Zhipu CEO Zhang Peng

TMTPOST -- Zhipu AI, one of China’s leading foundation model startups and a member of the so-called “Six Tigers” of domestic AI, has secured another 1 billion yuan ($140 million) in strategic funding as it moves closer to a public listing.

The investment—led by Pudong Venture Capital Group and Zhangjiang Group—was announced Tuesday during Zhipu’s Open Platform Industry Ecosystem Conference in Shanghai. The first tranche of the funding has already been completed. As part of the deal, the parties are also partnering with Shanghai Instrument and Electronics and SPD Group to co-develop next-generation AI infrastructure in the city.

This marks Shanghai’s first state-led capital injection into Zhipu and comes on the heels of similar strategic investments by state-owned entities in Hangzhou, Zhuhai, and Chengdu earlier this year. In total, Zhipu has raised more than 3 billion yuan ($410 million) in state capital since March, as regional governments across China race to back national AI champions ahead of a potential wave of IPOs.

Zhipu completed its corporate restructuring earlier this year and has reportedly entered the IPO guidance phase, becoming the first among China’s major foundation model startups to formally begin preparations for a stock market debut. CEO Zhang Peng has emphasized that while the company is moving toward a listing, the ultimate timeline will depend on external market conditions.

Zhipu’s latest investment deal also extends its integration into Shanghai’s AI ecosystem. The company was previously inducted into the Zhangjiang Group-led “Model Power Community” in March, a public-private initiative designed to foster collaboration across upstream and downstream players in the AI value chain.

The collaboration includes Zhipu’s Model-as-a-Service (MaaS) platform, which offers localized “model pools” powered by its proprietary GLM series. According to the company, the goal is to build sovereign-grade infrastructure for Chinese AI systems that can scale across industrial and governmental use cases.

Tuesday’s announcement also unveiled the next phase of that strategy: a co-investment effort with the Z Fund to support startups developing intelligent agents. Zhipu will funnel hundreds of millions of yuan into the new “Agents Pioneer Program,” which includes a revamped Agent Aggregation Platform within its MaaS framework.

“The future of AI is not just about model performance—it’s about a fundamental reshaping of the production paradigm,” Zhang said at the event. “In this new era, developers, designers, and entrepreneurs are co-creating the ecosystem.”

In tandem with the funding announcement, Zhipu also released GLM-4.1V-Thinking, the latest version of its multimodal large model optimized for cognitive reasoning tasks. The 10-billion parameter model outperformed all other models in its class across 23 of 28 benchmark tests and matched or beat Meta’s Qwen-2.5-VL—despite the latter using seven times more parameters.

A lightweight version, GLM-4.1V-9B, was also launched to provide deployment flexibility in enterprise and government scenarios, highlighting Zhipu’s push for performance efficiency in constrained environments.

Zhipu’s rapid advances are drawing global attention—and concern. Last week, OpenAI mentioned Zhipu by name in a blog post, warning of intensifying geopolitical competition in the so-called “AI for Nations” arena.

OpenAI pointed to Zhipu’s growing cooperation with Belt and Road countries like Indonesia and Vietnam, where the Chinese firm is helping local governments build sovereign large models and joint AI innovation centers. OpenAI characterized Zhipu’s ambitions as a challenge to the U.S. tech ecosystem, describing it as a “Chinese version of OpenAI for Nations.”

Zhipu is also reportedly working to lock in Chinese technical standards across emerging markets, part of a broader effort to reduce reliance on American models. The firm was added to the U.S. Commerce Department’s Entity List in January for its role in advancing China’s AI capabilities.

Zhipu has also clashed with domestic rival DeepSeek, choosing not to cooperate on foundation model training and instead pursuing its own independent architecture.

As China intensifies its support for homegrown AI leaders, Zhipu appears well-positioned to lead a new wave of Chinese AI listings. With state-backed investors from Hangzhou, Zhuhai, Chengdu, and now Shanghai, the company has locked in both capital and political backing. And with GLM-4.1V’s performance edging closer to much larger Western models, Zhipu is signaling it may be ready to compete on the global stage—not just in benchmarks, but in geopolitics.