Moonshot and DeepSeek head toward listings at 50 and 163 times revenue

Rhodium's six lines add to $10.5 billion, and OpenAI's $40 billion plus Anthropic's $65 billion is $105 billion, so the tenth holds to the decimal ^[1][2][3]. Two of those six lines belong to ByteDance and Alibaba, and between them they are $6.4 billion of the total, or 61 percent ^[7][12]. The four startups being priced off these numbers share $4.1 billion, which is 3.9 percent of the American pair ^[13][14].

Invert Rhodium's ratios and you have the prices behind them. Moonshot at 50 times $1 billion is a $50 billion mark; DeepSeek at 163 times $500 million is $81.5 billion ^[4][5]. The same method puts OpenAI at $1.36 trillion and Anthropic at $1.365 trillion, within $5 billion of each other ^[6]. Anthropic books 130 times DeepSeek's revenue and carries a multiple 7.8 times smaller ^[7][11]. Rhodium said valuations relative to revenue "appear exorbitant for Moonshot and DeepSeek at present" ^[9].

The denominator is the soft part on both sides. ARR as Rhodium describes it is a recent month multiplied by 12 ^[2], so one good month resets the annual figure for everyone in the table. Z.ai's own guidance shows how fast that moves: $3 billion by December against $1.8 billion now is $250 million a month against $150 million ^[6][16][8]. If it lands and nobody else changes, the Chinese total is $11.7 billion, or 11.1 percent ^[9].

Usage and revenue are wired together differently in the two markets. The Chinese models are open, so anyone with capable enough hardware downloads and runs them without the developer, and Rhodium said the labs are exploring ways to take a larger cut from the third parties offering access ^[17]. US models are mostly closed, and the cost per task for the leading OpenAI and Anthropic models is far higher than for the Chinese ones, according to Artificial Analysis ^[18].

Logan Wright, the Rhodium partner who wrote the report with research analyst Endeavour Tian ^[21], said "The financing gap means it will be far more difficult for Chinese frontier AI labs to scale sustainably" ^[19]. He said government funding "has been helpful on the hardware side of the buildout of compute capacity, but similarly will probably balk at direct funding for the frontier labs" ^[20]. Rhodium put state-affiliated sources above 60 percent of equity investment into Chinese AI chips and servers ^[22]. The state money bought machines. The labs raise from the equity market, and Z.ai's Hong Kong shares are back at levels seen this spring after briefly more than tripling over the summer, up more than 5 percent on Thursday morning after a decline earlier in the week that followed news of its second major fundraise in two months ^[23]. MiniMax has spent recent months struggling to hold above its IPO-day gains ^[24].

I'd expect the multiple to give before the revenue does, because a Hong Kong buyer can sell and the holders of a private round mostly cannot. The counter sits in the same numbers: 163 times is computed on $500 million, and an eightfold rise in that base takes the ratio to roughly 20 times with the price unchanged ^[10]. Rhodium could reference only the latest figures available from this summer, and usage of Chinese models has climbed steeply from low levels earlier in the year ^[15]. Anthropic is reportedly expected to list in the US next month, while OpenAI has pushed its own plans into next year ^[12]. US tech stocks tumbled after executives of leading American AI companies warned over the weekend about the risks of developing the technology too quickly, and leaders of the major Chinese labs have not commented ^[25].