#KimiConfidentiallyFilesForHKIPO

I think the “bubble” side is weighing more heavily right now – but it’s not an entirely empty story.

According to current reports, Moonshot AI made a discreet A1 investment in Kimi’s company HKEX and is advancing its final pre-IPO funding round at approximately $50 billion in upfront valuation. Kimi’s ARR rose from around $100 million+ in early March to approximately $300 million in mid-June; this is reportedly a several-fold increase in institutional ARR after K3.

What the numbers say:

$50 billion / $300 million ARR ≈ 167x ARR.

That’s incredibly high. The over $300 million figure is the ARR from mid-June, meaning K3 revenue hasn’t fully materialized yet. A current analysis also shows the same calculation as approximately a 167x revenue multiple.

Moonshot
2025 Final Valuation ~$4.3 Billion
Subsequent Valuations ~$30–35B
Targeted Upfront Payment ~$50 Billion
ARR ~$300 Million
Valuation / ARR ~167x
Valuation Increase ~11.6x

Supporting the Valuation Story

K3 isn't just a "good benchmark" story.

* Some sources report that K3 is in high demand as a weighted model with 2.8 sports parameters.

* Moonshot is negotiating revenue-sharing agreements to offer K3 on Microsoft, Amazon, and Google's cloud platforms. This could be significant in terms of realization and monetization.

* The fact that ARR increased from $100M to $300M in a few months shows that there is real commercial competition.

* The appetite for AI/robotics IPOs in Hong Kong is also quite strong right now.

But why is $50 billion dangerous?

The real problem is the gap between the valuation and the actual valuation.

A valuation of $50 billion based on a $300 million ARR means the investor believes:

“We’re not investing in today’s Moonshot, but in several massive Moonshots in the coming year.”

However, this is only achievable if the ARR continues very rapidly, generates high gross profit margins, and Kimi gains a lasting advantage over its competitors.

In general, the model advantage offered by the AI ​​model is not guaranteed to be permanent. In China, competitors like DeepSeek, Alibaba/Qwen, and Zhipu are also progressing very aggressively. Therefore, K3’s technical capacity alone does not justify the $50 billion valuation.

Another venture capital intensity: running frontier models with 2.8T parameters creates significant computational costs. This is one of the main reasons Moonshot is in revenue-sharing negotiations with large cloud providers.

My framework:

$50 billion = not “value,” but the price of a “future success option.”

Roughly speaking:

* <$10-15 billion: Quite interesting in my opinion.
* $15-25 billion: Defensive if growth continues.
* $25-35 billion: Requires very strong K3 monetization.
* ~$50 billion: Too much futures are priced in.

Therefore, if the IPO actually reaches around $50 billion, my first impression would be "quality company, valuation like a bubble."

However, the critical point is this: an IPO above $50 billion is not the same as private financing above $50 billion. The IPO value will be much more decoupled from the market after the IPO and the company's final disclosure data for H1 2026, including gross margin, net loss, R&D/accounting expenses, and corporate ARR.

Moonshot AI = real growth + real technology + currently overly aggressive valuation. If I were you, I wouldn't be following the K3 hype, but rather the fact that ARR didn't enter the $300M → $1B+ entry phase. What would make $50B reasonable isn't model benchmarking, but repeatable and high-margin revenue.