DeepSeek orders 160,000 Huawei chips for a 1-gigawatt site
Eugene / Chips and Infrastructure desk
DeepSeek just placed one of the largest known orders for domestically produced Chinese AI accelerators.
The order
Per Bloomberg, DeepSeek ordered at least 160,000 Huawei Ascend 950DT chips for a 1-gigawatt data center in Ulanqab, Inner Mongolia. At full load the site will draw as much power as 750,000 homes.
Inference on Huawei, training on Nvidia
The split matters. The Ascend chips are intended for inference, the work of serving trained models to users, while DeepSeek’s training continues to run on Nvidia hardware. The order is a bet that domestic silicon can carry the volume side of the business even while the frontier side still depends on US-designed processors.
The open question is delivery. An order of this size against a chip line that is still ramping means the timeline, not the intent, will determine when the gigawatt actually lights up.
Sources
Related
Chinese state hackers more than doubled their attacks after adopting DeepSeek
Taiwanese research firm TeamT5 told Bloomberg on August 24, 2026 that state-affiliated Chinese hacking groups have more than doubled the number of attacks they carry out since delegating routine tasks to open-source AI models and using them to develop malicious software. DeepSeek is the model of choice; TeamT5 chief analyst Charles Li attributes that to it being relatively powerful with very low cyber guardrails. The report names specific uses: Grimfengxi generated exploit code, Huapi targeted a Taiwanese company's email system, and Teleboyi collected 1,000 IP addresses and mapped a target's domains. Researchers say they have not yet observed a more expensive model such as Kimi K3 used in an attack. The UK AI Security Institute warned in May 2026 that models' cyber capabilities are doubling every few months.
DeepSeek is building a gigawatt of compute in Inner Mongolia
Bloomberg reported on July 30, 2026 that Hangzhou-based DeepSeek is adding one gigawatt of compute in Ulanqab, Inner Mongolia, about 350 kilometers northwest of Beijing, building its own campus while leasing extra capacity, with part expected online by end of 2027 or start of 2028. The chips are undecided: Nvidia, Huawei, DeepSeek's own future silicon, or a mix. It would be the largest AI facility any Chinese company operates.
DeepSeek shipped V4 Pro and raised API prices up to 12x
On August 13, 2026 DeepSeek shipped DeepSeek-V4-Pro-0813, the production version of its agent-focused flagship, and announced API price increases effective August 16 at 16:00 UTC. V4 Pro output tokens rise from $0.87 to $3.96 per million during peak hours, and cache-hit input jumps from $0.003625 to $0.044 per million, roughly 12x. Off-peak hours run at half the peak rate. The same day, Cointelegraph reported OpenAI and Anthropic cutting prices under pressure from Chinese rivals.
ANOTHER News is published by ANOTHER, an AI-native content agency. Daily coverage also runs on Instagram.