Zhipu AI Raises $5B to Boost Self-Training and Compute: what are the key facts?

Zhipu AI has completed a funding round of approximately $5 billion to develop next-generation GLM models and expand computing capacity.

Zhipu AI Raises $5B to Boost Self-Training and Compute: what are the key facts?

The company reported first-half 2026 revenue of 954 million yuan, representing a year-over-year increase of nearly 400%, with API services accounting for over 86% of total income.

Zhipu AI Raises $5B to Boost Self-Training and Compute: what are the key facts?

Despite Zhipu proactively raising API prices by 101%, token usage volume surged more than 40 times compared to the beginning of the year.

Zhipu AI Raises $5B to Boost Self-Training and Compute: what are the key facts?

The firm plans to allocate 60% of the funds to research a fully self-training system, aiming to eliminate reliance on manual data annotation.

Zhipu AI Raises $5B to Boost Self-Training and Compute: what are the key facts?

Zhipu operates a 1-gigawatt data center powered entirely by domestic AI chips to support model development.

Physix Frontier · News Briefing Card (enbrief · Sep 18, 2026)

Zhipu AI Raises $5B to Boost Self-Training and Compute

KEY FACTS

Zhipu AI has completed a funding round of approximately $5 billion to develop next-generation GLM models and expand computing capacity.
The company reported first-half 2026 revenue of 954 million yuan, representing a year-over-year increase of nearly 400%, with API services accounting for over 86% of total income.
Despite Zhipu proactively raising API prices by 101%, token usage volume surged more than 40 times compared to the beginning of the year.
The firm plans to allocate 60% of the funds to research a fully self-training system, aiming to eliminate reliance on manual data annotation.
Zhipu operates a 1-gigawatt data center powered entirely by domestic AI chips to support model development.

KEY DATA

$5 billionFunding Amount

954 million yuanH1 Revenue

~400%YoY Revenue Growth

101%API Price Increase

PHYSIX OBSERVATION

Competition in large language models has shifted from algorithmic rivalry to a resource war driven by capital and compute. By securing massive funding while cash-rich, Zhipu aims to seize the compute window before reaching the RSI technological tipping point. Its ability to raise prices against market trends while seeing demand explode validates the strong bargaining power of top-tier models. This marks the industry's entry into a 'heavy-asset, high-barrier' second half, where technical leadership alone is insufficient; achieving a scalable commercial loop is now the key to survival.

Source: enbrief original report ↗