Moonshot's Kimi AI Powered by 20,000 Nvidia Chips from Alibaba
Moonshot built on 20k Nvidia chip cluster from Alibaba
Moonshot has secured a computing agreement with Alibaba to utilize a cluster of 20,000 Nvidia chips for its Kimi AI models. This partnership highlights China's ongoing dependence on Western semiconductor technology to drive its artificial intelligence advancements. The massive hardware infrastructure supports the recent launch of Kimi K3, an open-source model that competes directly with leading systems from OpenAI and Anthropic.
Chinese AI champion Moonshot has a computing power agreement with Alibaba Group Holding Ltd. for the use of around 20,000 Nvidia Corp. chips, underscoring China's continued reliance on Western semiconductors to fuel its AI development.
- HarHarVeryFunny
Interesting if true - that Moonshot can train a ~3T SOTA model on only 20K NVIDIA GPUs, while others like Musk (who freely admits to distilling OpenAI's models) struggle to build a competitive 1T model (Grok 4.5) with massively more compute (Colossus-1 100-250K GPUs, Colossus-2 500K+ GPUs).
I guess at least partly a reflection of all the optimizations in the Kimi 3 architecture.
In the recent leaked DeepSeek investor meeting, they also mentioned only having a 20K GPU cluster (unclear if NVIDIA, or Huawei).
- a-priori
It kind of confirms a hypothesis I have that the next phase of AI development will be about getting smaller (in terms of model size and compute), because smaller is more capital efficient for training (allowing faster iteration and more iteration cycles for a given amount of capital), allows for denser inference (more inference for a given amount of compute hardware), and allows for more edge inference applications.
The goal will be to develop smaller models with more efficient architectures, that have similar or even better performance than larger models.
- infecto
My thesis is still that model building has no moat. Folks continue to migrate around between the big labs. There is a lot of value in having good taste around the harness and how the models are used. The medium to long term winners will be the folks that control the compute.
- idoxer
- ycui7
K3 is natively trained to mxfp4, if they cannot get a hold of Blackwell chip, it is meaningless. Hopper does not do native 4-bit floating math.
Either they have Blackwell with native 4-bit floating math, or they use have Chinese domestic NPU that support mxfp4 natively.
The article’s statement does not make sense.