The large-scale model market in china has emerged from the initial chaos after undergoing twenty months of baptism. The layouts of major manufacturers are becoming clearer, preparing for a new round of growth.
According to asia vets & finance APP, Omdia stated that the large-scale model market in china has emerged from the initial chaos after undergoing twenty months of baptism. The layouts of major manufacturers are becoming clearer, preparing for a new round of growth. Omdia predicts that asia pacific's artificial intelligence will grow to 24.2 billion in 2025 and nearly double to 58.4 billion in 2028. Some manufacturers focusing on their own software services have already embedded large models into various daily applications using multimodal capabilities. Others are focusing on empowering enterprises and artificial intelligence developers.
Chief analyst of ai at Omdia, Lianjie Su, stated that any comments on large model manufacturers that only focus on parameter and benchmark score comparisons no longer reflect the real situation in the market.
Cloud giants such as Alibaba (09988), Baidu (09888), and Tencent (00700) have actively released their own enterprise RAG and intelligent body products to help companies quickly deploy large model applications, achieving cost reduction and efficiency improvement. Based on the needs of leading domestic enterprises, enterprise-level RAG features multimodal data processing, high-precision search, optimized generation result reranking, and modular capabilities. The intelligent body can deploy a framework for multi-agent collaborative cooperation to handle complex tasks.
Meanwhile, led by Chinese companies like Lenovo (00992), Xiaomi (01810), and Ubtech (09880), large models are also beginning to move from the cloud to the edge, changing the interaction mode between humans and terminals on smartphones, computers, vehicles, and robots. Large models will ultimately branch out into various applications, relying on cloud-edge-end computing power to achieve automation and intelligence.