DeepSeek Plans 1GW AI Data Center in Inner Mongolia to Rival U.S. Giants
DeepSeek is scaling up its AI infrastructure with a massive 1GW data center project in Inner Mongolia, intensifying the global tech race.
Stock photo for illustration only, not from the actual event
- DeepSeek aims to construct a 1-gigawatt AI data center in Ulanqab, Inner Mongolia.
- Partial operations are projected to launch between late 2027 and early 2028.
- Ulanqab offers an average temperature of 4°C, dramatically cutting server cooling costs.
- The project faces scrutiny amid strict U.S. export controls and geopolitical tensions.
The artificial intelligence landscape is heating up as DeepSeek moves forward with plans to build a massive 1-gigawatt AI data center in Ulanqab, Inner Mongolia. This monumental undertaking marks the company's largest infrastructure project to date and is poised to become one of China's most expansive AI computing hubs once completed, positioning the firm to better contend with leading U.S. technology giants.
The strategy involves a mix of self-built facilities and leased computing power from external providers, with initial operations anticipated to go live by late 2027 or early 2028. While hardware specifications regarding chip models remain unconfirmed, Ulanqab has rapidly emerged as a prime hub, attracting over a dozen firms to seek local government approval for similar projects.
Stock photo for illustration only, not from the actual event
The city's year-round cool climate, averaging around 4 degrees Celsius, provides an immense advantage by drastically reducing the primary operational expense of server cooling—the exact reason why several other major tech players have established roots there. Despite setting records domestically, this scale still lags behind American counterparts like OpenAI and Anthropic, which are pumping tens of billions of dollars into 3 to 5-gigawatt facilities. Meanwhile, fellow Chinese firm Z.AI is also developing a 1-gigawatt project relying entirely on domestic chips, backed by DeepSeek's recent $7 billion funding round that valued the company at roughly $50 billion as it eyes an IPO.
China's aggressive push into localized AI infrastructure highlights a calculated strategy to circumvent foreign hardware restrictions while optimizing operational expenditures. By leveraging geographic advantages like cold climates for natural cooling, Chinese labs are effectively lowering the cost barrier to compete against western alternatives.
Nevertheless, regulatory hurdles loom large as the Trump administration investigates whether DeepSeek is bypassing restrictions to utilize restricted Nvidia Blackwell chips in Inner Mongolia. Compounding these pressures are accusations from OpenAI and Anthropic regarding model distillation practices, alongside fierce domestic rivalry from companies like Moonshot, Alibaba, and MiniMax, all while Chinese AI services maintain a tenfold cost advantage over their American counterparts.
Source: Techsauce
Found something wrong in this article? Report an issue with this article
Comments
Leave a Comment