Kimi: Due to a shortage of computing power, subscriptions for new users on the C-end will be suspended from today, and all existing computing power will be allocated to serving already subscribed users.

```

Products under Moonshot AI's portfolio, Kimi, have announced the suspension of subscription services for new C-end users, concentrating all existing computing power to ensure the experience of subscribed users. This decision reflects the explosive growth in demand following the release of the new model, and puts the supply-demand conflict for AI computing power under the spotlight.

In its announcement on July 19, Kimi stated that since the release of Kimi K3, user request volume in the past 48 hours has significantly exceeded estimates and is approaching the carrying limit of current clusters. The company decided to suspend new C-end user subscriptions effective immediately, while rapidly expanding computing resources. Once new resources come online, subscription slots will be gradually opened until normal subscriptions are fully resumed.

The immediate reason for suspending subscriptions is the surge in user demand triggered by Kimi K3’s strong performance in various public benchmark tests. Previously, Goldman Sachs partner and head of equities Rich Privorotsky described the release of Kimi K3 as a deeper industry signal, warning that the "era of compute expansion" may have reached its end, and that the AI valuation logic is facing remodeling.

Kimi K3 Spurs Demand, Compute Pressure Triggers Subscription Circuit Breaker

Kimi K3 topped the Arena comprehensive text ranking with 1679 points in the frontend code arena, surpassing Claude Fable 5, and achieved a leap from 18th to 1st place compared to the previous generation. In the frontend field, it ranked first in six out of seven subcategories including branding and marketing, reference design, data and analytics.

Moonshot AI also claims that Kimi K3 outperformed OpenAI and Anthropic flagship products in several public programming and agent benchmarks, with costs about 40% lower than comparable top US models. This combination—top performance paired with low cost—directly drove an influx of users far beyond expectations, ultimately triggering a proactive circuit breaker for subscription access.

Kimi stated in the announcement that it "sincerely apologizes" to users who did not get the expected experience, and promised that the rights of all subscribed users will not be affected. The company also announced that when subscriptions resume, it will split Kimi main entitlements (including Kimi Web, Kimi APP, Kimi Work) from Kimi Code entitlements to more precisely match computing resources and ensure user experience.

Goldman Sachs: Expansion Is No Longer the Only Winning Path

From Goldman Sachs's perspective, the significance of the Kimi K3 event goes far beyond a single product's market performance.

Reportedly, Goldman Sachs partner Rich Privorotsky pointed out that the truly thought-provoking issue is: how a Chinese lab unable to match the largest Western pre-training compute scales quickly narrowed the gap with top US models through architectural innovation, synthetic data, reinforcement learning, and post-training techniques. He clarified that this does not prove the "expansion law" has failed, but rather that "expansion is no longer the only winning path."

This judgment directly targets the core pillar of the current AI investment narrative—namely, the logic that computing power scale determines the upper limit of model capabilities. If this logic is shaken, investment expectations for building large-scale computing infrastructure will face repricing. AI analyst Kim Isenberg commented, "The whole rulebook has changed, and this will trigger 'red alerts' at certain institutions."

Low-Cost Open Source Strategy Challenges Compute Investment Assumptions

Moonshot AI plans to release Kimi K3 with open weights on July 27, allowing enterprises and governments to deploy customized systems. This open source strategy is very similar to DeepSeek's previous approach, again raising concerns in the market about a repeat of the "DeepSeek moment."

The combination of low cost, high performance, and open source poses a direct challenge to current AI computing power pricing authority. In the previous DeepSeek surge, the market had already undergone a round of questioning the necessity of large-scale compute investments.

For investors, Goldman Sachs's warning provides an important analytical framework: compute expansion itself is no longer a sufficient condition for AI competition; innovation in architectural efficiency and training methods is becoming the new variable. This logic shift will have a profound impact on the valuation system of the AI industry chain upstream and downstream.

Risk Warning and DisclaimerThe market has risks; investments should be made cautiously. This article does not constitute personal investment advice and does not take into account the particular investment goals, financial situation, or needs of individual users. Users should consider whether any opinions, views, or conclusions in this article fit their specific situation. Investing based on this is at your own risk. ```