China's artificial intelligence (AI) companies Z.ai and MiniMax could remain loss-making until 2030 despite rapid revenue growth, a forecast showed. The high cost of computing resources needed to develop AI models and expand services was cited as a key obstacle to improving profitability.

Hong Kong's South China Morning Post (SCMP) reported on Monday that Ellie Jiang, head of Asia internet and software research at Macquarie Group, said she expects losses at Z.ai and MiniMax to potentially continue through 2030.

The main burden is the cost of computing resources. Jiang said China's shortage of computing resources is 2 to 3 times more severe than the global supply shortage. That is because U.S. restrictions on exports of Nvidia advanced processors limit Chinese AI companies' ability to secure high-performance chips.

Jiang said Macquarie's forecasts for the two companies were calculated relatively conservatively, but assumed a high level of revenue growth. She said annual recurring revenue (ARR) is expected to rise quickly as paid users increase and companies expand adoption of generative AI.

The two companies' ARR is in fact rising steeply. Z.ai said in its recent first-half results that ARR reached $1.6 billion as of the end of August. MiniMax also disclosed that its chief executive, Junjie Yan (옌쥔제), said last month that ARR in August stood at $800 million.

But revenue gains are not immediately translating into improved profitability. That is because spending continues on computing infrastructure investment and research and development (R&D) costs needed for AI model training and service operations.

The market reaction was mixed. As concerns grew over long-term competitiveness and profitability, MiniMax shares fell 5.59 percent on its Hong Kong listing on Monday, while Z.ai shares slid 10.02 percent. The gap between rapidly rising revenue and cost burdens has emerged as a key investor focus.

Some analysis also said it is difficult to immediately interpret current losses at Chinese AI companies as a failure to monetise AI businesses. With the AI market still at an early stage, companies can prioritise investment to secure market share and develop technology while accepting a certain level of losses, it said.

Jiang said Chinese companies developing open-weight models are also improving monetisation. She said they can expand revenue through various channels such as API fees, building enterprise AI systems and commercial licensing contracts.

But criticism also continued that the current scale of revenue is still insufficient to justify the massive spending. Jiang said this does not mean the current situation will persist, but it is difficult at this point to say the companies are generating sufficient returns relative to costs.

The issue of securing computing resources is also being cited as a major task on the ground. Tang Jie (탕제), a co-founder of Z.ai, said in a recent interview with Qiushi, an official journal of the Chinese Communist Party, that the shortage of computing resources has become severe as AI application scenarios expand, and securing sufficient computing power remains an important challenge. HSBC also said in a report last month that MiniMax needs to increase additional investment to maintain competitiveness. It said future revenue growth potential also depends on whether the company can secure sufficient computing resources.

The market is therefore focusing on new indicators beyond ARR to assess the value of AI companies. ARR was used as a key metric in the early high-growth stage, but as companies scale up, there is a view that metrics are needed that more clearly show actual profitability and cost structures.

Jiang said investors are looking for more predictable and reliable figures to use as a benchmark for valuing companies, adding that the price-to-sales ratio (PSR) and gross profit can also be key valuation metrics.

Competition in China's AI industry is also expected to affect profitability. Jefferies said China's large language model (LLM) market has already become overcrowded and that it prefers full-stack cloud service platforms such as Alibaba Group Holding and ByteDance over independent AI labs. It said these companies have relative advantages in internally secured computing resources and data, investment capacity, and cloud infrastructure and services, making it easier to turn AI technology into actual revenue.

Ultimately, competitiveness among Chinese AI companies is unlikely to be determined by revenue growth alone. How efficiently they manage costs while securing sufficient computing resources, and how quickly they turn rapidly growing ARR into actual profit, are expected to be key variables that will determine corporate value going forward.