Open-source AI models may undermine closed-model monopolies and benefit cloud providers.

Huo Xing Finance reports that on August 3, Citrini analyst Jukan published a post stating that the ongoing advancement of open-source AI models and the weakening dominance of proprietary models could become a pivotal turning point for the cloud computing industry, enhancing the commercial value of AI infrastructure for cloud providers. Jukan noted that previously, cloud providers faced a core concern: after making massive investments in GPU procurement and data center construction, they might merely serve as infrastructure suppliers for a handful of proprietary model companies, which retained control over user access and pricing. However, as open-source models improve in performance and reduce in cost, the AI application paradigm is shifting from “a single high-performance model handling all tasks” to “layered model routing.” Complex reasoning tasks will still be handled by top-tier proprietary models, while the majority of routine tasks can be efficiently managed by low-cost small models or open-source alternatives. Jukan believes that with increased model substitutability, cloud providers will gain greater access to users, traffic orchestration rights, and pricing power. In the future, proprietary models may no longer function as “paywalls” atop cloud infrastructure but instead resemble compute resources that cloud platforms can freely orchestrate. Moreover, competition from open-source models does not imply reduced hardware demand. Jukan stated that lower inference costs could drive rapid growth in AI query volumes, while model compression, inference optimization, intelligent routing, and proprietary ASIC chips will reduce the amount of general-purpose GPU resources required per token. He argues that the long-term growth of AI infrastructure depends on whether demand growth outpaces efficiency gains. If token usage increases faster than improvements in algorithms and chip efficiency, data center utilization and return on capital can remain high, encouraging continued investment by cloud providers in compute infrastructure. Jukan concluded that the true bull case for AI infrastructure does not simply rely on “cheap models benefit cloud providers, whose growth benefits hardware,” but rather on open-source models reducing monopolistic profits at the model layer, enabling cloud providers to enhance the monetization efficiency of compute resources through orchestration and vertical integration—ultimately creating a positive feedback loop between cloud computing and hardware investment.