DeepSeek has finally put "making money" on the table. On September 24, The Information reported that DeepSeek's annualized revenue run rate has reached $1 billion, more than doubling in just a few months. The news comes from two people familiar with the matter: CEO Liang Wenfeng disclosed the figure at a recent investor meeting.
The doubling engine: a 2.3x to 4.5x price hike
The direct engine of the doubling is price hikes. This year DeepSeek raised model prices by roughly 2.3 to 4.5 times, and combined with the models' sustained popularity, revenue growth beat most expectations. A caveat: run rate is an annualized projection at the current pace — it doesn't mean $1 billion was actually collected over the past year. But it does mean this company that rose to fame on "being cheap" has, for the first time, cash flow of a serious magnitude.
A $7.5 billion raise in talks: the valuation logic of the open-source route rewritten
The same sources also say DeepSeek is negotiating a new funding round that could reach $7.5 billion. If it closes, it would be one of the largest financings in Chinese AI startup history — and it would mean the market's valuation logic for the "open-source plus low price" route is being rewritten: cheap no longer means unprofitable.
Going against the grain in the price war
The timing is intriguing. Just as everyone is cutting flagship model prices and the industry price war rages, DeepSeek is going the other way — raising prices first, then handing in a $1 billion report card. It shows that at least in inference-hungry market segments, customers are willing to pay more than before for "cheap but good"; price wars were never just about the numbers, but about the cost-performance of "unit intelligence". Earlier benchmarks showed DeepSeek's coding-task costs at about one-fifteenth of rivals' — its room to raise prices comes precisely from that cost advantage.
For the open-source ecosystem, this is a positive signal: the V4 family proves open models can stand on their own commercially. The suspense is left for the next round — once the $7.5 billion lands, whether DeepSeek pours it into compute, into AGI, or into the next price war, the market will soon see the answer.
