China’s artificial intelligence industry is discovering that attracting users is only half the battle. Z.AI, formerly known as Zhipu AI, has missed sales expectations as an increasingly aggressive price war among Chinese AI companies puts pressure on the economics of selling advanced models.
The result illustrates a growing contradiction at the heart of China’s AI ambitions. Companies are producing increasingly capable models at lower prices, helping accelerate adoption, but the same strategy is making it harder for AI developers to generate enough revenue to cover the enormous cost of research, computing infrastructure and talent.
Zhipu AI reported first-half revenue of 953.9 million yuan ($142 million), representing a 400% increase from a year earlier. That is rapid growth, but the company’s sales performance still fell short of expectations, highlighting how quickly the competitive environment is changing.
China’s AI Price War Intensifies
Chinese technology companies have increasingly used low prices to attract customers to their AI models.
The strategy has been encouraged by intense competition from companies including DeepSeek, Alibaba, Tencent and other emerging developers. Instead of charging premium prices for access to advanced models, companies are increasingly offering inexpensive or heavily discounted services.
That makes AI more accessible to businesses and consumers.
It also creates a difficult business environment for the companies providing the technology.
When competitors continuously cut prices, an AI company can increase the number of users while generating relatively little additional revenue.
The industry can therefore experience spectacular growth in usage without seeing equally spectacular growth in profits.
Revenue Is Growing Rapidly
The sales miss should not obscure the speed at which Z.AI is expanding.
First-half revenue increased fourfold from the previous year, demonstrating that demand for its AI services is growing rapidly.
The company has been attempting to commercialize its GLM family of models by selling AI services to businesses and developers.
It has also emphasized lower-cost products and the use of Chinese-made chips.
That approach fits Beijing’s broader goal of building a domestic AI ecosystem that is less dependent on foreign technology.
But revenue growth alone does not guarantee financial success.
AI models require significant computing resources, and maintaining competitive performance can require continuous investment in chips, data centers, software and researchers.
Cheap AI Changes the Market
The price war is changing the economics of artificial intelligence.
In previous years, investors often assumed that the most capable AI models could command premium prices because developing them required enormous amounts of capital.
Chinese companies are challenging that assumption.
If models become increasingly interchangeable, customers may have little reason to pay substantially more for one provider.
Instead, companies can switch between competing models based on price, performance and availability.
That gives customers more bargaining power.
It also puts pressure on AI developers to reduce costs.
The Cost of Computing
One of the biggest expenses for AI companies is computing.
Training large models requires enormous amounts of processing power.
Serving those models to millions of users can also become expensive because every query requires computing resources.
The economics become particularly difficult when companies reduce prices aggressively.
A lower price can increase demand, but if the cost of providing each additional query remains high, higher usage does not necessarily translate into higher profits.
AI companies therefore have to improve efficiency at the same time as they expand.
Chinese Chips Become More Important
Z.AI has been increasing its use of domestically produced chips as part of its commercialization strategy.
That has both economic and strategic significance.
China has faced restrictions on access to some of the most advanced foreign semiconductor technology.
Those restrictions have encouraged Chinese companies to develop domestic alternatives.
Using Chinese chips could eventually reduce costs and improve supply security.
However, domestic hardware must also become powerful and efficient enough to compete with leading international systems.
The AI price war therefore extends beyond software.
It is also a competition over computing infrastructure.
The DeepSeek Effect
DeepSeek has played an important role in changing expectations about AI pricing in China.
Its low-cost models demonstrated that competitive AI services could be offered at prices far below what many investors had expected.
That forced competitors to respond.
Once one major company cuts prices, others face a difficult choice.
They can maintain higher prices and risk losing customers, or cut prices and sacrifice revenue per user.
The result can be a race toward lower prices across the industry.
More Users, Lower Revenue Per User
This creates an important distinction between adoption and monetization.
A company may attract millions of additional users because its model is cheap.
But if those users pay very little, revenue may not rise as quickly as usage.
That is particularly challenging for companies whose operating costs increase with every additional request.
The long-term winners may therefore not simply be the companies with the most users.
They may be the companies that can serve those users at the lowest cost.
AI Agents Could Change the Economics
One potential solution is moving beyond basic chatbot access.
Chinese AI companies are increasingly developing AI agents capable of performing tasks for businesses.
Instead of charging users simply for asking questions, companies could charge for completing useful work.
That could create a much larger commercial opportunity.
An AI system that handles customer service, software development, research or administrative tasks could generate considerably more economic value than a simple chatbot.
The challenge is making those systems reliable enough for companies to trust them with important operations.
Corporate Customers Matter
Enterprise customers could become especially important for Z.AI.
Businesses may be willing to pay more for customized models, security, integration and specialized AI applications.
That could provide a more sustainable revenue stream than competing purely on consumer chatbot prices.
However, corporate customers are also highly cost-conscious.
If several Chinese AI providers offer similar performance, businesses can negotiate aggressively.
That keeps pressure on margins.
China’s Broader AI Strategy
The price war must also be understood in the context of China’s national AI ambitions.
Beijing wants China to become a leader in artificial intelligence while reducing dependence on foreign technology.
Lower AI prices can actually help advance that objective.
Cheap models encourage businesses, developers and consumers to experiment with AI.
More widespread adoption can stimulate demand for domestic chips, cloud services, software and AI applications.
From a national economic perspective, that could be beneficial even if individual AI companies struggle to earn large profits.
Investors Face a Difficult Question
For investors, the central question is no longer simply which company has the best AI model.
It is which company has the strongest business model.
A technically impressive model can still produce disappointing financial results if competitors offer similar capabilities for substantially less.
That makes pricing power increasingly important.
Investors will also need to evaluate computing costs, customer retention, enterprise demand and the ability to generate recurring revenue.
A Potential Race to the Bottom
The biggest risk is that Chinese AI companies become trapped in a cycle of continuous discounting.
Each company wants to gain market share.
Each fears losing customers to competitors.
Prices fall.
Customers benefit.
But companies may struggle to generate sufficient returns on their investment.
That could eventually force consolidation.
Smaller companies without sufficient capital may find it difficult to keep funding expensive model development.
The industry could ultimately become dominated by a smaller number of companies with stronger balance sheets and access to computing infrastructure.
The Price War Could Benefit Consumers
From a consumer perspective, the situation looks very different.
Cheaper AI services mean more people can access powerful technology.
Small businesses can use advanced models without paying the prices that once restricted AI to large corporations.
Developers can experiment more freely.
That could accelerate innovation across the economy.
The price war may therefore be economically beneficial in the short term even if it creates financial problems for AI developers.
US Companies Face a Different Challenge
The developments in China also matter for American technology companies.
US firms have generally pursued higher-priced AI services and enormous infrastructure investments.
If Chinese companies demonstrate that similar capabilities can be delivered at dramatically lower costs, American providers could face pressure to reduce prices.
That could affect the economics of the global AI industry.
The competition is therefore not only between Chinese companies.
It is increasingly a competition over the global cost of artificial intelligence.
Efficiency Could Become the Real Moat
The next stage of AI competition may focus less on model size and more on efficiency.
Companies that can deliver similar performance using fewer computing resources have a major advantage.
Better algorithms, specialized chips, improved software and optimized infrastructure can all reduce the cost of serving AI models.
That could allow companies to offer low prices without destroying their margins.
In other words, the winner of the price war may not be the company willing to charge the least.
It may be the company that can afford to charge the least.
Conclusion
Z.AI’s sales miss demonstrates the difficult economics emerging in China’s rapidly expanding artificial intelligence industry.
The company’s first-half revenue surged 400% to 953.9 million yuan, showing that demand for its technology is growing quickly.
But rapid revenue growth is not enough when competitors are aggressively cutting prices.
Chinese AI companies are increasingly competing on affordability, with companies such as Z.AI emphasizing lower-cost products and domestic computing technology.
That strategy could accelerate AI adoption across China.
It could also undermine the profitability of the companies building the technology.
The contradiction is central to the industry’s future.
China wants AI to become widely available, while AI companies need to earn enough money to finance research, computing infrastructure and talent.
Lower prices help achieve the first goal but can make the second much harder.
For customers, the price war is largely positive.
Businesses and consumers can access increasingly capable AI tools at lower costs.
For investors, however, the situation is more complicated.
The key question is whether rapid adoption can eventually translate into sustainable margins.
Z.AI’s experience suggests that this transition will not be automatic.
The company has achieved extraordinary revenue growth, but the competitive environment is forcing it to fight for customers in a market where price is becoming one of the most powerful weapons.
The next phase of China’s AI race may therefore be less about proving that its models can compete with the world’s best.
That battle is already producing increasingly capable systems.
The harder challenge will be proving that those systems can make money.
If the price war continues, AI companies will have to find new sources of revenue, improve computing efficiency or develop products that customers are willing to pay significantly more for.
AI agents and enterprise applications could provide that opportunity.
So could advances in domestic chips and infrastructure that lower the cost of running models.
Ultimately, the strongest AI company may not be the one with the most impressive model.
It may be the one that can deliver competitive performance at the lowest sustainable cost.
That is the economic test now facing Z.AI and China’s broader artificial intelligence industry.
