Observed Signal · Jul 2, 2026 · Technical Release · Source: Hello China Tech · Impact: 4/5 · Sentiment: Positive
Large Language Models (LLM) & AI Market: China demonstrates domestic AI training at scale
Multiple Chinese organisations in mid‑2026 reported large-scale model training runs using domestically produced processors, challenging the view that high-performance training still required Nvidia hardware. Meituan said it pre-trained LongCat‑2.0 (a trillion-parameter model) on a 50,000‑chip domestic cluster. A Huawei‑included research consortium reported full-parameter post‑training of DeepSeek‑V4‑Pro (1.6T MoE) on over 1,000 Ascend 910C chips, and Huawei’s Pangu Ultra runs and variants were also trained on large Ascend NPU clusters. Baidu reported training ERNIE 5.1 on a Kunlunxin-powered cluster, with industry reports linking that effort to Kunlun P800 chips and a claimed 97% effective training rate. Financial Times coverage (citing Morgan Stanley data) says some Chinese chips already outperform Nvidia’s H20 on several metrics. Together these datapoints indicate Chinese domestic chips have entered the AI training stack, though questions remain about training efficiency compared with best‑in‑class Nvidia solutions.
Multiple large-scale training runs on domestic Chinese processors indicate a structural shift in AI training capability that tests export-control assumptions and could affect global model‑training supply chains and competitive dynamics with major chip vendors.
Wichtigste Kernpunkte & Evidenz
- Meituan released LongCat-2.0 and says the trillion-parameter model was pre-trained from scratch on a 50,000-chip cluster powered entirely by domestically produced processors.
- A research consortium including Huawei completed full-parameter post-training of DeepSeek-V4-Pro (a 1.6-trillion-parameter mixture-of-experts model) on a cluster of over 1,000 Ascend 910C chips with zero interruptions.
- Huawei’s Pangu Ultra 135B was pre-trained on 8,192 Ascend NPUs with 13.2 trillion tokens; Pangu Ultra MoE 718B trained on roughly 6,000 NPUs and reported 30% MFU.
- Baidu said a key version of ERNIE 5.1 was trained on a cluster powered by its Kunlunxin chip unit; industry reports link this to the Kunlun P800 and cite a 97% effective training rate.
- The Financial Times (citing Morgan Stanley data) reports Chinese AI chips already outperform Nvidia H20 on several metrics including processing performance and memory bandwidth.
Verknüpfte Unternehmen
8 verknüpfte UnternehmenBernstein Research
Ein führendes, unabhängiges Joint Venture für erstklassiges institutionelles Equity Research und globale Cash-Equities-Brokerage-Dienstleistungen.
“Bernstein analysts described Huawei’s recent chip advances as another “DeepSeek moment.”...”
Huawei
Diversifizierter Technologiekonzern, der Endgeräte, Cloud-Infrastruktur, App-Distribution und Werbelösungen umfasst.
“Separately, a research consortium including Huawei completed full-parameter post-training of DeepSeek-V4-Pro......”
NVIDIA
Ein führendes Unternehmen für Accelerated Computing, das KI-Software, Cloud-Infrastruktur und Gaming-Technologien bereitstellt.
“For 2 years, the working assumption among investors and policymakers has been clear: Chinese AI chips could handle inference... Training... ...”
Baidu
Baidu ist ein führender chinesischer Technologiekonzern, der ein hochintegriertes Ökosystem aus KI-gestützter Websuche, datengesteuerten AdTech-Lösungen und hochskalierbarer Enterprise-Cloud-Infrastruktur betreibt.
“Baidu, the search and AI group, says a key version of its ERNIE 5.1 model was trained on a cluster powered by its Kunlunxin chip unit......”
Meituan
Chinesische Super-App für lokale Dienstleistungen, die On-Demand-Lieferungen, E-Commerce und Händlerwerbung in einer Plattform vereint.
“In late June, Meituan, China’s largest on-demand services platform, released LongCat-2.0....”
DeepSeek
DeepSeek ist ein führender LLM-Entwickler, der hocheffiziente KI-Modelle über eine performante API und Consumer-Chat-Schnittstellen bereitstellt.
“a 1.6-trillion-parameter mixture-of-experts model from Chinese AI lab DeepSeek...”
Morgan Stanley
Globale Bank und Vermögensverwaltung für Institutionen, Unternehmen und anspruchsvolle Anleger.
“According to Morgan Stanley data compiled by the Financial Times, Chinese vendors’ AI chips already outperform the Nvidia H20......”
Financial Times
Premium-Plattform für globale Wirtschaftsnachrichten, fundierte Marktanalysen und hochpräzise B2B-Werbeumgebungen für anspruchsvolle Zielgruppen.
“According to Morgan Stanley data compiled by the Financial Times, Chinese vendors’ AI chips already outperform the Nvidia H20......”
Ontology Mapping & Concepts
Track Real-Time Market Signals & Shifts
Set up custom watchlists to receive automated, evidence-grounded executive digests whenever material signals or shifts occur across your tracked landscape.