Chinese artificial intelligence companies Z.ai (also known as Zhipu AI) and MiniMax are expected to continue operating at a loss until at least 2030, despite rapid revenue growth, according to Ellie Jiang, head of Asia internet and software research at Macquarie Group. Jiang outlined the challenges facing these firms during the Macquarie Asia Technology Conference 2026, emphasizing the significant costs associated with developing and maintaining advanced AI models.
One of the key constraints is the high expense of computing power required for training large AI models. This issue is particularly acute in China, where restrictions imposed by the United States on access to Nvidia’s most advanced processors have intensified a domestic shortage of computing resources. Jiang described China’s “compute crunch” as being two to three times more severe than the global average, putting additional pressure on local developers.
Macquarie’s forecast remains cautious, projecting that both Z.ai and MiniMax will remain unprofitable through the end of the decade, even as annual recurring revenue (ARR) continues to climb. Jiang noted that while Z.ai aims to reach ARR of approximately US$2.4 billion by the end of 2026, Macquarie’s estimate is somewhat higher at close to US$3 billion. Z.ai reported an ARR of US$1.6 billion by the end of August, while MiniMax announced an ARR milestone of US$800 million during the same period. Despite these gains, neither company has yet achieved profitability.
The widening gap between growing revenues and sustained losses highlights the considerable ongoing investments these AI developers must make in infrastructure, model training, and research and development to maintain a competitive edge. Investors have so far tolerated these losses given the early stage of AI market adoption, but skepticism is rising regarding the long-term viability of specialized AI labs. Analysts at Jefferies have described China’s large language model sector as “overcrowded” and have expressed a preference for full-stack cloud service companies, such as Alibaba Group Holding and ByteDance. These larger players benefit from greater computing resources, extensive data access, stronger financial positions, and diversified revenue opportunities, giving them an advantage over stand-alone AI firms.
Both Z.ai and MiniMax have faced share price declines recently, with MiniMax’s Hong Kong-listed shares falling 5.59% and Z.ai’s losing just over 10% in one day. Industry experts underscore that continued funding is essential for frontier AI labs to keep pace with technological advancements, but current spending levels are not yet fully justified by revenues.
Jiang stressed that ongoing losses should not be interpreted as an inability to monetize AI technology. Monetization is improving even among Chinese companies developing open-weight AI models, which allow broader user access to the underlying code. These firms generate revenue through API usage, enterprise deployments, and commercial licensing, while increased user activity contributes valuable data for refining models. However, Jiang acknowledged that revenue growth does not yet completely offset costs but suggested that profitability might be attainable over time.
Z.ai co-founder Tang Jie highlighted the challenges posed by growing demand for AI applications, which has exacerbated shortages in computing capacity. HSBC analysts similarly noted that MiniMax needs to increase investment to secure sufficient computing power to support revenue expansion.
As the AI market matures, analysts are seeking more reliable valuation metrics beyond ARR, which, while useful during the rapid growth phase, may not fully capture the evolving financial dynamics of AI enterprises. Alternative measures under consideration include price-to-sales ratios and gross profit metrics.