CoinWorld reports:
According to an analysis article cited by Wall Street Watch, the profit distribution in the AI industry chain is being reshuffled as the Token economy expands. The Chinese market is experiencing rapid growth in usage, but revenue conversion remains relatively slow; in contrast, the US retains some incremental value upstream and at the software end, supported by high capital expenditures and subscription systems.
The article suggests that the current order of profit realization is becoming clearer: upstream computing power accounts for initial profits, midstream model prices continue to decline, and downstream applications begin to absorb the space created by falling costs. The differences between China and the US lie not only in technological capabilities but also in who can truly retain this value on their balance sheets.
Growth in Usage Not Synchronized with Monetization
China's average daily Token usage has surged from about 100 billion times at the beginning of 2024 to approximately 100 trillion times by the end of 2025, yet the annual revenue from public cloud MaaS remains only about 3.07 billion yuan. The article mentions that reasons include a significant amount of usage occurring in large companies' internal scenarios, external trading prices being suppressed, and the application layer not yet fully transitioning to Token-based pricing.
In contrast to the revenue scale, the investment in computing power continues to rise. The article states that as of the second quarter of 2026, the rolling 12-month capital expenditure to operating cash flow ratio for the four major US cloud companies has risen to between 0.63 and 1.05, with some companies' free cash flow under significant pressure, relying more on capital markets for funding.
Upstream First to Capture Scarcity Premium
The article believes that upstream computing power remains the most stable profit capture point currently. The scarcity premium brought about by tight supply has directly boosted revenues related to chips and data centers. In the US market, bottlenecks are more related to power access and data center capacity, with scarcity primarily cleared through pricing.
In the Chinese market, constraints are more related to the supply of high-end computing chips. The article notes that under the backdrop of export controls, new demand is accelerating domestic substitution, with local manufacturers now holding over 40% of the AI accelerator card market share, and new computing power continues to concentrate on hubs like "East Data, West Computing."
Application Layer Begins to Capture Dividends
The article argues that midstream models are rapidly commodifying. As open-source models continue to close the capability gap, the substitutability of Tokens at the same level is increasing, with competition becoming more directly price-driven. The calling price for capabilities equivalent to GPT-4 has been compressed to about one-fortieth of what it was in the past.
In this context, the midstream faces dual pressures: transaction prices are below listing prices, and a significant portion of costs is tied up in high depreciation amortization, continuously compressing profit margins. The article predicts that the segments that can truly retain profits in the long term will shift more towards both ends: one end relies on computing power suppliers that dilute costs through high utilization, while the other end consists of application companies that can maintain pricing power through customer switching costs.
Diverging Monetization Paths in China and the US
The article states that the US market is more capable of embedding AI incremental value into high-priced software subscription systems, with cutting-edge model capabilities being packaged into existing SaaS products, leading to ongoing charges. In contrast, the willingness to pay for software in the Chinese market is relatively low, with midstream vendors often pricing based on computing power costs, causing the dividends from low-priced Tokens to flow more quickly to the application layer.
However, the flow of dividends to the application layer does not equate to profits being secured. The article believes that the key to whether downstream can truly retain profits lies in whether sufficient switching costs are established, such as industry entry, data integration accumulation, and system integration depth. As Token or outcome-based pricing gradually becomes more widespread, the customer stickiness and actual retention of application companies will also be reflected more quickly in quarterly data.
This content is provided for general informational purposes only and doesn't constitute financial, investment, legal, or tax advice. Any events, rewards, online promotions, or related information mentioned herein should not be considered a recommendation, solicitation, or invitation to purchase, sell, trade, or otherwise deal in any crypto assets. Crypto assets are highly volatile and may result in loss. The availability of WEEX services, products, and related events may vary by region. You are responsible for ensuring that your participation is in accordance with applicable local laws and regulations.





























