention artificial intelligence as one priority among many. It positions AI as the central nervous system of China's entire economic future, a cross-cutting infrastructure to be woven into factories, hospitals, farms, energy grids, and government services. The shorthand for this vision is "AI+", a formula that signals AI is no longer a laboratory curiosity but the oxygen of industrial transformation.

This comes at a pivotal moment. Global AI competition has intensified dramatically over the past year, and the old assumption that China trails the United States by a comfortable margin is crumbling under the weight of new data. Chinese AI models now dominate global usage rankings on platforms like OpenRouter, where all six of the top models by weekly token consumption came from Chinese developers during the first week of April 2026 (Global Times, 04/08/2026). On Hugging Face, the world's largest open-source AI platform, Chinese models have overtaken American ones in both monthly and cumulative downloads, accounting for a plurality of 41% of all downloads over the past year (Hugging Face, 03/2026). Meanwhile, DeepSeek, the Chinese lab that stunned the industry in early 2025, is reportedly preparing to launch its V4 model entirely on Huawei's domestically designed Ascend chips, potentially becoming the first frontier AI system built to run without a single NVIDIA GPU (Reuters, 04/03/2026).

Hence, the question this raises is not abstract. It is perhaps the most consequential question in technology today: Is China on the verge of building a fully self-sufficient AI ecosystem, one that renders Western export controls obsolete and reshapes global technology power? Or is this progress, however impressive, still constrained by structural bottlenecks that the Five-Year Plan cannot simply plan away?