GLOBAL RESEARCH ARCHIVE
Initiation: Riding an established supply chain
Research evidence excerpt
Initiation: Riding an established supply chain
Macquarie Equity Research Iluvatar
Investment thesis
Iluvatar has a strong chip design track record among domestic AI chip players, as it wasLongest track record in
founded in 2015 and began focusing on GPGPU chip design in 2018. Its first chip product, TGChina AI chips
Gen 1, reached mass production in Sep 2021, becoming the first successful GPGPU product
in China. This, coupled with the company's relatively established customer relationships
(especially with domestic cloud leaders), should lead to sustainable outperformance going
forward, in our view. Hence, we initiate coverage with an Outperform rating.
Wide product range with a fast iteration strategy
Iluvatar stands out among peers for its fast iteration strategy. Its TG series, focussed on
training functions, reached mass production in Sep 2021, with the second generation
hitting mass production in 4Q23, and the third-generation launching in 3Q24. Additionally, it
launched its ZK series, focusing on inferencing, and also equipping edge and video processing
capabilities. It has rolled out two generations of ZK chips thus far, giving Iluvatar one of
the largest product portfolios among domestic AI chip players. These products also come
in different form factors, including PCIe cards, OAM (open accelerator modules) and high-
density integration systems.
Medium-term strategies
Iluvatar has stated that it plans to focus on: 1) R&D of next-generation GPGPU chips; 2)
development of a software stack to achieve CUDA-compatibility; 3) development of large-
scale computing clusters; 4) collaboration with ecosystem partners; 5) joint R&D with key
value-chain collaborators; and vi) develop the open-source community for domestic AI chips.
Why we are confident
The English excerpt is extracted automatically from the cited source page and may contain layout or recognition errors. It is never batch translated.
Open report viewer