Woofun AI reports that Citrini analyst Jukan asserts the CUDA ecosystem's competitive advantage is diminishing due to AI-driven code generation and advanced languages like TileLang. Jukan cites DeepSeek founder Liang Wenfeng, noting that self-developed compilers significantly reduce dependence on NVIDIA software despite using its GPUs for V3 model training. Liang Wenfeng quantified the gap between Chinese and American chips—hardware efficiency is about 4 times higher, with a time lag of about two years—and revealed that DeepSeek is closely collaborating with Huawei, expecting to receive around 16,000 Huawei AI chips. The Huawei 950 SuperNode can replace the workload of the GB200/GB300.

Jukan characterizes this shift as the end of the CUDA moat, leading to a bearish outlook on NVIDIA. Conversely, Jukan remains bullish on AMD, arguing that AI coding advancements will accelerate the ROCm ecosystem, narrowing the gap with CUDA. AMD recently announced active use of Claude Code for chip design and software engineering when investing in Anthropic. This analysis suggests systematic erosion of NVIDIA's developer stickiness from the software side.