Author | Jiang Xu For More Financial Information | BT Finance Data Pass The main text is 2,576 characters long and is estimated to take 9 minutes to read.

In March 2026, China's average daily Token calls reached 140 trillion, a more than thousandfold increase from early 2024. (Source: East Money, May 2026)

During the same period, the API call prices of mainstream AI large models generally fell by 80%-99%—DeepSeek, Kimi, Tongyi Qianwen, and others saw their cost per million Token calls drop from tens of yuan in 2023 to just a few cents.

On the surface, this seems like a win for users: AI is getting cheaper, and computing power is becoming more accessible.

Yet, at the same time, another set of numbers is moving in the opposite direction: Computing power rental company East Sunny Cloud Intelligence secured a major computing power service contract worth RMB 16-19 billion. (Source: East Sunny Announcement, May 2026); GPU rental prices rose from $1.70 per hour in October 2025 to $2.35 per hour in March 2026, a nearly 40% increase. (Source: Leton Electronics Annual Report, May 2026).

But in my view, the most noteworthy aspect this time is not the size of any single company's order—but rather: The pricing model in the computing power industry is undergoing a quiet shift, one that is reshaping profit distribution across the entire AI industry chain.

Some might say, isn't this just economies of scale? Lower prices due to higher volume?

Actually, no. As usual, I aim to explain this clearly in one article.

1

Why Is AI Getting Cheaper While Computing Power Gets More Expensive?

To understand this, we first need to grasp a concept: the computing power inflation paradox.

On the surface, cheaper AI calls should imply greater computing power supply and lower prices. But the reality is—cheaper AI leads to explosive growth in usage, and the rate of usage growth far outpaces the rate of price decline, causing total demand for computing power to surge.

Cross-industry analogy: This is like highway tolls dropping in price, only to see more cars on the road and toll booths collecting more revenue overall. In 2023, most companies used fewer than 1 million Tokens per month. By 2026, an ordinary AI programming tool could see over 100 million Token calls in a single day.

The explosion in total demand completely offsets the decline in unit price and, in turn, drives scarcity of computing power.

This is the essence of the computing power inflation paradox: Unit prices fall, but total spending rises; users feel it's cheaper, but the industry's computing power bills are skyrocketing.

2

The Shift in Pricing Models Is the Real Key

But simply saying "higher volume, lower prices, but greater total demand" doesn't fully explain why computing power companies' profits are rising. The deeper change lies in the pricing model.

Traditionally, computing power was priced based on "fixed-duration leasing"—like renting a parking space, you pay monthly regardless of use.

By 2026, the computing power rental industry is rapidly switching to a "revenue sharing based on Token calls" model—like a gas station, you pay for what you use, with higher usage leading to higher payments.

I call this shift: The transfer of pricing power from "parking spaces" to "gas stations." Parking fees buy space. Gas station fees buy consumption. When consumption grows exponentially, gas stations fare far better than parking spaces.

Under the fixed-duration model, a GPU server's annual revenue is fixed. Under the Token-based model, as long as the AI applications running on it see increased usage, the computing power company's revenue grows accordingly. And in 2026, AI application usage is growing exponentially.

3

Three Specific Outcomes of Computing Power Inflation

【Outcome 1】Scarcity of computing power is rising, not falling Intuitively, with increasing production capacity for computing chips, computing power should become less scarce. But there's a timing mismatch: Demand is growing much faster than capacity expansion. Google raised its 2026 capital expenditures to $180-190 billion, with Amazon and Microsoft following suit, totaling over $450 billion among the three. (Source: East Money, May 2026) These funds are building data centers that won't be delivered until 2028. Yet AI application demand is exploding in 2026.

【Outcome 2】Pricing power in the AI application layer is shifting toward the computing power layer Previously, it was assumed that the bulk of profits in the AI industry chain would come from the "model" layer. But computing power inflation is boosting the bargaining power of the computing power layer. Cross-era analogy: This is somewhat like the dot-com bubble of the 1990s, when companies selling network cables and servers ultimately profited more than those creating web content—the "picks and shovels" outperformed the "gold miners." In the AI era, computing power is the pick and shovel.

【Outcome 3】The Token economy has created a new profit distribution mechanism The emergence of the Token-based revenue sharing model allows computing power companies to directly participate in the commercial monetization of AI applications. This transforms their business model from "selling leases by time" to "participating in the AI application ecosystem's revenue sharing"—a systemic upgrade of the computing power layer's position in the AI industry value chain.

4

Where Can the Computing Power Inflation Paradox Be Applied?

The true value of this framework lies in its transferability.

Applied to personal electricity bills: Household appliances are becoming more energy-efficient, yet why is household electricity consumption rising? Because efficiency lowers the barrier to use, leading to more devices and longer usage times. Utility companies' total revenue doesn't decline—it rises. This is structurally identical to computing power inflation.

Applied to bandwidth economics: In the 4G era, per-unit data prices fell sharply, but operators' total revenue didn't collapse because usage growth offset the price declines. The same applies to 5G: Unit prices drop, but the number of connected devices and data transmission volumes expand exponentially.

Universal framework: As long as two conditions are met—falling unit prices and explosive usage growth—this "inflation paradox" will emerge. Understanding this framework helps identify industries experiencing similar structural shifts on the supply side.

5

What Does This Mean for You?

If you're an ordinary AI user: AI will indeed become cheaper, but as computing power inflation filters down to the application layer, you'll notice premium features adopting tiered pricing (as seen with Doubao's paid version). Free offerings will become more basic, and truly valuable tasks will require payment. This is structural.

If you're a business decision-maker: Computing power costs are becoming a core expense for AI application companies and will grow linearly with usage under the Token-based model. Effective budgeting for computing power costs is a new imperative in AI-era financial planning.

If you follow AI industry investments: The assumption that "the AI application layer will be most profitable" needs reevaluation. The computing power layer's bargaining power is rising, driven by changes in pricing models rather than one-time hype. Of course, for individual stocks, this is not investment advice.

This article has been heavily simplified. In reality, pricing models vary widely across different types of computing power and customer segments, and the Token-based model is still in its early stages.

What this article offers is a framework to understand the computing power inflation paradox: Falling unit prices → Explosive usage growth → Rising total demand → Sustained scarcity → Shift in pricing power from buyers to sellers. With this framework, when you see "AI getting cheaper" and "computing power companies landing major contracts" simultaneously, you'll understand these events are not contradictory.

This article is for information sharing and industry analysis only and does not constitute any investment advice, investment analysis, or trading solicitation. Market risks exist; invest cautiously. Anyone making investment decisions based on this article assumes all risks and outcomes, with the author and publishing platform bearing no legal responsibility.

This article is an original work by BT Finance and may not be used, reproduced, disseminated, or adapted without permission. Infringement will result in legal action.