The open-source AI race just got more crowded. Alibaba launched a new AI model on Monday, Qwen3.8-27B, built to run directly on laptops, and released the full weights of its most powerful model, Qwen3.8 Max, making it freely downloadable. The move lands just a week after Meta announced its own laptop-ready open-source models under the name Muse Glimmer, as the U.S. tech giant tries to position itself as the American alternative to more proprietary labs like OpenAI and Anthropic.
The numbers suggest Alibaba already has real developer traction. The company says Qwen3.8-27B matches the performance of models ten times its size, and Hugging Face — one of the largest hubs for downloading open-weight AI models — reported that Qwen-based models now account for 151,448 derivative projects, roughly 2.6 times Meta’s entire footprint in the category. Analysts see the real fight shifting away from massive data centers toward “edge” AI that runs locally on devices like laptops and phones, which can be faster and more private. Alibaba, alongside Chinese peers DeepSeek and Moonshot, has built a commanding lead in that specific race.
For investors, this is a data point worth tracking on two fronts. Alibaba (BABA) is emerging as the clearest proxy for China’s on-device AI ambitions, and continued developer adoption could eventually translate into cloud and hardware revenue. On the U.S. side, watch how Meta responds — its willingness to keep matching open-weight competitors with its own free models is a signal of how seriously it’s taking the competitive threat, and its infrastructure spending plans are worth monitoring for margin impact. Chipmakers building for efficient on-device inference, such as Qualcomm and Arm, are another indirect way to play the shift toward edge AI. The bigger picture for portfolios: the AI trade is broadening beyond a handful of U.S. mega-caps, and investors should keep an eye on which companies actually convert model downloads and developer mindshare into paying customers over the next few quarters.
