DeepSeek (深度求索) has reportedly restarted its second funding round, seeking CNY 50 billion at a pre-money valuation of around CNY 500 billion, as the Chinese AI company simultaneously strengthens its position in the global large-language-model market. The financing is expected to be signed in late August, according to Chinese financial media reports.
The new round would mark another major capital injection for one of China's most prominent foundation-model developers. DeepSeek completed its first external financing earlier this year, raising CNY 50 billion at a valuation of more than CNY 350 billion. If the latest round is completed as planned, the company will have raised more than CNY 100 billion across two rounds, giving it substantial resources for model training, computing infrastructure and global developer expansion.
The financing comes as DeepSeek's latest model gains traction among developers worldwide. DeepSeek V4-Flash ranked first by token usage on OpenRouter, a global platform that aggregates access to models from multiple AI providers, during the week of July 27 to August 2. The model's strong adoption highlights how Chinese foundation models are increasingly competing not only on benchmark performance but also on price, inference efficiency and real-world developer usage.
DeepSeek has built much of its international momentum around this cost-performance strategy. Rather than relying solely on ever-larger models and computing budgets, the company has focused on improving model efficiency and lowering inference costs. V4-Flash continues that approach, combining stronger agent and coding capabilities with aggressive API pricing, helping DeepSeek attract developers outside China and compete directly with models from OpenAI, Anthropic and Google.
The combination of large-scale financing and rising global usage points to a new stage in China's AI competition. Capital is increasingly being converted into computing capacity and faster model iteration, while lower inference costs help generate more developer adoption and token consumption. For DeepSeek, the potential CNY 50 billion round would therefore be more than a financing event: it could reinforce a cycle of capital, compute, model improvement and global usage, as Chinese AI companies seek a larger role in the international foundation-model ecosystem.