DeepSeek

Artificial Intelligence · Large Language Models · China Tech

Funding

DeepSeek reportedly hits $1B revenue run rate, nears $7.5B raise

A 2.3-4.5x price hike didn't scare off developers, proving efficient Chinese models can still command premium pricing without losing share.

• DeepSeek's annualized revenue run rate has hit $1 billion, more than double the under-$500 million pace of a few months ago, CEO Liang Wenfeng told investors as the company finalizes a roughly $7.45 billion second funding round.
• Nearly all of that revenue comes from developers paying for API access to DeepSeek's models; the free consumer chatbot app generates no revenue at all.
• Growth was fueled largely by a recent price hike of 2.3 to 4.5 times across models, which Liang said did not shrink the customer base even as prices remain among the cheapest for major AI models.
• DeepSeek's API business posted an 82.9% gross margin over the first seven months of the year, reportedly outpacing margins Anthropic and OpenAI posted in comparable periods.
• DeepSeek still directs more than 70% of its compute to training new models versus under 30% to inference, and Liang expects to bring in Huawei-made training chips as early as Q4 2026.
• Rivals are scaling just as fast: Chinese lab Z.ai reportedly reached $1.8 billion in ARR, while MiniMax's run rate jumped from $150 million in February to $800 million by August.
• DeepSeek aims to close the round by the end of October and has hired CITIC Securities to prepare a listing on Shanghai's STAR Market, though IPO timing and terms remain undecided.
• Crossing $1 billion in run-rate revenue shows a Chinese frontier lab converting cheap, efficient models into real profitability, not just cost disruption, sharpening the competitive stakes against OpenAI and Anthropic.