Walls and wiring

While the US is building a small yard with high fences, China has opened its AI ecosystem to all

When Western pundits analyze China’s burgeoning artificial intelligence strategy, the default lens is often geopolitical. Following the recent World AI Conference 2026 in Shanghai, headlines in the West framed Beijing’s push for open-source models, the establishment of the World AI Cooperation Organization and the rollout of public-good AI tools such as the MAZU meteorological system as a “Trojan horse” for digital hegemony.

But viewing China’s global AI outreach solely through a Cold War prism is a profound misunderstanding of how the Chinese economic engine actually works. Strip away the geopolitical rhetoric, and what emerges is not a sudden, state-mandated ideological crusade, but the continuation of a highly sophisticated, battle-tested international business strategy that Chinese enterprises have been perfecting for nearly three decades.

To understand China’s AI playbook, one must look back to the telecommunications boom between 1998 and 2008.

At the turn of the millennium, Western telecom giants such as Ericsson and Nokia approached developing nations in Africa, Asia and Latin America with expensive, rigid landline infrastructure models. For many of these nations, the capital expenditure was prohibitive. Then came Huawei and ZTE. Rather than competing on the West’s terms, Chinese companies offered flexible, cost-effective GSM mobile networks. More importantly, they didn’t just sell equipment; they provided “turnkey” solutions — handling network design, construction, local training and pairing those services with financing from Chinese State-owned banks.

Many Global South countries effectively leapfrogged large-scale fixed-line infrastructure and moved rapidly into the mobile age, and Chinese telecom companies gained substantial market share by offering affordable, adaptable solutions tailored to local needs.

Today, history is rhyming in the realm of AI.

Western AI laboratories, such as OpenAI and Anthropic, are the new Ericssons. They offer premium, closed-garden API services that are prohibitively expensive for startups and governments in the developing world to afford at scale. China, conversely, is offering the AI equivalent of the GSM leapfrog. By championing open-weight models such as Qwen, DeepSeek and Kimi, and pairing them with cloud infrastructure, regional data centers and the pledge of 5,000 AI training opportunities for developing countries under the WAICO framework, China is lowering the barrier to entry. The message to the Global South is clear: countries do not need to spend billions building their own sovereign supercomputers; they can build their digital future on China’s open, comparatively accessible infrastructure.

Western skeptics frequently ask: Why give away the “crown jewels” of AI model weights for free? Is this just State-subsidized dumping?

The answer lies in China’s domestic internet history — specifically, the legendary e-commerce war between Taobao and eBay in 2003. When eBay charged Chinese merchants listing fees via EachNet, Alibaba’s Taobao countered by making basic storefronts entirely free. This “free entry” model drained eBay of its user base, allowing Taobao to achieve total market dominance before eventually monetizing the ecosystem through value-added advertising and premium services.

We are now witnessing the “Taobao Playbook” applied to global AI. Take the pricing strategy of Moonshot’s Kimi K3 model for example. The model’s weights are free to download for researchers, students and small enterprises. However, if a company uses the model to power a commercial API service and crosses a revenue threshold of $20 million, a paid commercial license is triggered.

This is not charity; it is a strategy to build a broader technological ecosystem. While US companies attempt to harvest immediate, high-margin API revenues (the “eBay model”), Chinese AI companies are using open-weight access to attract developers worldwide and expand the adoption of their technological platforms. If the Global South’s digital economy increasingly relies on Chinese foundational models, significant commercial opportunities could emerge in the future.

China’s genius lies in “selective adaptation”. China rarely adopts Western templates wholesale; it indigenizes them to fit local institutional voids and market realities.

In the AI sector, there has been a globalized adaptation of China’s “Triple Helix” university-industry-government model of innovation. In the West, the government often plays a peripheral regulatory role. In China’s global AI push, the State provides the diplomatic umbrella, sovereign financing and macro-level trust, while agile private tech companies execute aggressive, market-driven ecosystem plays. It is a combination of State-backed strategic patience with hyper-competitive commercial execution.

For the international community, the implications are significant. Washington’s “small yard, high fences” strategy — relying on chip export controls to cage China’s AI ambitions — may slow down China’s progress at the absolute cutting edge of hardware. But sanctions are unlikely to eliminate the appeal of a business model that aligns with the development needs of much of the Global South.

While the West builds walled gardens and restricts access, China is actively wiring the rest of the world into its digital ecosystem. To understand China’s AI ambitions, observers must stop reading geopolitical thrillers and start reading business history. The script was written nearly three decades ago in the telecom sector, and it is currently being executed with devastating commercial precision in the age of AI.

The author is a former executive vice-president of the Hungarian Communications Authority, a communications researcher at China-CEE Institute (Budapest) and a professor at Pazmany Peter Catholic University in Budapest, Hungary.

The author contributed this article to China Watch, a think tank powered by China Daily. The views do not necessarily reflect those of China Daily.

Contact the editor at editor@chinawatch.cn.