From "subscription hell" to precise payment: A history of the evolution of internet pricing models

Written by: Sumanth Neppalli, Nishil Jain

Compiled by: AididiaoJP, Foresight News

Programmatic Payment vs Advertising

There are two completely different schools of thought in the crypto space. One view holds that everything is a market, and pricing things can give us clarity. The other view believes that crypto is a better financial technology infrastructure. Because, like all markets, there is no single truth. We are merely sorting through all possible patterns.

In today's topic, Sumanth will analyze how a new payment standard is evolving online. In short, it poses a question: what would happen if you could pay for each article? To find the answer, we go back to the early 1990s and see what happened when AOL tried to price internet access by the minute. We explore the path of Microsoft's SaaS subscription pricing.

In this process, we explained what x402 is, who the key participants are, and what it means for platforms like Substack.

The business model of the internet is disconnected from the way we use it. In 2009, the average American visited over one hundred websites per month. Today, the average user opens fewer than thirty applications each month, but spends much more time on them. Back then, it was around half an hour a day, now it's close to five hours.

Winners like Amazon, Spotify, Netflix, Google, and Meta have become aggregators, gathering consumer demand and turning roaming behavior into habits. They price these habits in the form of subscriptions.

This works because human attention follows patterns. Most of us watch Netflix most nights; we order from Amazon every week; the Prime membership bundle includes delivery, returns, and streaming services for $139 a year. Subscriptions eliminate a lot of ongoing pain. Amazon is now pushing ads to subscription users to increase profit margins, forcing users to either watch ads or pay more. When aggregators cannot justify the subscription model, they revert to an ad-based model like Google, which monetizes attention rather than intent.

Let's see what the ads are about now:

Robots and automation now account for nearly half of internet traffic. This is primarily driven by the rapid adoption of artificial intelligence and large language models, which make it easier to create and scale robots.

API requests account for 60% of the dynamic HTTP traffic processed by Cloudflare. In other words, machine-to-machine communication has taken up the majority of the traffic.

We designed today's pricing model for a purely human internet, but the current traffic is machine-to-machine and bursty. Spotify on the way to work, Slack during work hours, Netflix in the evening. Advertising assumes eyeballs, with people scrolling, clicking, and thinking. But machines have neither habits nor eyeballs. They have triggers and tasks.

Content pricing is not only a function of market constraints but also a function of the underlying distribution infrastructure. Music has existed in the form of albums for decades because physical media needed to be bundled. The cost of burning one song or twelve songs onto the same CD is nearly the same. Retailers require high profit margins, and shelf space is limited. In 2003, when the medium of distribution shifted to the internet, iTunes changed the accounting unit to songs. You could purchase any song from iTunes for $0.99 on your computer and sync it with your iPod.

The splitting of bundles has increased discovery, but it has also eroded revenue. Most music fans purchase popular songs rather than those ten filler tracks, compressing the per capita income of many artists.

Then, when the iPhone was introduced, the infrastructure changed again. Affordable cloud storage, 4G, and global CDNs made accessing any song instant and smooth. Phones are always online, providing instant access to an infinite number of songs. Streaming re-bundled everything at the access layer: for just $9.99 a month, you could listen to all recorded music.

Music subscriptions now account for over 85% of music revenue. Taylor Swift is unhappy about this: she was forced back to Spotify.

Enterprise software follows the same logic. Because the product is digital, vendors can charge for the exact resources used. B2B SaaS vendors provide predictable access to services on a monthly or annual basis, usually charging “per seat” and offering tiers with limited features, such as $50/user/month, plus $0.001 per API call.

Subscription covers predictable human usage, while metering addresses sporadic machine usage.

When AWS Lambda runs your function, you pay precisely for what you consume. B2B transactions typically involve bulk orders or high-value purchases, leading to larger transaction sizes and significant recurring revenue from a smaller, more concentrated customer base. Last year, B2B SaaS revenue reached $500 billion, twenty times that of the music streaming industry.

If most consumption is currently driven by machines and is sporadic, why are we still pricing it like in 2013? Because we designed today's infrastructure for the occasional choices made by humans. Subscriptions have become the default choice, as a month's decision outweighs a thousand small payments.

It is not cryptocurrency that has created the underlying infrastructure capable of supporting micropayments today. There are factors in that direction, but the internet itself has become such a massive beast that it requires new methods for pricing usage.

Why did the micropayment fail?

The dream of paying for content by the cent is as old as the internet itself. The Millicent protocol from digital device companies promised transactions below a cent in the 1990s. Chaum's DigiCash underwent banking pilots, and Rivest's PayWord has already solved cryptographic issues. Every few years, this elegant idea is rediscovered: what if you could pay $0.002 for each article and $0.01 for each song, exactly their cost?

America Online learned this lesson at a considerable cost in 1995.

Source - “The Case Against Microtransactions”

They charge dial-up internet fees by the hour. For most users, this is objectively cheaper than a fixed-rate subscription. However, customers hate it because it brings a mental burden. Every minute online feels like a meter is running, and every click carries a tiny cost. People can't help but record each micro-cost as a “loss,” even if the amount is small. Every click turns into a micro-decision: Is this link worth $0.03?

When America Online switched to unlimited plans in 1996, usage doubled overnight.

People pay more to reduce thinking. “Paying precisely for the content you use” sounds efficient, but for humans, it often feels like anxiety with a price tag.

Audrey Dziko summarized this point in his 2003 paper “The Case Against Micropayments”: people pay more for fixed-rate plans not because they are rational, but because they crave predictability over efficiency. We would rather pay an extra $30 a month for Netflix than optimize every $0.99 rental. Later experiments, such as Blendle and Google One Pass, attempted to charge $0.25 to $0.99 per article but ultimately failed. Unit economics do not work unless a considerable proportion of the reader base converts, and the user experience increases cognitive burden.

Subscribe to Hell

If we crave the simplicity of subscriptions, why are we complaining about subscription hell today? A simple way to understand pricing is to ask yourself how often you feel the pain that the product alleviates.

The demand for entertainment is infinite. The black line in the chart represents this constant pain point, which is a shared dream of users and companies—a flat, predictable, and constant pain curve. This is why Netflix transformed from a quirky DVD mailing service into a member of the elite FAANG club. It offers endless content and eliminates billing fatigue.

The simplicity of subscriptions has reshaped the entire entertainment industry. As Hollywood studios watched Netflix's stock soar, they began to reclaim their content libraries to build their own subscription empires: Disney+, HBO Max, Paramount+, Peacock, Apple TV+, LionsGate, and so on.

Fragmented content libraries force users to purchase more subscriptions. If you want to watch anime, you need a Crunchyroll subscription. To watch Pixar movies, you need a Disney subscription. Watching content has become a portfolio-building issue for users.

Pricing depends on two things: how accurately the underlying infrastructure can measure and settle usage, and who must make the decision each time consumption value is assessed.

One-time payments are very effective for rare, sudden events. Buy a book; rent a movie; pay a one-time consulting fee. Pain hits hard once, then disappears. This model works when tasks are infrequent and the value is clear. In some cases, this pain is even desirable, and we romanticize the trips to the theater or bookstore.

Accurate measurement of usage means that the price will align with the unit of work. That's why you wouldn't pay for half a movie. The value there is ambiguous. Figma cannot extract a fixed portion from your monthly output; creating value is difficult to measure.

Even if it's not the most profitable, charging a monthly fee is much easier.

The calculations are different: the cloud can observe every millisecond. Once AWS can measure execution at such fine granularity, renting an entire server no longer makes sense. Servers only start when needed, and you only pay for the time they are running. Twilio has done the same for telecommunications: one API call, one SMS segment, one charge.

Ironically, even in places where we can measure perfectly, we still charge like cable TV. We run on meters in milliseconds, but funds flow through monthly credit card subscriptions, invoice PDFs, or prepaid “points” buckets. To achieve this, each vendor makes you go through the same hurdles: creating an account, setting up OAuth/SSO for authentication, issuing API keys for authorization, storing cards, setting monthly limits, and praying not to be overcharged.

Some tools allow you to pre-charge points. Other tools, like Claude, limit you to lower-tier models when you reach your quota.

Most SaaS live in the green “predictable pain” range. They are too frequent for one-time purchases and too stable to justify precise measurement by events; the conventional practice is tiered pricing. You choose a plan that fits your typical monthly usage, and upgrade when usage exceeds the limit.

Microsoft's 1TB limit per user is an example that distinguishes light users from heavy users without the need to measure every file operation. The CFO restricts the number of users who need access to higher levels by allocating permissions.

The chaotic middle ground

A simple method for organizing pricing models is to use a two-dimensional graph, where the x-axis represents usage frequency and the y-axis represents usage variance. Here, variance refers to volatility, the degree of fluctuation in the pattern of a single user's activity over time. Watching two hours of Netflix most nights is low variance; an AI agent that makes 800 API calls in ten seconds and then remains silent is high variance.

In the lower left corner, there is a one-time purchase. When tasks are rare and predictable, a simple buy-and-go pricing is effective, as you move on after feeling the one-time cost.

The top left corner is a chaotic random network, with irregular news frenzies, link jumping, and low payment willingness. Subscriptions seem excessive, while small payments per click collapse under decision-making and transaction friction. Advertising has become the financing layer, aggregating millions of tiny, inconsistent views. Global advertising revenue has exceeded $1 trillion, with digital advertising accounting for 70% of spending, which indicates how many online lives exist in this low-commitment corner.

The bottom right corner is a very meaningful place to subscribe. Slack, Netflix, and Spotify align with human daily rhythms. Most SaaS lives here, separating heavy users from light users through tiers. Most products offer a free tier to encourage users to start using their products, and then gradually shift their usage from the top left to the bottom right through daily, consistent habits.

Subscriptions account for approximately $500 billion in global annual revenue.

The upper right corner is where the modern internet's focus lies, including LLM queries, proxy operations, serverless bursts, API calls, cross-chain transactions, batch jobs, and IoT device communication. Usage is both continuous and volatile. The fixed seating fee misprices this reality but lowers the psychological barrier to starting to pay. Light users overpay while heavy users are subsidized, causing revenue to drift further away from actual consumption.

This is why seat-based products have quietly shifted towards a metered model. Retain the basic plan for collaboration and support, but charge for heavy loads. For example, Dune offers a limited number of credits per month. Small, simple queries are cheap, while larger queries that run for a longer time consume more credits.

Cloud services standardize billing per millisecond for computing and data, while the API platform sells points that scale with actual work. It is moving towards linking revenue to the smallest observable unit of network energy. In 2018, less than 30% of software adopted usage-based pricing. Today, usage-based pricing has nearly reached 50% by eroding commission-based pricing, while subscriptions still dominate at 40%.

If expenditures are shifting towards consumption, then the market tells us that pricing hopes to stay in sync with work. Machines are rapidly becoming the largest consumers of the internet, with half of consumers using artificial intelligence for searches. Moreover, the content created by machines now exceeds that created by humans.

The problem is that our infrastructure still runs on an annual account basis. Once you register with the software provider, you can access their dashboard, which contains API keys, prepaid credits, and end-of-month invoices. This is fine for habitual humans; but for sporadic software, it becomes cumbersome. Theoretically, you can set up monthly recurring billing using ACH, UPI, or Venmo. However, these require batch processing to be used, as their fee structure collapses under high-frequency traffic below one cent.

This is where cryptocurrencies are crucial to the economics of the internet. Stablecoins provide you with programmable, global, and granular payments down to a fraction of a cent. They settle in seconds, operate around the clock, and can be held directly by agents rather than being trapped behind a bank user interface. If usage becomes event-driven, settlements should be too, and cryptocurrencies are the first infrastructure capable of keeping up.

What is X402 in reality?

x402 is a payment standard that works in conjunction with HTTP. It utilizes the 402 status code, which has a history of several decades and is reserved for small payments.

x402 is just a way for sellers to verify that a transaction has been completed. Sellers who hope to accept on-chain payments with no Gas through x402 must integrate facilitators like Coinbase and Thirdweb.

Imagine Substack charging $0.50 for a high-quality article. When you click the “Pay to Read” button, Substack returns a 402 code that includes the price, accepted assets (like USDC), the network (like Base or Solana), and the strategy. It looks something like this:

Your Metamask wallet authorizes $0.50 by signing a message and passes it to the facilitator. The facilitator sends the text to the chain and notifies Substack to open the article.

Stablecoins make accounting easy. They settle at internet speed, have small denominations, and do not require opening accounts with each vendor. With x402, you don’t need to pre-load five points buckets, switch API keys between different environments, or find out at 4 AM that your quota has been exhausted, causing your job to fail. Human bills can remain on credit cards where they are most effective, while all the sudden, machine-to-machine paths become automated and inexpensive in the background.

You can feel the difference in agent-based checkout. Imagine you're trying out new fashion styles on Daydream (AI fashion chatbot). Today, the shopping process will redirect you to Amazon so you can pay using your saved card information. In the world of x402, the agent understands the context, retrieves the merchant's address, and pays from your Metamask wallet without leaving the conversation thread.

The interesting part about x402 is that it is currently not a single entity; it is composed of various layers that you would expect to find in real infrastructure. Anyone who builds AI agents using Cloudflare's Agent Kit can create bots priced by operation. Payment giants like Visa and PayPal are also adding x402 as supported infrastructure.

QuickNode has a practical guide for adding an x402 paywall to any endpoint. The direction is clear: unify “proxy checkout” at the SDK level and make x402 the way to proxy payments for APIs, tools, and ultimately retail purchases.

Integration x402

Once the network supports native payments, an obvious question is where it will first emerge. The answer lies in high-frequency usage areas with transaction values below $1. This is the domain where subscriptions overcharge light users. This monthly commitment forces light users to pay the minimum subscription fee to start using it. x402 can settle each request at machine speed, with granularity as low as $0.01, as long as blockchain fees remain feasible.

Two forces make this transformation feel urgent. On the supply side, the “tokenization” of work is experiencing explosive growth: LLM tokens, API calls, vector search, Internet of Things pings. Every meaningful action on the modern web has already been attached to a small, machine-readable unit. On the demand side, SaaS pricing leads to absurd waste. Approximately four out of ten licenses are idle because finance teams tend to pay per seat, as this is easy to monitor and predict. We measure work at the technical layer but charge humans at the seat layer.

Event-native billing with a cap is a way to align these two worlds without scaring off buyers. We can have a soft cap, ultimately reconciling to the best price. A news website or developer API charges per request throughout the day, and then automatically refunds to the published daily cap.

If The Economist announces “$0.02 per article, daily cap of $2,” a curious reader can browse 180 links without having to do mental math.

At midnight, the protocol settles all fees to $2. The same model applies to the developer interface. News agencies can charge for each LLM scrape to sustain future AI browser revenue. Search APIs like Algolia can charge $0.0008 per query, with a daily usage total of $3.

You can already see that consumer AI is moving in this direction. When you reach Claude's message limit, it doesn't just say, “You've reached your limit, come back next week.” It offers two paths on the same screen: upgrade to a higher subscription or pay per message to complete what you're doing.

What is missing is a programmable infrastructure that allows agents to automatically make a second choice per request, without the need for UI pop-ups, cards, or manual upgrades.

For most B2B tools, the actual final state looks like “subscription base price + x402 bursts”. The team retains a basic plan linked to the number of users for collaboration, support, and backend use. Occasional recalculations (build minutes, vector search, image generation) are billed through x402 instead of forcing an upgrade to the next tier.

Double Zero aims to provide faster and cleaner internet through dedicated fiber optic sales. By routing traffic through their proxy, you can be charged per gigabyte at x402 pricing, with a clear SLA and cap. A proxy that requires low latency for trading, rendering, or model jumping can temporarily enter the fast lane, pay for that specific burst, and then exit.

SaaS will accelerate the shift to usage-based pricing, but with a safety net:

Customer acquisition and activation costs are reduced. You can make money on the first call. Those temporary developers can still pay $0.03. Agents prefer vendors that can pay instantly.

Income scales with actual usage rather than seat expansion. This is the way to cure 30-50% seat waste in most organizations. Heavy loads shift to burst billing with caps.

Pricing becomes a product interface. “An additional $0.002 for the fast lane per request,” “half-price batch mode,” these are all knobs that startups can experiment with to increase revenue.

The locking effect weakens. As the ability to try out vendors without the need for integration efforts and time increases, switching costs decrease.

A world without advertisements

Micro-payments will not eliminate advertising; they will narrow the area where advertising is the only viable model. Advertising still performs well in the realm of casual intent. x402 prices for interfaces that advertising cannot reach, and occasionally individuals may choose to pay for a good article without subscribing for a month.

X402 reduces payment friction; it may change the future when it reaches a certain scale.

Substack has 50 million users, with a conversion rate of 10%, which means 5 million subscribers pay about $7 per month. When the number of paying subscribers doubles to 10 million, that could be when Substack starts to earn more revenue from small payments. With low friction, more casual readers can turn to pay per article, accelerating the revenue curve.

The same logic applies to any seller with high variance and low-frequency sales: when people use a product occasionally rather than habitually, pay-per-use feels more natural than committing to a long-term plan.

This is somewhat similar to my experience visiting local badminton courts. I play two or three times a week, usually in different places with different friends. Most of these courts offer monthly memberships, but I don't like being tied down to one place. I enjoy having the freedom to decide which court we go to, how often I go, and to skip a session when I'm tired.

Don't get me wrong, I know this varies from person to person. Some people prefer to stick to the nearest court, some like to have a subscription that encourages them to form a routine, while others may want to share one with friends.

I cannot represent offline payments, but with x402, this individuality can be reflected in the digital world. Users can set their payment preferences through strategies, and companies can respond with flexible pricing models to adapt to everyone's habits and choices.

The real shine of x402 lies in the agency workflow. If the past decade was about transforming humans into logged-in users, then the next decade is about transforming agents into paying customers.

We have succeeded halfway. AI routers like Huggingface allow you to choose among multiple LLMs. OpenAI's Atlas is an AI browser that runs tasks for you using LLMs. x402 integrates as the missing payment infrastructure into that world. It is a way for software to settle small bills with other software at the precise moment the work is completed.

However, just having infrastructure does not constitute a market. Web2 has built a complete scaffolding around card networks. Bank KYC, merchant PCI, PayPal disputes, credit card blocks due to fraud, and chargebacks when errors occur. Currently, agency commerce lacks any of these. Stablecoins plus x402 provide agencies with a payment method, but they also strip away the built-in recourse that people are accustomed to.

When your shopping agent buys the wrong flight, or your research robot runs out of data budget, how do you get your money back?

This will be the content we will delve into next: how developers can use x402 without worrying about future failures.