Analysis
DeepSeek, the Chinese AI lab whose low-cost models rattled US AI stocks in early 2025, has resumed its second funding round, targeting roughly 50 billion yuan (about $7 billion) at a pre-money valuation near 500 billion yuan (about $70 billion), with a signing expected in late August, according to The Information and Bloomberg. The company is separately preparing for a possible IPO as soon as this year.
In the same window, DeepSeek told developers to expect "significant" increases to its API pricing, without specifying exact numbers or an effective date -- the second pricing move in a month, after it introduced peak and off-peak pricing tiers in mid-July. DeepSeek currently charges $0.14 per million input tokens and $0.28 per million output tokens for its V4 Flash model, pricing that remains well below OpenAI's and Anthropic's comparable tiers even after the coming increase.
The two moves together read as a company transitioning out of a pure market-share land-grab phase: raising a large round while simultaneously signaling that the rock-bottom pricing that built its user base isn't sustainable at the volume DeepSeek is now serving. For US AI labs that have spent a year competing partly against the threat of Chinese price undercutting, a DeepSeek price hike -- even a modest one -- narrows that specific competitive pressure, though DeepSeek's post-hike pricing would likely still undercut every major US lab by a wide margin.