Breaking

DeepSeek Raises $7.4 Billion, Fueling Low-Cost AI Push Against OpenAI and Anthropic Abroad

July 17, 2026 at 00:46 EDT

Funding & M&A
Foundation Models
Open Source

DeepSeek, the Hangzhou-based Chinese startup that shook the AI industry with its efficient open-weight models, has closed a roughly $7.4 billion first external funding round, giving it the war chest to undercut US rivals on price in overseas markets.

May 2026 · DeepSeek

China's DeepSeek Raises $7.4B to Undercut US AI on Price

The frugal open-weight startup closes its first external round at a ~$50–52B valuation — a war chest built to compete on cost in overseas markets, structured to keep founder Liang Wenfeng firmly in control.

$7.4B

First external round (~50B yuan)

~$52B

Post-money valuation — China's top AI startup

−75%

Reported price cut on V4 models

5-yr

Lock-up, no investor voting rights

The Cost Gap

Up to 52× cheaper than US incumbents

On high-volume, agentic workloads, developers report savings of 50× or more versus American providers.

1×

DeepSeek cost

up to 52×

US equivalent cost

Flagship API Pricing · per million tokens

DeepSeek-V4-Pro

1.6T MoE · 49B active · 1M-token context

$0.44

input

$0.87

output

DeepSeek-V4-Flash

lightweight, lowest cost tier

$0.14

input

$0.28

output

Why it lands

Open weights — self-host via Hugging Face & GitHub
Optimized to run on domestic Huawei chips
Huge budget relief for high-volume agentic use

The catch

US market restrictions blunt direct impact there
Cheap tokens ≠ better margins — agents burn volume
Competitive at benchmarks, not the absolute frontier

What's next

$7.4B round closed → ~$1.5B follow-on at higher valuation → Eventual public listing

With billions in the bank and prices set far below rivals, DeepSeek intends to compete on cost wherever it is allowed to sell.

Continue reading

The rest of this article is for AI News Blitz readers. Choose an option below to keep reading.