Moonshot AI has suspended new subscriptions for its Kimi K3 artificial intelligence model. This halt comes after a surge in demand overwhelmed the Beijing-based company's existing infrastructure within days of the model's launch. Moonshot AI stated on X late Sunday that Kimi K3 had received “far more love than we expected,” pushing demand “close to the limits of our current capacity” over the past 48 hours.
The firm is now prioritizing existing subscribers, temporarily pausing new ones while it works to add capacity. This scramble for infrastructure underscores the intense competition for market dominance in the global artificial intelligence sector, where capital accumulation is the driving force.
Who Profits in the AI Race
Kimi K3's rollout late last week aimed directly at challenging established AI models from U.S. corporations like Anthropic and OpenAI. Moonshot described its model, with 2.8 trillion parameters, as the world's largest open-source AI model. This release immediately “rattled U.S. rivals,” according to the base article, and exerted “pressure on American technology stocks.”
Investors now fear that “cheaper Chinese AI models could undercut pricing power and demand at U.S. AI companies,” signaling a direct threat to the profit margins of Western capital. The intensifying U.S.-China tech race highlights the challenges for Chinese AI models in serving a rapidly expanding user base both domestically and internationally. Chinese open-source, lower-cost advanced AI models, including DeepSeek’s latest V4, have been gaining global attention.
About 1 year and 6 months ago, DeepSeek's V4 model “shook world markets,” establishing China as a serious competitor to the U.S. These Chinese frontier AI models are largely open source, allowing broader access for examination and development, a strategy that challenges proprietary Western models and their control over intellectual property.
Infrastructure and Capital Investment
Lian Jye Su, chief analyst at Omdia, a technology research and advisory group, noted that new model releases typically generate massive interest, which can strain existing compute infrastructure. Su stated that Moonshot AI “does not have sufficient compute chips to serve the current surge in demand.” He attributed this primarily to Moonshot not fully anticipating K3’s popularity.
Su also highlighted that K3 is “very demanding” in terms of compute requirements, making the allocation of such resources both “challenging and expensive.” This expense acts as a significant barrier, concentrating control over advanced AI in the hands of those with vast capital to invest in infrastructure. Despite infrastructure strains, Kimi K3 topped the rankings for front-end coding capability among major advanced AI models on Arena, an AI model evaluation platform, following its public release. This technical achievement fuels the competitive drive for market share and future profits.
Other Chinese tech giants are also aggressively expanding their AI portfolios. Over the weekend, Alibaba previewed its latest Qwen3.8 Max AI model, boasting 2.4 trillion parameters, which the company claims is one of the world’s most powerful models, second only to Anthropic’s Fable 5. About 1 month ago, Chinese startup Zhipu, or Z.ai, launched its GLM-5.2 model, which has seen rapid global adoption. This flurry of activity demonstrates the fierce global competition for control over the next generation of digital infrastructure and the immense profits it promises.