China's Low-Cost AI Models Gain Edge in Price War
Left 100%
Center 0%
Right 0%
2 left · 0 center · 0 right
What happened
Research firm Rhodium Group published estimates on Thursday that all Chinese AI models combined generate about 10 percent of the annual recurring revenue reported for OpenAI and Anthropic, using an industry metric that annualizes a recent monthly figure. Rhodium cited roughly $40 billion for OpenAI and $65 billion for Anthropic, versus far smaller figures it listed for Chinese firms including ByteDance, Alibaba, Z.ai, Moonshot, MiniMax, and DeepSeek. The same report said valuations relative to revenue look high for Moonshot and DeepSeek compared with the U.S. firms. Z.ai told investors its latest ARR was $1.8 billion and raised its year-end ARR forecast to $3 billion. Rhodium partner Logan Wright said a financing gap would make it harder for Chinese frontier labs to scale sustainably without a favorable equity market.
Sourcing check · Mostly on the record · 3 articles
1 of 2 articles cite a named source for the story. It was on the record from the first report.
What these terms mean and how they're counted
- On the record · names a source 1 of 2
- On the record means the article cites a source by name for the story's main claim: a person, or an institution speaking officially, such as a company statement or a named agency's spokesperson. The opposite is an unnamed source. The count is how many of the counted articles cite at least one such source; the names are everyone named across them, each once however the articles spell it. Here: Rhodium Group and Z.ai
- Primary evidence · cites a document or data 0 of 2
- Primary evidence means the article cites a record you could check yourself: a court filing or ruling, a dataset or report, a recording or transcript, an official statement or press release, market data. A quote is not a document. The count is how many articles cite one. Here: None cited.
- Credits another outlet 0 of 2
- The article says the story comes from another outlet's reporting (“first reported by Reuters”). Those articles relay the story; they are not independent confirmation. Here: Each article presents its own reporting.
- Unnamed sources 0 of 2
- The main claim is attributed to people who are not identified: “people familiar with the matter”, “officials said”. The count is how many articles do this.
- How it developed
- When the first report landed and whether it named anyone; when the first named source and the first document appeared. Here: First report Sep 17, 09:00 UTC by CNBC, naming its source.
- Not counted 1
- Opinion, analysis and video pieces, and articles too short to assess. Every count above is out of the counted articles. An article republished under a new headline is one article.
This is about who is cited, not whether the story is true. Each source below says what it names. How we count
ONE-SIDED THIS WEEK.
Only left-leaning outlets have covered this story in the last two days;
New York Post reported it on 2026-09-11.
Omitted — what each side leaves out
Unpacked
Right-leaning outlets had not covered this as of publication, so a reader there would not learn that U.S. AI labs still collect far more model revenue than every Chinese model combined. Left-leaning coverage reports that China's models together make only about 10% of OpenAI and Anthropic's take, according to CNBC's account of a Rhodium Group study. That coverage also records Chinese adoption rising fast without matching sales. Without that comparison, China's cheap models can be read as a commercial success rather than a usage wave that has not yet paid.
The emphasis among left-leaning outlets pulls apart. Valuations of Chinese startups take the lead in CNBC, which calls them "exorbitant" against current sales, while Bloomberg leads with budget open-weight models giving China an edge.
If Chinese models are cheap and widely adopted, what share of global AI tasks do they actually run?
Bottom line
OpenAI alone is booked at $40 billion in CNBC's Rhodium figures, more than all Chinese AI models combined. That comparison has not appeared in right-leaning coverage.
The receipts — 5 facts one side's coverage lacks
- All of China's AI models combined generate only about 10% of the revenue reported for OpenAI and Anthropic
- DeepSeek's estimated valuation-to-revenue ratio is 163x
- More than 60% of equity investment in Chinese AI chips and servers came from state-affiliated sources
- The cost per task for leading OpenAI and Anthropic models is far higher than that of Chinese models
- Bank of America said the strongest long-term economics are likely to remain with AI chips, semiconductor equipment, memory, and large cloud platforms rather than standalone AI model developers
The Left View
CNBC, citing Rhodium, frames the gap as adoption that has not yet become revenue, with Chinese ARR totals around one-tenth of OpenAI plus Anthropic, and valuation-to-revenue ratios it calls exorbitant for Moonshot and DeepSeek versus lower multiples for the U.S. labs. It also reports Z.ai’s investor ARR figures, notes U.S. models are mostly closed and cost more per task, and quotes Wright on dependence on China’s equity market and limited direct government funding for frontier labs. Bloomberg frames China’s low-cost, open-weight models as gaining ground and asks whether Beijing can convert adoption into global AI leadership. Business Insider, citing Bank of America, frames a new phase of China’s AI price war: competition shifting from blanket token cuts toward cost per completed task, with basic inference commoditized and longer-term profits more likely in chips, equipment, memory, and cloud platforms than in standalone model developers.
Our Take (balanced)
The story is substantive: it rests on a named research group’s published estimates, named companies, specific ARR and multiple figures, and an on-the-record Z.ai forecast—not on a single anonymous claim. Established is that Rhodium issued those comparisons and that Z.ai stated $1.8 billion ARR and a $3 billion year-end target; the cross-company revenue totals and 50x/163x versus 34x/21x multiples remain Rhodium estimates based on annualized monthly metrics, not audited public filings. No right-leaning outlet has reported this. Watch for the full Rhodium report, listing or IPO filings from the named firms, and any restated ARR or audited revenue that can confirm or revise the estimates.
3 sources
More stories
The week's bottom lines, in your inbox
One email a week: the five stories that mattered and what they actually mean. Free.