Analysis
DeepSeek's annualized revenue run rate has reached roughly $1 billion, more than double the level of a few months earlier, according to PYMNTS citing reporting from The Information. DeepSeek CEO Liang Wenfeng shared the figure directly with investors as the company finalizes its second major funding round.
What's New Since Pulse's Last Coverage
Pulse previously reported that DeepSeek had engaged CITIC Securities to prepare a Shanghai STAR Market listing. This update adds the concrete numbers that filing left open:
- Pre-IPO valuation -- ~$74.5B, reported before this round
- Annualized revenue -- $1B, DeepSeek's first disclosed figure at this scale
- Pre-IPO round size -- $7.5B (50B yuan), now finalizing
- Implied post-money -- ~$75B (500B yuan), consistent with the earlier estimate
Why Revenue Doubled: Pricing Power, Not Just Volume
The revenue jump followed API price increases of 2.3x to 4.5x depending on the model, and DeepSeek's own reporting to investors says demand held up despite the price hikes -- a meaningfully different signal than volume growth alone would send. Any AI lab can grow revenue by cutting prices and chasing volume; DeepSeek grew revenue by raising prices sharply and keeping most of its customer base, which is a much stronger data point about the actual value customers place on its models. DeepSeek's prices remain among the lowest of any major AI lab even after the increase, meaning this is a lab moving up from extremely cheap toward merely cheap, not repricing into premium territory.
The Numbers In Context
A targeted $75 billion valuation against roughly $1 billion in annualized revenue implies a revenue multiple near 75x -- a number that puts DeepSeek in the same speculative-growth territory as the fastest-growing US foundation-model labs, despite DeepSeek building its entire technology stack under US export controls that restrict its access to the most advanced Nvidia training chips. That combination -- premium-multiple valuation, discount-brand pricing, constrained chip access -- makes DeepSeek's underlying unit economics a genuinely unusual case worth more scrutiny than a comparably priced US lab with unrestricted chip access would warrant.
DeepSeek's most direct competitive comparison is Alibaba's Qwen family and other Chinese open-weight labs, all of which are also racing to capture enterprise API demand inside and outside China with aggressive pricing; internationally, DeepSeek competes against OpenAI, Anthropic and Google for the same enterprise API spend, and CNBC reporting earlier this year found the share of US enterprise AI tokens routed to Chinese models had risen roughly tenfold in eighteen months -- direct evidence DeepSeek's low-price positioning has been winning real workloads outside China, not just at home.
What to watch: whether DeepSeek discloses gross margin alongside future revenue updates, since margin is the number that will actually reveal whether this is a durable, profitable business at scale or a company still buying market share even after this month's price increases, and whether the Shanghai listing proceeds on the timeline CITIC Securities was originally engaged for.