A recent analysis from Rhodium Group indicates that the combined annualized recurring revenue (ARR) of seven leading Chinese AI companies amounts to just 10% of what the two dominant US players generate. The report identifies DeepSeek, Moonshot AI, Zhipu AI, and MiniMax among the Chinese firms covered.

According to Rhodium's data, the total ARR for Chinese AI models between March and August 2026 reached approximately $10.7 billion. In stark contrast, OpenAI and Anthropic collectively reported disclosed revenues exceeding $100 billion.

Among Chinese enterprises, ByteDance is leading the pack, with its AI business generating an ARR of roughly $4 billion in July, while Alibaba recorded $2.4 billion in August. The remaining players reported significantly lower figures.

Despite the gap, Chinese companies are experiencing robust revenue growth. For instance, Z.ai's ARR surged from $74 million in January to $1.6 billion by August, marking a roughly 20-fold increase, while MiniMax grew about four-fold during the same period. However, revenue expansion alone does not yet translate into commensurate profitability.

Massive Infrastructure Spending

Rhodium estimates that Chinese firms will pour 932 billion yuan ($139 billion) into AI infrastructure in 2026, more than double the previous year's figure. By 2027, that number could exceed 1.2 trillion yuan ($193 billion). Meanwhile, US data center investments are projected at around $800 billion in 2026. This indicates that China is scaling up its infrastructure at a pace far outpacing the growth of AI model revenues.

The combined free cash flow of China's three major cloud providers—Alibaba, Tencent, and Baidu—turned negative in the first half of 2026, hitting minus 16 billion yuan. Just a year earlier, that figure stood at a positive 170 billion yuan.

Reliance on Equity Financing

In the US, major tech companies are increasingly leaning on debt financing: the net debt inflow for the five largest American cloud firms jumped from $90 billion in 2025 to $163 billion in the first half of 2026. Chinese enterprises, by contrast, depend more heavily on equity financing and bank loans.

Up to August 21, 2026, equity capital flowing into China's AI sector reached 282 billion yuan. Funding growth has been especially pronounced for leading labs such as Z.ai, MiniMax, DeepSeek, and Moonshot AI: they collectively raised about 9 billion yuan in private equity and venture capital in 2025, whereas IPO and private placement proceeds in the first eight months of 2026 totaled 179 billion yuan.

Roughly 25% of direct AI investment in China in 2026 comes from state funds and related entities. In the chip and server segment, state-owned or state-backed investors account for 47% of funding, with an additional 14% coming from banks, predominantly state-owned. In total, more than 60% of investment in this segment carries a state or quasi-state origin.

High Market Valuations

The report also highlights valuation discrepancies. Rhodium compared the market caps of these companies to their ARR, finding that several Chinese developers trade at higher multiples than OpenAI and Anthropic. The US pair boasts market-cap-to-ARR ratios of about 34x and 21x, respectively. Z.ai stands at roughly 46x, Moonshot AI at around 50x, and DeepSeek at a striking 163x.

Moonshot AI is particularly notable: its valuation ballooned from about $4 billion at the end of 2025 to $50 billion by August 2026. Still, profitability for Chinese developers trails their US counterparts. DeepSeek's company-level gross margin is estimated at 45% for July, compared to Anthropic's 65%. Z.ai and MiniMax recorded gross margins of roughly 26% and 18%, respectively, in the first half of 2026.

Rhodium attributes this partly to lower pricing in China. Even after recent hikes, most Chinese frontier models cost between $0.04 and $0.50 per task, whereas top-tier Claude models range from $2 to $4, and the priciest GPT models go for $1 to $2.

Funding as the Key to the AI Race

The report's authors contend that China's AI sector faces the same fundamental issue as the US: companies are expanding capital expenditures and computing capacity faster than they can generate sufficient cash flow. But China's financing model has its own particular characteristics.

To keep investing, firms must retain access to equity markets and bank credit. If valuations continue to rise faster than revenues, a downturn in the stock market could choke off their ability to secure new capital. Rhodium projects that Chinese AI investment will keep growing through 2026–2027, but identifies financing as one of the critical constraints on industry scale-up, alongside the supply of advanced chips.

Morgan Stanley estimates that monetizable revenue from China's consumer-grade AI could reach 294 billion yuan (about $44 billion) by 2030.