DeepSeek levantou 7,35 bilhões de dólares na primeira rodada de financiamento, deixando a Alibaba para trás! Apenas o dinheiro com as "menores condições"

China AI startup DeepSeek launches its first external funding round, aiming to raise up to 7.35 billion USD (500 billion RMB), with post-investment valuation expected to exceed 51.5 billion USD (3.5 trillion RMB), becoming the largest single funding in China’s AI history. Alibaba’s negotiation to connect its ecosystem failed, Tencent’s proposed 20% subscription was also rejected, founder Liang Wenfeng plans to personally invest up to 20 billion RMB, acquiring about 40% of the current round.
(Background: Huang Renxun: Blackwell and Rubin should not be allowed to acquire China, NVIDIA China market share remains zero and they refuse to back down)
(Additional context: Anthropic spent 200 billion USD over five years on Google Cloud, two AI startups take half of the orders from the four major cloud giants)

Table of Contents

Toggle

• Key Summary
• Alibaba rejected, Tencent cut, national team expected to lead investment
• V4.1 launched in June, core researchers already poached
• Pure research mode has reached its limit
• Frequently Asked Questions

Key Summary

• DeepSeek’s first external funding round aims to raise up to 7.35 billion USD (500 billion RMB), founder Liang Wenfeng personally invests 2 billion RMB, holding 40%
• Alibaba’s negotiation failed due to demands to tie ecosystem; China National Big Fund expected to lead with the loosest terms
• V4.1 scheduled for June release, native support for MCP, but core researchers have left for Xiaomi and ByteDance

DeepSeek, which has not sought external funding for three years, officially opens its doors. According to The Information, this China AI startup incubated by quant fund Fantom Quant is launching its first external funding round, aiming to raise up to 50 billion RMB (about 7.35 billion USD), with a post-investment valuation expected to surpass 35 billion RMB (about 5.15 billion USD). If completed, it will be the largest single funding in China’s AI industry history.

But even with the opening, founder Liang Wenfeng’s goal is “the least conditional money.”

Alibaba rejected, Tencent cut, national team expected to lead investment

The largest investor in this round is not an external institution but founder Liang Wenfeng himself. He plans to personally invest up to 20 billion RMB, acquiring about 40% of the shares. China’s National Integrated Circuit Industry Investment Fund (under the third phase of the Big Fund) is expected to become the second-largest investor, with relatively relaxed terms.

Alibaba and Tencent both actively negotiated but hit a wall. Alibaba hoped to strengthen its AI ecosystem through investment, deeply integrating business units like Tongyi Qianwen into products such as Taobao and Gaode. DeepSeek insists on technological independence and does not accept ecosystem tying, resulting in no agreement.

Tencent proposed subscribing for up to 20% of shares, but was rejected due to the large share transfer.

Liang Wenfeng has long refused external equity financing. Previously, DeepSeek was fully supported by Fantom, with Liang holding about 84% directly and indirectly. Although this is the first time opening the door, the bottom line is “the offer with the fewest conditions.” The conditions from the Big Fund are the loosest and most aligned with DeepSeek’s approach.

V4.1 launched in June, core researchers already poached

The pace of commercialization driven by funding is accelerating. DeepSeek plans to release V4.1 in June, focusing on enterprise tools, native support for MCP (Model Context Protocol), and adding image and audio processing capabilities (generation still limited to text). This is DeepSeek’s first clear move to target the enterprise market with tool-level products.

But talent loss is more urgent than the schedule. Core contributor Luo Fuli left for Xiaomi in November last year, taking over the MiMo large model team. In April this year, the MiMo-V2.5 released by the team surpassed DeepSeek-V4-Flash in multiple evaluations. Another core researcher, Guo Daya, joined ByteDance’s Seed team in April, responsible for Agent (L8 level), with rumors of a salary “close to 100 million RMB” (ByteDance denies specific figures). Reports indicate ByteDance has poached nearly 70 people from DeepSeek over the past year.

Competitors offer salaries more than twice that of DeepSeek. A large part of the new funding will be used to increase employee compensation and issue stock options, establishing market-based valuation for equity, making paper wealth more liquid.

Pure research mode has reached its limit

In February this year, DeepSeek publicly stated it was “completely funded by Fantom, with no immediate plans for fundraising.” Three months later, soaring computing costs and fierce talent wars forced a change in stance.

Compared to US model companies, Anthropic and OpenAI’s cloud computing contracts total nearly 2 trillion USD. DeepSeek’s reputation for training cutting-edge models at very low cost is well-known, but maintaining technological leadership requires not just efficiency but also absolute scale of computing power and competitive salaries to retain talent.

Embracing national team funding, launching enterprise tools, and pricing employee equity, DeepSeek faces self-sustainability and enters larger global competition.

Frequently Asked Questions

How large is this funding round for DeepSeek?

Up to 50 billion RMB (about 7.35 billion USD), with a post-investment valuation exceeding 35 billion RMB (about 5.15 billion USD), making it the largest single funding in China’s AI history. Founder Liang Wenfeng personally invests about 40%.

Why couldn’t Alibaba invest in DeepSeek?

Alibaba wanted to tie DeepSeek’s technology into its product ecosystem (Taobao, Gaode, etc.), but DeepSeek insisted on technological independence and minimal conditions, leading to failed negotiations. The China National Big Fund is expected to lead with the loosest terms.