DeepSeek’s reported $1 billion run rate: the growth, price hikes and caveats

Reuters reported on September 24, 2026, that DeepSeek’s annualised revenue run rate had reached $1 billion. The figure, attributed to The Information and two people with direct knowledge, was more than double the reported pace of less than $500 million a few months earlier. A run rate projects a company’s current revenue pace across a year; it does not mean DeepSeek had already earned $1 billion over a full year. 13

What the $1 billion figure measures

A revenue run rate is an estimate based on how a business is performing at a particular time. It answers a conditional question: how much revenue would the company make over a year if it continued at that pace? Reuters included this explanation in its report. 1

That makes the $1 billion figure a snapshot extrapolated over a year—not an audited annual revenue total. Reuters said it could not immediately verify The Information’s report. 3

Growth and the role of higher model prices

The Information’s sources put DeepSeek’s run rate at less than $500 million a few months earlier, so the reported pace had more than doubled by September. The Information also said that raising model prices by 2.3 to 4.5 times partly drove the revenue growth. The report does not establish that price increases were the only factor. 13

Funding plans and a possible listing

Separate coverage said DeepSeek was pursuing a second funding round targeting 50 billion yuan at a 500 billion yuan valuation, with a goal of completing it by the end of October. It also described a possible listing in Shanghai. These were reported plans, not completed transactions; Reuters said it could not immediately verify the underlying report. 321

The available reporting supports describing a possible Shanghai listing, but does not establish that a Shanghai STAR Market IPO had been formally set or approved. 21

Computing resources and DeepSeek-V4.1-Flash

Reuters’ report said DeepSeek was allocating more than 70% of its computing capacity to developing new models. The available excerpt does not give a precise share for inference, the process of running models to generate outputs, so a specific training-versus-inference split cannot be confirmed from it. 1

Reuters’ DeepSeek news page said the company described DeepSeek-V4.1-Flash as the smallest model in its new architecture family. The available report excerpt does not specify the model’s aims or provide further detail about its intended uses. 2

What is established—and what is not

The central claim is a reported run rate: $1 billion annually at the current pace, more than twice the level reported a few months earlier. Higher model prices were said to have contributed. But the figure came from sources cited by The Information, and Reuters said it could not immediately verify the report. The run rate should therefore be understood as a reported estimate, not a confirmed annual revenue result. 13