Tokens are expensive, burning a hole in people's hearts.
This is not just the voice of those currently obsessed with Vibe Coding; even Silicon Valley behemoths who were previously fervently advocating for Tokenmaxxing have begun to restrict tokens for their own employees.
However, a counterintuitive point is that currently, those of you using AI subscriptions are actually using tokens that have already been subsidized by AI giants, with the highest subsidy possibly reaching up to 70 times the subscription fee!
What is even more worrisome is that both AI leaders, OpenAI and Anthropic, have entered the IPO sprint phase. Once these two companies go public,
will the remaining companies start raising their average revenue per user, similar to the "subsidy war" of the internet era, in order to bring token prices back to rational levels?
The good news is that this scenario may not occur. Recently, Bill Maris, the founder of Google Ventures, raised a question on the All-In podcast:
If Google decides to slash token prices by another 80%, how will OpenAI and Anthropic respond?
Similarly, not long ago, the startup team at Agnes AI, during a live broadcast with Geek Park, elaborated on the potential arrival of a "Token Free Era."
So, will the future price of tokens rise or fall? And what does this mean for those already addicted to AI?
01 The Token Subsidy Battle is in Full Swing
Why is it said that the current price of tokens is actually not expensive?
Because at least in the AI subscription model, the current prices of various AI companies have already been discounted to a "rock-bottom price" after subsidies.
Recently, SemiAnalysis conducted a detailed evaluation of the actual token value consumed and subscription fees under OpenAI and Anthropic's subscription models.
SemiAnalysis did a simple yet effective thing—actually used AI to perform various tasks under the subscription plans of various AI platforms, and then publicly priced these tasks using APIs to calculate the value of these tasks in tokens. The results are as follows:
Notice a pattern: The more expensive the package, the higher the subsidy multiplier. This in itself indicates that these high-end packages are not meant to make money— they are a form of "reverse pricing," using the most aggressive losses to retain the most heavy users. Because heavy users are developers, they are enterprise decision-makers. Once they are tied to a platform, they will bring along their entire team and product line.
Why continue to operate at such a loss? The standard answer is: burn money to gain scale first, then raise prices to recoup losses. This is how the mobile internet operates— Didi and Uber subsidized billions of RMB in taxi fares, and once the subsidy ended, the fares increased; Meituan subsidized numerous meal deliveries, and after the subsidy ended, the delivery fees went up.
This logic is valid under one key assumption: a lock-in effect is established during the subsidy period.
Didi can raise prices because drivers cannot leave the platform's order flow, and passengers cannot leave the platform's drivers. Meituan can raise prices because merchants cannot do without its traffic and delivery network. At the end of the subsidy, users are already "locked" into the ecosystem, with an extremely high switching cost.
However, in the AI arena, there is a fundamental difference from the internet— Tokens have almost no lock-in effect.
If Claude were to raise prices, developers could migrate their API calls to GPT or Gemini within a day— interfaces across providers are becoming more standardized, many development frameworks even have built-in multi-model switching capabilities. It's simpler for regular users: just switch to a different website. AI is unlike ride-hailing with a local driver network, unlike food delivery with a delivery system, or unlike social media with a friend network. A token is just a token, regardless of who produces it, it's the same thing.
This means that once the subsidy stops, users can instantly churn. Subsidies are not about "building barriers," more like "maintaining a pulse"— as long as someone offers a lower price, users will leave.
And this is not even taking into account a new variable that is causing everyone's bills to spiral out of control: AI Agent.
When you chat with ChatGPT, a conversation may consume a few thousand tokens. But when you have an AI Agent perform a complex task— write and debug a piece of code, analyze a dozens-page document and generate a report— after one round, the token consumption is "5 to 30 times" that of a normal conversation.
According to developers' tests, a single Agent programming session on the $100 Claude Max plan can burn nearly $100 worth of tokens. Uber's CTO recently revealed that the company burned through its entire 2026 AI budget in just four months.
The question is, can this kind of Token subsidy war continue? Who is likely to still be standing after the chaos to see the end?
Bill Maris believes the answer is obviously the traditional giants.
02 Token as a weapon
To understand the true brutality of this subsidy war, we first need to see a structural asymmetry—the ammunition sources of the warring parties are completely different.
Google's annual ad revenue exceeds $300 billion. This isn't money from investors, nor money burned through financing; it's an automatic cash printer running every day. Every day, billions of people around the world open the search engine, watch YouTube, use Gmail, and the ad revenue automatically flows into the account. Google doesn't need roadshows, doesn't need to please analysts, doesn't need to explain to anyone why it's spending this money.
Google uses ad profits to subsidize AI tokens, much like someone who owns an oil well engaging in a price war at a gas station—his oil comes from his own land, while the opponent's oil is bought with bank loans.
OpenAI and Anthropic are those people buying oil with loans.
OpenAI has raised over $180 billion in total financing, with a latest valuation exceeding $850 billion. Anthropic has raised over $130 billion. This money comes from venture capitalists and strategic investors—they don't give money for charity, they expect these companies to go public and hope to receive generous returns upon exit.
However, the real trouble starts after going public. An IPO means financial statements are public worldwide. Every quarter, Wall Street analysts focus on revenue, profit, user acquisition costs, and marginal costs. When they calculate that for every $1 of subscription fee received, you actually lose $70—no matter how brilliant the growth story, the stock price won't hold up.
Bill Maris was very straightforward in a podcast about this logic. In his own words: "If I were Google and decided to arbitrarily cut the token price by 80%, what would happen to the business models of OpenAI and Anthropic?"
The host asked how likely that scenario was. Maris didn't hesitate: "100%. Capital as a weapon, tokens as a weapon."
This is not analyst speculation. Bill Maris, the founder and CEO of Google Ventures and a former Google VP of Special Projects who incubated Waymo and Google X, has seen Google in action firsthand. Everyone in the room understands: this is not an assumption, this is him having witnessed Google fight.
The scenario he painted is quite simple: Google announces an 80% price cut for the Gemini API. What would enterprise customers do? If the product quality is similar — Gemini has already performed on par with Claude and GPT in many benchmarks — but the price is a quarter of what it used to be, would you stick with the expensive one?
Maris himself provided the answer: "If you're a company, and you can go to Google and Gemini and pay 80% less for essentially the same product, why wouldn't you? And that puts tremendous pressure on those companies."
Meanwhile, OpenAI and Anthropic have almost no symmetrical countermove. They cannot follow suit with price cuts — there's no money-printing press; every dollar comes from investors. They also can't rely on maintaining a premium through technological differentiation — the gap between large models is rapidly closing, today you're three months ahead, three months later you're caught up. It's not like the generational tech gap between the iPhone and Nokia. The moat between AI models is more like a dam made of sand, easily flooded when the tide rises.
In Bill's narrative, Google has a strong advantage, but in the AI world, can Google really monopolize? Meta can open-source a free model at any time, China has DeepSeek and Byte, Amazon is pushing its own model. When you slash the token price to dirt cheap, competitors don't disappear — they also start cutting prices.
The AI war may have no winner.
03 The "Infinite Game" of Tokens?
Even those who are not well-versed in history can somewhat discern the endgame of the current AI war:
The first scenario is the "Internet Service" playbook — the story of Didi, the story of Amazon: start with subsidies, then monopolize, and finally raise prices to harvest. In this playbook, today's price war is only the prologue, and eventually one or two winners will dominate the majority of the market and have pricing power. If this is the case, the current massive losses are a worthwhile investment — just like how Amazon lost money for twenty years before becoming the dual-monarch of e-commerce and cloud computing.
The second scenario is the "Hydropower and Coal" script. The token becomes a standardized basic resource, just like electricity, bandwidth, and cloud storage. No one can maintain pricing power in the long term because the product differences are too small, and the switching costs are too low. Competition continually drives prices down to the cost line, and profit margins approach zero. Eventually, the government may intervene with regulation—just as it did with electricity and telecommunications a hundred years ago.
The divergence of the two scenarios depends on one word:
Lock-in.
Didi can raise prices because passengers are locked into the driver network, and drivers are also locked into the order flow. Amazon can raise prices because merchants are locked into its logistics and traffic ecosystem.
The lock-in effect is the cornerstone of the "lose first, earn later" pattern.
But as already repeatedly argued earlier, the AI token almost lacks lock-in. API standardization, with switching costs close to zero. The core requirement for the establishment of the first scenario does not exist in this product that is a token.
If the second scenario, the endgame of "Hydropower and Coal" infrastructure, is closer to reality, what we are witnessing is not a war that will eventually determine a winner, but a relentless race without end.
Meituan's founder Wang Xing once described this competitive state. His insight was that some competitions have no concept of "winning." The goal of participants is not to defeat opponents but to ensure that they are always at the table. Because as long as you are at the table, you can continue to raise funds, hire people, and iterate. Leaving the table is the only way to lose.
Looking at today's AI landscape through this framework, many seemingly contradictory things suddenly become clear.
OpenAI's latest valuation exceeds $800 billion, not because training models require that much money. It needs this much money to continue the price war. Fundraising is not to win but to "qualify to continue playing."
Google is preparing to reduce token prices by 80%, not to eliminate OpenAI and Anthropic. It is to ensure that it remains a core player in the AI era—just as it ensured it was not left behind in the mobile era through free Android.
And Anthropic has raised the pricing of the API for its latest flagship model, Fable 5, to twice that of the previous generation—$10 per million tokens for input and $50 per million for output—seemingly "increasing prices" but actually proactively selecting enterprise customers willing to pay for high-end capabilities because it knows deep down that the consumer-side subsidy war cannot defeat Google.
With each price war, the scale of AI usage expands. This scale expansion means more data, more scenarios, and more developers entering the ecosystem. This, in turn, makes the models of all participants stronger. The warring parties use the war itself to attract resources to upgrade themselves—not a zero-sum game of life and death, but a process where everyone becomes stronger through competition, yet none are likely to earn windfall profits.
Does this sound like the eventual form of the electricity industry?
140 years ago, Edison and Westinghouse both thought they were competing for a winner-takes-all market. They wagered their entire fortunes on "whoever defined the standard for electricity would own electricity." But the fate of electricity taught us a simple lesson:
When a technology is important enough, universal enough, and standardized enough, it no longer belongs to any one company. It belongs to the infrastructure.
The competition in AI, on the surface, is Google versus OpenAI versus Anthropic, a showdown of model capabilities, a battle of financing scales. But zooming out, the true role of this competition is: it is accelerating AI towards an infrastructure level that no company can monopolize.
When Bill Maris said "It will definitely happen," he may not have just been predicting that Google would lower prices. He may have been unwittingly predicting a larger trend—In the world of AI, tokens will eventually not belong to anyone. Just like today, no one "owns" electricity.
For OpenAI and Anthropic, this means an unsettling thing: even with a technological edge, even with astronomical funding, the future they are chasing of "making big money with AI" may not have existed from the beginning. What they are facing is not a temporary price war, but a structural destiny—what they are striving to build may fundamentally be the next generation's water, electricity, and highways.
And for users, to some extent, this may be good news. Because as long as the Token subsidy war continues, people can still enjoy the "good deal" of a $20 cost and $400 computing power.
Welcome to join the official BlockBeats community:
Telegram Subscription Group: https://t.me/theblockbeats
Telegram Discussion Group: https://t.me/BlockBeats_App
Official Twitter Account: https://twitter.com/BlockBeatsAsia
