The short version
- Construction of large-scale data centers is accelerating in rural China to support national AI ambitions despite hardware restrictions.
- Beijing promotes open-source models as a competitive advantage and a counter to US closed-system approaches, seeking greater influence in global governance.
- Environmental concerns regarding water usage and energy sources persist alongside reports of village relocations for these new facilities.
A significant expansion of digital infrastructure is underway in the grasslands of Inner Mongolia, where traditional agriculture now shares space with massive construction sites. In Ulanqab, a region historically known as China’s potato capital, workers are erecting large rectangular facilities designed to house computing power for artificial intelligence platforms. This development occurs against the backdrop of high-level diplomatic meetings between US and Chinese leaders, who are discussing AI safety and other geopolitical issues. The physical growth of these centers reflects a broader strategic push by Beijing to embed AI technologies across ninety percent of its industries and society by 2030.
The construction pace has been described by local workers as rapid, with multiple phases of development already underway or planned. These facilities are critical to China’s economic and national security objectives, serving as the backbone for a technological contest with the United States. The sites are located in sparsely populated provinces including Ningxia, Gansu, and Guizhou, chosen for their cooler climates, lower land costs, and existing renewable energy capacity. This geographic distribution aims to mitigate some environmental impacts while leveraging regional advantages for power-intensive computing tasks.
Beyond physical infrastructure, China is cultivating a domestic talent pool to sustain this growth. Academic observers note a shift in migration patterns, with fewer Chinese students and researchers heading to the United States and more returning home. Some attribute this trend to immigration restrictions imposed by the Trump administration and a desire for greater government support and resources available domestically. Researchers who previously studied abroad are finding opportunities to apply their work to fields like human health and medicine within China’s expanding ecosystem.
The strategic approach to AI development in China differs markedly from that of its American counterparts. While US firms such as OpenAI and Anthropic often rely on closed-model approaches where design elements remain proprietary, Beijing is encouraging the creation of affordable, open-source models. This strategy allows other entities to modify and adapt the technology, fostering rapid innovation. Proponents argue this method has helped Chinese companies overcome challenges related to limited access to advanced semiconductor chips, which are restricted by US export controls.
Tensions over these technological disparities remain high. US officials and industry leaders warn that a Chinese lead in AI could pose significant dangers to global security. Dario Amodei, CEO of Anthropic, has advocated for maintaining bans on selling advanced equipment to China, citing the risks associated with unchecked development. Conversely, state-aligned media in Beijing have dismissed these concerns as part of a Cold War-era playbook intended to contain China’s rise. These conflicting narratives complicate diplomatic efforts to establish cooperative frameworks for AI governance.
Despite the rhetoric, both nations face pressure to address the rapid pace of technological advancement. Industry insiders have raised alarms about the speed at which AI capabilities are evolving, prompting calls for regulatory intervention. President Trump has emphasized the competitive stakes, asserting that dominance in AI is crucial for national success. Meanwhile, Chinese leadership continues to advocate for a global regulatory framework, seeking a larger voice in setting international standards. The upcoming summit at the White House will test whether collaboration on safety measures is feasible amidst these deep-seated strategic rivalries.
The environmental and social costs of this infrastructure boom are not negligible. While China aims to power most data centers with green energy by 2030, parts of the grid in these regions still rely on coal. Water scarcity remains a concern in arid areas like Inner Mongolia, where cooling systems for servers demand significant resources. Additionally, there have been reports of entire villages being relocated to make way for construction, though details regarding compensation and the exact number of displaced residents are difficult to verify due to limited public discussion on the topic.
As the physical structures rise in Ulanqab and other regions, they symbolize a broader shift in the global AI landscape. The competition is no longer just about software algorithms but also about the tangible assets that support them. With hundreds of data centers being built for major tech firms like Huawei, ByteDance, and DeepSeek, China is solidifying its position as a key player in the industry. The outcome of this race will likely influence not only technological innovation but also international relations and regulatory norms for years to come.
Sources behind this briefing
Go to the original reporting
- BBC Technology↗A new skyline of data centres is rising from the rolling plains of rural China
