Competition among industry leaders over token supply has evolved from a focus on raw computing scale to output efficiency. Lenovo has unveiled its Lenovo Wanquan Hybrid Token Factory brand and announced plans to build its first Wanquan Hybrid Token Mother Factory in Zhangjiakou. Under the plan, the project will initially deploy a 15,000P heterogeneous computing cluster and an approximately 100-megawatt intelligent computing center, with an annual token output target exceeding 2 quadrillion; by 2030, key metrics are expected to grow more than tenfold compared with the existing baseline. In recent months, Alibaba, ByteDance, Tencent, and other companies have also ramped up their investments in intelligent computing and model services. The competitive focus in the AI infrastructure market is shifting from simply stacking hardware to overall performance—including computing power scheduling, multi-model compatibility, and supply stability.