DeepSeek Rumored to Start New Funding Round

量子位 · wechat · 2026-07-15

According to the Financial Times and others, DeepSeek is meeting new investors to launch its second funding round. The pre-money valuation is around $71 billion, up roughly 37% from the post-money valuation of $52 billion in the first round. The company could file for an IPO as early as the end of this year, targeting a 2027 IPO.

Why This Round is Happening

The author believes the core driver is compute expansion:

Agents require continuous inference, consuming more compute;
Long contexts demand larger memory and infrastructure;
DeepSeek aims to lock in computing resources for the next few years.

Where the Funds Will Go

Building proprietary data centers;
Purchasing more AI chips;
R&D for next-gen AI systems like Agents.

Hiring and Infrastructure Signals

The article notes DeepSeek is aggressively hiring, including:

IDC design engineers for GW-level data center site selection, layout, and construction;
Training/inference framework engineers;
High-performance operator and communication compiler engineers;
Supercomputing cluster R&D engineers.

Additionally, Reuters reports DeepSeek is secretly developing its own AI inference chips, having contacted chip designers, foundries, and memory suppliers.

Extra Info

The first round reportedly raised over 50 billion RMB;
Liang Wenfeng personally invested about 20 billion RMB;
The Bloomberg Billionaires Index estimates his net worth has risen to $36 billion.

Original post →

DeepSeek Rumored to Start New Funding Round

More from Venture