Prospects have emerged that the price structure of enterprise software (SW) could be shaken as low-cost artificial intelligence (AI) models from China and Asia spread. Analysts say that as cheap AI models quickly replace general-purpose tasks, the premium billing structure of high-priced frontier models and SW built on them could also come under pressure.
Gartner Inc., a global IT advisory firm, analyzed in a recent report titled “China Will Not Replace Western AI. Instead, It Will Reshape Enterprise Software Pricing” that as inexpensive AI models appear in large numbers, the need for companies to use high-priced models for general tasks is decreasing.
Gartner analyzed that Asian AI led by China and open-weight models are pulling down the “cost floor” of the global AI market. For repetitive tasks where results can be verified relatively easily—such as translation, information extraction, classification, summarization, code generation, and customer support drafting—it is expected that using a combination of frontier models, low-cost models, and small language models (SLMs) depending on the task will expand rather than relying on a single high-performance frontier model.
“The entry of Chinese AI into the West is not a direct replacement of models, and as low-cost Chinese and Asian AI models spread, the overall cost of using AI will decrease,” Gartner said. “In the future, the value of software that selects, manages, and connects multiple models according to enterprise tasks will grow larger than the model itself.”
This trend is already appearing in the market. According to OpenRouter, a global AI model platform, in June, DeepSeek recorded a token market share of about 20% as of early June, rising to a top-tier model since mid-May, and the usage proportion of Chinese models such as Xiaomi, MiniMax, and Tencent also rose together. OpenRouter's weekly token usage also increased nearly fivefold in about five months.
Accordingly, Gartner projected that the way enterprises adopt AI will shift from purchasing a specific model to choosing and using multiple models according to task characteristics. “The center of value moves back to software,” Gartner explained. “The competitiveness of software that selects and connects multiple models according to tasks and manages permissions, security, and governance will become more important than the AI model itself.”
This change is also expected to affect the AI billing system of enterprise SW. Gartner projected that by the third quarter of 2027, vendors charging extra fees per user for general-purpose copilots using a single AI model or family of models without proving business outcomes will face pricing pressure. The analysis indicates that in the future, actual business outcomes—such as document processing, error reduction, and reduction in working hours—will become more important than receiving a blanket selection simply for featuring AI capabilities.
