DeepSeek's annualized revenue doubles to $1 billion in a few months, with a second round of $50 billion financing expected to be completed by the end of October.

动察 Beating AI News Flash: The Information, citing two people with direct knowledge, reported that DeepSeek's annualized revenue has reached approximately $1 billion. Just a few months ago, that figure was less than $500 million, effectively doubling in a short period. Revenue mainly comes from model APIs, while the free chat app still generates no advertising or subscription revenue.

Part of this growth came from price increases. DeepSeek previously raised some API prices substantially, with increases of about 2.3x to 4.5x depending on the model and time period. Liang Wenfeng recently told investors that customer numbers did not decline after the price hikes and demand remains very strong. Previously disclosed financial data showed that DeepSeek's API business gross margin reached 82.9% in the first seven months of this year.

DeepSeek is also advancing a second funding round, planning to raise about RMB 50 billion at a valuation of about RMB 500 billion. This plan had already been reported, and the latest development is that the company hopes to complete the financing by the end of October. DeepSeek is also preparing for an IPO on the Shanghai Stock Exchange's STAR Market.

However, the company still allocates most of its computing power to training. Liang Wenfeng said that more than 70% of compute is used to train new models, leaving less than 30% for inference. DeepSeek is trying to get more small models to run directly on gaming GPUs, continuing to reserve high-end chips for training.