Kimi Suspends C-end Membership Sales Due to Power Shortage, Highlighting GPU Shortage Issues
Kimi announced the suspension of C-end membership sales due to tight computing power.
This move further confirms the current supply shortage of high-end GPUs and other computing resources.
Market mechanisms show that the explosive demand exacerbates the computing power bottleneck, with capital accelerating towards GPU supply chains, data center expansions, and inference optimization companies. Beneficiaries are AI platforms with stable computing power reserves or self-built clusters, while those under pressure are small and medium model service providers and rapidly growing C-end applications.
Source: Public Information
ABAB AI Insight
After the previous K3 release, Kimi had already suspended new user subscriptions due to demand exceeding expectations. This suspension of C-end membership continues to reflect the reality that computing power has become a major bottleneck, similar to the capacity management strategies of cutting-edge labs like OpenAI after the release of popular models.
In terms of capital pathways, the company prioritizes ensuring existing loads and urgently expanding capacity, shifting resources from user growth to infrastructure construction, motivated by the need to maintain service stability and long-term competitiveness. Chinese AI companies are intensively financing to layout computing power.
Similar to the recurring computing power shortages faced by global AI labs, the generative AI industry is currently in a structural contradiction phase where model capabilities are rapidly advancing while infrastructure lags severely.
This essentially belongs to the restructuring of the industry chain: the computing power shortage forces platforms to temporarily shrink C-end operations, with the mechanism being that the exponential growth in training and inference demand outpaces supply speed, leading capital to accelerate towards GPUs, energy, and data centers, thereby reshaping AI service pricing power and infrastructure barriers.
ABAB News · Cognitive Law
The stronger the model, the more severe the computing power shortage.
Suspending sales to maintain experience, will resume volume after expansion is completed.
Whoever solves the GPU issue first will hold the pricing power in the next phase.