August 12, 2026, (Inside AI) — Chinese AI company Zhipu is approaching 7 million registered API users on its MaaS open platform, adding roughly 2 million new users since early July. The surge highlights accelerating demand for model-as-a-service access in China's competitive AI landscape.
The growth comes alongside a major infrastructure expansion. Zhipu has activated more than 50,000 domestically developed AI chips to handle the rising inference workload. The chips, sourced from Chinese manufacturers, signal Beijing's push for semiconductor self-sufficiency amid US export controls.
Zhipu opened its previously restricted Coding Plan for purchase on July 31, following a price increase. The plan, which offers specialized coding model access, was previously limited to select partners. The move suggests Zhipu is monetizing its developer ecosystem more aggressively as competition intensifies.
The 7 million API user figure refers specifically to registered accounts on the platform's model APIs, not active monthly users. Zhipu has not disclosed daily active users or revenue metrics. The company competes with Baidu's ERNIE Bot, Alibaba's Tongyi Qianwen, and ByteDance's Doubao, all vying for enterprise and developer adoption.
China's AI chip sector has seen rapid growth since 2022 export restrictions limited access to Nvidia's advanced GPUs. Domestic alternatives from companies like Huawei, Biren Technology, and Cambricon have filled the gap, though they often lag in performance. Zhipu's deployment of 50,000 such chips indicates a large-scale bet on indigenous hardware for inference tasks.
Zhipu did not specify which chip vendor it used. Industry analysts note that Huawei's Ascend series is among the most mature options for large-scale AI inference. However, scaling to tens of thousands of chips requires significant software optimization, a challenge Chinese firms have been tackling with mixed results.
The API user growth reflects broader trends in China's generative AI market. According to a 2025 report by the China Academy of Information and Communications Technology, the country's AI platform user base grew 120% year-over-year, driven by enterprise demand for large language models.
Zhipu's pricing strategy has evolved rapidly. The company initially offered free trials and low-cost tiers to attract developers. The July price hike for the Coding Plan suggests a shift toward sustainable revenue, though the company has not disclosed profitability. Rival platforms have also adjusted pricing, with some offering free tiers for basic models.
The company was founded in 2019 and has raised over $340 million from investors including Alibaba and Tencent. It is known for its GLM series of models, which compete with Meta's Llama and OpenAI's GPT in Chinese-language tasks. Zhipu has positioned itself as a national champion in AI, aligning with government goals for technological sovereignty.
The activation of 50,000 domestic chips also underscores the energy demands of large-scale inference. Data centers housing such hardware require substantial power and cooling, raising questions about environmental impact and operational costs. Zhipu has not disclosed its data center locations or energy sources.
In a related development, Chinese regulators have been tightening oversight of generative AI services, requiring security assessments and content moderation. Zhipu's platform must comply with these rules, which could affect API usage patterns. The company has not commented on how regulations impact its user metrics.
Zhipu's expansion comes as the US considers further chip restrictions, potentially targeting even lower-performance GPUs. This could accelerate Chinese firms' reliance on domestic chips, making Zhipu's early adoption a strategic advantage. However, the long-term performance gap remains a concern for cutting-edge model training.
