The company generated roughly 475 million yuan (about $70.7 million) in revenue in the first seven months of this year, or about 10 times its revenue for all of last year, according to the report.
The company recorded a net loss of about 715 million yuan (about $106 million) between January and July, versus 935 million yuan for the whole of 2025 (about $139 million), the report said. DeepSeek is in talks with investors about a funding round that aims to raise 50 billion yuan (about $7.4 billion) at a valuation of 500 billion yuan (about $74.4 billion).
We’d love to be your preferred source for news.
Please add us to your preferred sources list so our news, data and interviews show up in your feed. Thanks!
While DeepSeek’s revenue is dwarfed by that of companies like OpenAI, it is still reaching a healthy gross profit margin, according to the report. This is due to its ability to keep the costs of running its models low improving the efficiency of its artificial intelligence infrastructure so that the AI systems can carry out tasks using fewer chips.
DeepSeek’s gross margin for the first seven months of 2026 came to 44.6%, and the gross margin for selling access to its models through APIs was 82.9%, the report said.
OpenAI, meanwhile, recorded $5.7 billion in revenue and a 39% gross margin during the first quarter, while Anthropic generated $11.5 billion in revenue in the second quarter and forecast that its gross margin in 2026 would rise from 40% last year to 63%, according to the report.
DeepSeek’s expenses are quickly growing as well, although still at a fraction of what’s happening with AI labs in the United States, the report said. DeepSeek spent about 11 billion yuan (about $1.6 billion) on AI infrastructure-related expenses during the first seven months of the year, which was almost another tenfold increase over 2025, when the company spent 1.2 billion yuan (About $178 million).
DeepSeek first came to prominence last year when it launched an AI model that shook Silicon Valley and Wall Street. That model offered performance comparable to that of U.S. rivals while using far fewer Nvidia chips.
This month brought news that the company is building a team to rival Anthropic’s Claude Code in the market for AI agents that automate tasks for business professionals.
For all PYMNTS AI coverage, subscribe to the daily AI Newsletter.
Add as Preferred Source
