Dernière minute
Newsgather
BackDeepSeek Expands Data Centre Operations in Inner Mongolia Ahead of V4 Model Launch
DeepSeek Expands Data Centre Operations in Inner Mongolia Ahead of V4 Model Launch
En développement
SCMP Tech11.04.2026Tech1 dk okumaChina

DeepSeek Expands Data Centre Operations in Inner Mongolia Ahead of V4 Model Launch

L'essentiel

  • Chinese AI startup DeepSeek is hiring for data center roles in Inner Mongolia, signaling an expansion of its computing infrastructure ahead of the V4 model release.
  • This move raises questions about its sourcing of AI chips.

Résumé généré par IA

Pourquoi c'est important

DeepSeek is a Chinese AI startup that gained prominence with its R1 model. The company is now preparing to release its V4 model, prompting speculation about its hardware infrastructure and chip sourcing.

Taille de police

Chinese artificial intelligence start-up DeepSeek is ramping up recruitment in Inner Mongolia, signalling a potential expansion of its data centre footprint ahead of the much-anticipated release of its V4 model.

The Hangzhou-based company is recruiting for two new roles – server maintenance engineers and delivery managers to oversee data centre launches – in Ulanqab, in Inner Mongolia, according to recent job postings.

The move marks the first time DeepSeek has publicly advertised on-site roles tied directly to its computing infrastructure. Previous recruitment efforts were largely focused on software engineers and AI researchers in hubs such as Hangzhou and Beijing.

The hiring push comes ahead of the expected debut of its next-generation flagship model, V4, more than a year after the release of its R1 model propelled the company into the spotlight.

The upcoming launch has fuelled industry speculation over how DeepSeek is sourcing and deploying computing power and AI chips for the new model.

A report by US news outlet The Information last week said V4 would run on Huawei’s latest Ascend 950PR processor, while a separate report by Reuters last month suggested the model had been trained on Nvidia’s Blackwell chips, the US firm’s most advanced processors to date.

À surveiller

Perspective IA — des possibilités, pas des certitudes

  • DeepSeek will officially announce the V4 model release date within the next two months.

    Probable · En quelques mois

  • Further details on DeepSeek's chip sourcing strategy will emerge, potentially confirming or refuting current reports.

    Probable · En quelques semaines

  • DeepSeek's data centre footprint in Inner Mongolia will expand beyond the initial roles advertised.

    Possible · En quelques mois

Questions ouvertes

  • What is the exact timeline for the V4 model release?
  • What are the specific performance improvements of the V4 model compared to R1?
  • What is DeepSeek's strategy for securing a consistent supply of advanced AI chips?
  • Will DeepSeek's data centre expansion in Inner Mongolia be sufficient for the V4 model's demands?

Sujets liés

This article was originally published by SCMP Tech.

Articles liés

从“秀功能”到“能干活”,人形机器人产业咋突破?
En développement·2 sa önce

从“秀功能”到“能干活”,人形机器人产业咋突破?

工业和信息化部、国务院国资委联合发文,启动人形机器人与具身智能实景实训专项行动,力争2026年底前部署超过1万台人形机器人进入商业使用场景。行动聚焦工业、服务、特种三大领域,旨在解决机器人从实验室走向实际场景中面临的续航、环境适应性、可靠性及成本等挑战,并通过实景实训迭代升级机器人“大脑”,推动产业从“秀功能”转向“能干活”。

中国新闻网
Plus sur ce sujetdeepseek