Under DeepSeek’s Impact, Moonshot AI Significantly Reduces Marketing Budget

According to the Chinese media outlet Jiemian, Chinese AI startup Moonshot AI has recently decided to significantly reduce its marketing budget, including pausing placements on multiple Android channels and ending partnerships with third-party advertising platforms.

Internally, Moonshot AI explained this adjustment as being influenced by “external factors and internal strategic changes.” As of the time of writing, the company has not provided any comments to media.

Kimi was the most talked-about AI product from a domestic startup before the rise of DeepSeek. Apart from its user experience, the external discussions surrounding Kimi mainly focused on its marketing and customer acquisition strategies. After exploring business ventures such as overseas consumer (user-end) products, Moonshot AI proactively chose to streamline its operations and focus on Kimi. In an interview last November, Moonshot AI’s founder and CEO, Yang Zhilin, emphasized that improving user retention was currently Kimi’s core objective.

However, faced with the reality of DeepSeek reaching 35 million DAUs (Daily Active Users) within a few weeks through organic traffic, Kimi, despite significant expenditure over more than a year, only achieved a daily active user count in the low tens of millions, indicating a challenging road ahead.

Nevertheless, this adjustment should not be solely interpreted as Moonshot AI’s negative response to the current situation. Fundamentally, the company is addressing the previously criticized issue of high customer acquisition costs.

In fact, DeepSeek’s most notable and acclaimed aspect this time is the reasoning model R1, significantly reducing inference costs compared to the OpenAI o1 model. Moonshot AI also made the right bet on the technological path, being among the first to introduce o1-class models within the “Six Tigers” and even launching the latest multimodal reasoning model k1.5 on the same day as R1. However, these models did not generate significant industry buzz.

A source familiar with the matter revealed that DeepSeek’s sudden success was a significant motivation for Moonshot AI internally. After reviewing the situation, the team believes in sticking to the fundamental SOTA (State-of-the-art) models. Moving forward, the team may retrain the basic models while seizing opportunities at the application level. One selected product direction internally is DeepResearch, an intelligent entity product aimed at the deep research field released by OpenAI in February.

According to public data, Moonshot AI has completed four rounds of financing, with a total funding amount exceeding 11 billion RMB. Reports suggest that nearly 200 million USD was converted into computing power vouchers in a funding round exceeding 1 billion USD with Alibaba‘s participation. This indicates that Moonshot AI’s cash reserves are close to 10 billion RMB.

SEE ALSO: Chinese City Shenzhen Introduced “AI Civil Servants”