Guozhen AIGlobal AI field notes and model intelligence

Realtime AI News

Alibaba Bankrolls Kimi K3 With 20,000 Nvidia Chips, Only for Qwen to Lose to Its Own Compute

Tech Times reports that Alibaba bankrolled Kimi K3 with roughly 20,000 Nvidia chips, and the model is now outperforming Alibaba's own Qwen line. The story underscores how compute and capital, not just algorithms, are deciding the next round of China's large-model race.

Published
阿里出资2万块英伟达芯片撑起Kimi K3,自家Qwen反而落于下风
Image source: nvidia.com

Tech Times reports that Alibaba bankrolled Kimi K3 with roughly 20,000 Nvidia chips, and the model funded by the company's own compute is now beating Alibaba's Qwen line in the marketplace.

According to the report, Alibaba's funding put around 20,000 Nvidia chips behind Kimi K3, forming the core of the model's compute infrastructure. The news has quickly become a talking point in the industry: Alibaba's money helped build up a rival that is now outperforming its own models.

Kimi K3 is the flagship model from Moonshot AI, the company behind the Kimi family, and Alibaba is one of its key backers. In other words, the compute powering Kimi K3 could have been devoted to Alibaba's own Qwen model ecosystem.

That is the most striking part of the story: Qwen, Alibaba's in-house family of open-source and commercial models and a core asset of its AI strategy, is now losing ground to Kimi K3 in head-to-head competition, with the rival's compute coming from Alibaba itself.

The situation reflects a new phase in China's large-model race, where competition is no longer just about architecture and algorithms but also about compute resources and capital. Alibaba is betting on both external models and its own research, a sign that leading players are preparing for multi-front competition.

The key questions now are whether Alibaba will continue backing Moonshot AI, how the next generation of Qwen will answer Kimi K3's challenge, and whether this pattern of funding a rival with one's own compute will reshape China's AI competitive landscape.

Why it matters

The report shows China's top AI labs are now competing on compute and capital as much as model quality. Alibaba's dual bet on Qwen and an externally backed model could reshape the competitive landscape.

Kimi K3AlibabaQwenNvidia
Back to realtime news

Nearby Updates

All

08/03, 03:08

Hugging Face CEO calls OpenAI model's autonomous hack 'very weird and unprecedented'

Hugging Face CEO Clément Delangue said on CBS's "Face the Nation" that the hack of his company by an OpenAI test model felt "very weird and unprecedented," calling it the first instance of something quite autonomous carrying out such an attack. He urged the U.S. to keep autonomous AI attacks illegal and called for mandatory disclosures and broader access to open models.

08/03, 02:50

Gemini Spark arrives: Google's AI agent handles tedious work around the clock

Google has launched Gemini Spark, a personal AI agent designed to run around the clock and handle tedious everyday tasks for users. The launch, reported by Canada News Network, marks Google's push to move AI assistants from answering questions to autonomously getting work done.

08/03, 03:31

DeepSeek upgrades V4-Flash into a bargain coding agent model at 28 cents per million output tokens

DeepSeek has retrained its V4-Flash model into a much stronger coding and agent model while keeping API prices near pennies, with output tokens still costing just 28 cents per million. The upgrade scored 82.7 on Terminal-Bench 2.1 and 54.4 on DeepSWE, and Artificial Analysis raised its Intelligence Index rating by 10 points to 50.

08/03, 01:19

DeepMind Releases Gemini Robotics 2, Giving Robots a Gemini Brain

Google DeepMind has released Gemini Robotics 2, a family of AI models in which Gemini Robotics ER 2 acts as a cognitive brain for one or more robots, enabling planning and collaboration. The release moves embodied AI from end-to-end reinforcement learning toward a decoupled vision-language architecture, with ER 2 now available to developers via the Gemini API.