DeepSeek Annualised Revenue Run Rate Doubles to $1 Billion Ahead of IPO

DeepSeek Annualised Revenue Run Rate Doubles to $1 Billion

Chinese AI startup DeepSeek's annualised revenue run rate has reached $1 billion, more than doubling from under $500 million just a few months ago, according to 1

1

1

1

. The report cites two people with direct knowledge of the matter, marking a significant milestone for the company that disrupted assumptions about Chinese AI capabilities last year. CEO Liang Wenfeng shared these revenue figures during a recent meeting with investors as DeepSeek prepares for its IPO on Shanghai Stock Exchange2

2

. A run rate projects annual income by assuming the company maintains its current earning pace, showing momentum rather than actual banked revenue.

API Price Increase Drives Revenue Growth

The dramatic revenue surge stems partly from DeepSeek's API price increase implemented in August. The company raised fees between 2.3 to 4.5 times depending on the model, yet Liang Wenfeng told investors that customer demand remained unaffected by the price hike3

3

. Despite these increases, DeepSeek's prices remain among the lowest for major AI models in the market. Most of DeepSeek's revenue comes from developers who pay to access the company's AI models through its API, while the chatbot remains free for users1

1

. The Information reported last month that DeepSeek earned approximately 475 million yuan ($70.7 million) in the first seven months of 2026, roughly ten times its revenue for all of 20251

1

.

Funding Round and IPO Preparations Accelerate

DeepSeek is finalizing its second funding round, aiming to raise 50 billion yuan ($7.5 billion) at a valuation of 500 billion yuan ($74 billion) by October-end2

2

. The company has hired CITIC Securities to prepare for a potential STAR Market IPO, though specific timing remains undecided5

5

. This funding round has not proceeded smoothly—DeepSeek paused it in July after leaked remarks by Liang Wenfeng went viral on Chinese social media, then restarted negotiations in August1

1

. The lab only began accepting outside investment this year, closing its first round of roughly $7 billion in June, with Liang himself providing a substantial portion of the capital1

1

.

Computing Capacity Focused on Model Development

Despite its commercial success, DeepSeek continues prioritizing model development over immediate profitability. Liang Wenfeng revealed to investors that the company allocates more than 70% of its computing capacity to training new AI models while reserving less than 30% for inference—running existing models4

4

. This strategy underscores DeepSeek's commitment to innovation and long-term competitiveness in the rapidly evolving AI landscape. Earlier this month, the company released DeepSeek-V4.1-Flash, its latest model designed for greater capability, faster inference speed, higher throughput, and scalability to larger models5

5

.

Cost-Efficiency Challenges Silicon Valley Assumptions

DeepSeek first gained prominence in early 2025 when it launched an AI model that shook Silicon Valley and Wall Street by delivering performance comparable to U.S. rivals while using far fewer Nvidia chips3

3

. This cost-efficiency advantage, combined with China's lower energy costs, has created an opening for Chinese AI labs to undercut American competitors on pricing. Companies worldwide are scrutinizing their AI spending as costs rise, making DeepSeek's affordable yet powerful models increasingly attractive. The company's ability to maintain lower prices even after significant increases positions it strategically as organizations seek to manage AI expenditures more effectively. Watch for how DeepSeek's IPO performance influences investor confidence in Chinese AI startups and whether its cost-efficiency model pressures U.S. companies to optimize their own infrastructure spending.