Cheap Chinese AI Models May Boost Industry Demand, Analysts Say

Price war reshapes AI economics

The artificial intelligence industry is experiencing dramatic price compression as competition intensifies between Chinese open-weight models and established U.S. providers. Large language model inference prices have plummeted from above $2 per million tokens in early June to just $1.20 this week, according to Silicon Data's LLM Token Expenditure Index, as first reported by the South China Morning Post.

The pressure has forced major players to respond aggressively. OpenAI recently announced an 80 percent discount on developer pricing for its lightweight GPT-5.6 Luna model and a 20 percent reduction for its mid-tier GPT-5.6 Terra offering. The price cuts contributed to a sharp sell-off in AI stocks last month as investors questioned whether U.S. technology companies were overvalued.

Why it matters

While falling prices initially spooked Wall Street, the cost compression could fundamentally expand the addressable market for AI services. Lower barriers to entry mean more enterprises can afford to deploy AI systems at scale, potentially driving infrastructure demand far beyond current levels—even as per-unit revenues decline.

Demand expansion through efficiency

Analysts are framing the price competition as a long-term positive for the industry. Morgan Stanley pointed to "Jevons Paradox," an economic principle where efficiency gains that lower costs ultimately increase total demand for a resource. Stephen Byrd, the firm's global head of thematic and sustainability research, wrote that the competitive push for efficiency "reinforces our fundamental view that demand for compute is likely to vastly exceed supply."

Silicon Data echoed this perspective, noting on social media that increased competition and lower prices are "good for consumer and enterprise users of AI and promotes much wider and faster AI adoption."

Three scenarios for market evolution

Morgan Stanley outlined potential outcomes from the open-weight versus closed-model competition. In a scenario where open-weight models dominate, prices would fall more drastically, spurring greater enterprise adoption. If closed systems prevail, a smaller number of labs would control the market with slower price declines due to oligopoly dynamics. A hybrid outcome would see closed frontier models handling complex workloads while open-weight alternatives manage high-volume, cost-sensitive tasks.

Across all three scenarios, cloud service providers including Microsoft, Amazon, and Google emerge as beneficiaries, according to the investment bank.

Cloud platforms show resilience

Recent earnings reports support the optimistic view. Microsoft's Azure and other cloud services posted 43 percent year-over-year revenue growth, while Amazon Web Services saw a 36.7 percent increase. Both companies' stocks rebounded strongly following their quarterly announcements, with Microsoft climbing nearly 28 percent and Amazon rising 20 percent since last week.

Nomura analysts suggested that competition from Chinese open-weight models strengthens cloud platforms' bargaining power by enabling them to host diverse advanced models, potentially making them more attractive than exclusive closed-source providers.

However, some observers warn of risks. IEEE Technology Blog author Alan Weissberger noted that while cheaper Chinese AI doesn't reduce raw demand for cloud platforms, it "lowers monetisation per unit of demand." If users route routine tasks to less expensive models, the same infrastructure may process more tokens while generating less revenue.

These details were first reported by the South China Morning Post.