The AI economy I saw at Stripe Sessions 2026

Writing: Gao Fei

Translation: AididiaoJP, Foresight News

In 1987, economist Robert Solow made a famous statement: “You can see the computer age everywhere but in the productivity statistics.” This remark puzzled economists for nearly a decade. It wasn’t until the mid-1990s that the contribution of computers to productivity finally became visible in the data.

By 2026, the same confusion is replaying with AI. Bubble theories come and go, scholars debate endlessly, companies hesitate, and macroeconomic signals remain blurry. But there is one place where AI’s impact on the economy is already undeniable.

Now, let’s look at Stripe.

In the past few days, I attended the Stripe Sessions held in San Francisco. Stripe processes transactions amounting to nearly 2% of global GDP, with annual payments reaching $1.9 trillion, and over 5 million businesses on its platform. 86% of the companies on Forbes AI 50 list are using Stripe. If the AI economy is a newborn, then Stripe is the heartbeat monitor in the delivery room. It records the baby’s heartbeat earlier and more accurately than almost anyone else.

A study released early 2026 by the Federal Reserve Bank of St. Louis shows that AI-related investments have contributed nearly 40% to the marginal GDP growth in the US, surpassing the peak contribution of the tech industry during the internet bubble era. When these investments translate into revenue, most settlements happen on Stripe. More importantly, Stripe is not just recording the heartbeat of the AI economy. At this year’s conference, it announced a new economic paradigm: Agentic Commerce, where agents become the primary actors in transactions. In a collective media interview, co-founder and President John Collison stated he expects agents to become the mainstream buyers in commercial transactions within 12 to 18 months.

Over two days, 288 products and features launched, and more than 10k attendees participated. A defining phrase ran throughout: Agentic Commerce. Below are my observations at Stripe Sessions 2026 and my personal reflections.

How fast is the AI economy really growing?

Before discussing agentic commerce, let’s first outline the overall landscape of the AI economy. Solow in 1987 said that computers left no trace in the statistics; nearly forty years later, AI’s presence is clearly visible in Stripe’s data.

On the first morning of the conference, CEO Patrick Collison presented some data. Since the pandemic, the number of new companies founded on Stripe each month has remained high but relatively flat. Starting early 2026, this curve sharply ascended vertically. The direct cause is that AI coding tools have drastically lowered the barriers to starting a business; many developers can now create billable products in days using “vibe coding.” Patrick described this as a broader phenomenon— the entire economy is re-platforming around AI. Maia Josebachvili, Stripe’s Chief Revenue Officer for AI, added an external comparison: until 2024, the number of app releases on the iOS App Store was declining. After AI coding tools emerged, release volume grew 24% month-over-month.

The change is not only in quantity but also in quality. Stripe Atlas is one of the easiest ways for founders to register a US company. Last week, it celebrated the milestone of 100k companies incorporated. At the conference, I heard astonishing data: companies registered via Atlas in 2025 generated twice the revenue at the same lifecycle stage compared to those in 2024. Companies founded in 2026, just a few months old, already have revenue five times that of the same period last year.

In the AI economy report on the first afternoon, Maia listed several names driving its rise. Lovable achieved $100 million in revenue within eight months and then surged to $400 million in the following eight months. Cursor reached $1 billion in annualized revenue in less than two years, then doubled to $2 billion in three months. Leading AI-native companies on Stripe grew 120% in 2025, and have grown 575% so far in 2026.

Consumer spending is similarly steep. The highest-spending users spend an average of $371 per month on AI products, exceeding the total monthly internet access, streaming, and mobile bills of an average American. I roughly calculated my own monthly token expenses, which already surpass my mobile bill.

Patrick also made a comparison: the growth rate of companies on Stripe is 17 times that of the global economy.

The next day, John Collison directly referenced the Solow paradox and explained it with a historical analogy. In 1882, Edison lit the first customer’s electric light in Manhattan. But during the subsequent thirty years of electrification, productivity hardly increased. The reason wasn’t that electricity was ineffective, but that factories at the time were designed around steam engines. Only after the entire factory was rebuilt did productivity improve. John’s judgment is that AI is at a similar stage. Change is happening, but old models haven’t fully absorbed it yet. “However,” he said, “I doubt AI will take thirty years.”

Stripe’s data seems to support his optimism. The AI economy is already exploding on its platform. Nearly every traditional company I’ve encountered at the conference’s top management level is urgently pushing AI deployment.

From day one, globalized

Besides speed, another impressive feature of these AI companies is that they are global from day one. Stripe has a saying: default to globalization.

Since I became an AI blogger, I often experience that AI content creation is timezone-agnostic. News from the other side of the Pacific carries the same weight as local news. The same applies to AI products. Large language models blur the traditional dependence on interface language and interaction habits. A unified chat window allows users worldwide to use products via natural language. In this sense, large language models have made a unified global software market possible for the first time.

Data from the conference confirms this observation. In previous SaaS waves, the fastest-growing companies covered about 25 countries in their first year, reaching 50 by the third year. AI companies follow a completely different pace: 42 countries in the first year, 120 in the third. Maia said Kazakhstan now appears on many AI companies’ market lists. In the “Indexing the Economy” sub-forum on the second day, Stripe provided a median figure: the top 100 AI startups sold in 55 countries in their first year.

Emergent Labs is a concrete example. Founded in the US in 2024, nearly 70% of its revenue now comes from overseas, with at least 16 countries contributing at least 1% each. Among leading AI companies, 48% of revenue is from markets outside their home country. Three years ago, this figure was only 33%. Global revenue is no longer supplementary but fundamental.

Speed and globalization are two core features of the AI economy, both directly related to Stripe. AI companies need to rapidly establish payment capabilities and be able to receive payments in 40 countries from the first week. That’s exactly what Stripe has been doing since its inception.

A brief note on Stripe’s founding background.

Stripe’s founders, Patrick and John Collison, are Irish and themselves cross-border entrepreneurs. At the conference, I met an Irish colleague who told me that in the eyes of Irish AI founders, these brothers are heroes. After moving to the US, they found online payments extremely difficult: connecting to payment systems required contracts with banks, PCI compliance checks, and multiple intermediaries, a process that could take weeks or even months.

So in 2010, these two young men in their early twenties dropped out of college, moved to San Francisco, and wrote a solution that allowed developers to accept payments with just seven lines of code. These seven lines coincided with the rise of mobile internet and SaaS. Shopify needed to help millions of merchants accept payments, Uber needed frictionless payments for riders, Salesforce needed to handle global subscriptions… all chose Stripe. As these global customers grew, Stripe built local capabilities in 46 countries, covering 195 markets and supporting 125 local payment methods.

For consumers, Stripe is not a company in the spotlight. It operates behind the scenes in Shopify checkout pages, OpenAI subscription confirmation emails, and Uber fee notifications. But this invisibility hasn’t prevented it from becoming the underlying financial infrastructure of the internet economy. In the AI era, this global financial infrastructure gives Stripe a first-mover advantage in serving AI companies expanding internationally.

At this year’s conference, I also met Stripe’s Global Product Lead, Abhi Tiwari. He took over this role just three months ago and moved to Singapore. Stripe has engineering centers in San Francisco, Dublin, and Singapore, and a Latin America office in São Paulo. Abhi told me many AI companies approach Stripe with the first words: “We default to globalization; it doesn’t matter where users are.” The old model of developing products at headquarters and then pushing them globally is being replaced by a new model where local teams build in-market.

Reaching global users is one thing; getting them to pay is another. The latter is much more complex because each market has its own currencies and payment habits. In this regard, Stripe mainly helps AI companies and other clients through two methods: local currency pricing and connecting local payment methods. The former shows Brazilian users prices in reais instead of dollars, increasing cross-border revenue by 18%; the latter enables Indian users to pay via UPI, Brazilian users via Pix, boosting conversion rates by over 7%. After integrating UPI, AI demo tool Gamma saw Indian revenue surge 22% in the month. At the booth, I also saw Chinese company MiniMax. I understand many Chinese outbound companies use Stripe’s financial services through overseas entities.

These AI-native companies also share another trait: very few personnel, often just solo founders. One or two people plus a handful of agents can support a truly revenue-generating global company. On the second day, Emily cited a statistic: the density of solo founders on Atlas is nearly 5,000 per million Americans, and more and more earn over $100k annually.

Emily used the term solopreneur. This reminded me of China’s rapidly growing One Person Company (OPC) wave. John explained this phenomenon using Ronald Coase’s theory of the firm: firms exist because internal coordination costs are lower than market coordination costs. But AI may be reversing this logic. When agents can help you discover services, integrate software, and handle payments, external coordination costs plummet. You no longer need a whole department of employees to do what previously required many.

From human economy to agent economy

The AI economy described above, no matter how fast it grows or how globalized, still involves human actors as primary transaction subjects. Humans buy AI products; humans use AI tools to start businesses. But the strongest signal I felt at this year’s Sessions is that Stripe’s next major focus is a different transformation: an economy where agents become market participants, called Agentic Commerce.

This shift is already subtly appearing in Stripe’s own data. Product and Business President Will Gaybrick showed some figures. For years, Stripe CLI (command-line interface) was used by a tiny, highly technical user base with almost no change. Starting in 2026, usage suddenly exploded. The reason is that agents don’t need fancy GUIs; simple CLI commands are often more useful. Maia’s data shows that in 2025, traffic from agents reading Stripe documentation increased about tenfold. If current trends continue, by year’s end, the number of agent reads of Stripe docs will surpass human reads. The API documentation Stripe spent over a decade refining has found its newest, most loyal audience.

If paying with agents sounds unfamiliar, consider two already happening scenarios.

First, shopping interfaces may be shifting toward model chat windows. Consumers now often use ChatGPT, Gemini, or research products on Instagram. The gap between research and purchase is compressed into a single interface. China has similar cases, such as buying milk tea within AI apps.

In a collective media interview, John Collison explained why this compression is hard to reverse, using his own experience buying a travel power adapter. If an agent completes the entire process from research to order and the product arrives in a few days, he won’t go to another website to fill out personal info from scratch, even if that site’s product is slightly better. Once a shopping agent completes the search, the next natural step is checkout.

A more interesting example is OpenClaw. Those familiar with the “Lobster” wave know it’s one of the hottest open-source autonomous agent frameworks. Users give instructions via messaging apps like Feishu, Telegram, or WhatsApp, and the agent autonomously executes tasks. The key is that OpenClaw can consume hundreds or even thousands of dollars worth of tokens per day. It manages token consumption and usage itself. Although manual authorization is often still needed, ultimately, the agent is spending tokens, which can be directly converted into money.

From managing token spending to agents paying directly—only one step away. At this year’s conference, Stripe demonstrated this step.

Demo: Agents buying and selling

On the main stage on the second day, a demo received multiple rounds of applause.

John Collison gave a simple instruction to an agent: research how AI demand affects the energy market. The agent begins searching, finds that Alpha Vantage has a needed energy market dataset priced at 4 cents, and autonomously completes the purchase and download using Tempo CLI’s stablecoin wallet because paying 4 cents with a credit card isn’t cost-effective. It then generates a complete analysis report. That’s already impressive. But John then told the agent: “Publish and sell this report. Set a reasonable price so other agents can find and buy it.” The agent checked Alpha Vantage’s licensing terms, confirmed commercial use was allowed, then built a website, published the report, and generated a command file for other agents to purchase the data via a request.

In just a few minutes, one agent completed the entire chain: research, procurement, production, compliance, publishing, pricing, and sales. It was both buyer and seller. After the demo, John said: “Agentic Commerce is here.”

The other two impressive demos from Day 1 included Will Gaybrick building an API review app that allows agents to obtain review services without being told any payment info. During execution, the agent automatically discovered the app used Machine Payments Protocol (MPP) and autonomously completed a $2 payment, with only one fingerprint authorization by a human. This zero-configuration payment discovery capability is core to MPP’s design. Developers don’t need to write separate payment logic; the agent finds it itself.

Next, Gaybrick combined Metronome (a real-time metering engine), Tempo (a blockchain designed for payments), and stablecoins to demonstrate streaming payments. An app charges in real-time based on AI token consumption, at $3 per million tokens. Multiple agents run simultaneously. The left dashboard shows rising token consumption; the right shows micro-payments in stablecoins flowing in sync. When opening the Tempo blockchain explorer, the total of $3.30 is composed of thousands of micro-payments, each only a few thousandths of a dollar. Neither credit cards, ACH, UPI, nor Pix can do this. Gaybrick announced on stage that this is the world’s first streaming payment system.

The return of micro-payments and new consumption logic

Shopping via chat windows and OpenClaw are examples of agents representing human consumption. But in a collective interview, Collison made a more ambitious prediction: agents could create entirely new demand.

He believes agents might enable a business model that has been discussed for years but never truly realized: micro-payments. Humans are not good at making extremely granular consumption decisions. Spotify replaced per-song payments with a $9.99 monthly subscription because no one wants to decide if a song is worth 15 cents each time they press play. Agents don’t have this cognitive burden. If this is correct, many business models that failed due to human cognitive friction could suddenly become feasible with agents. Maia also expressed a similar view in a one-on-one conversation with me. She said she recently spoke with dozens of AI founders, and pricing is the most frequently discussed topic when talking about agentic commerce.

Every transaction involves a buyer and a seller. If the buyer becomes an agent, what should merchants do?

In an interview, I asked Stripe Product Lead Jeff Weinstein: the saying “The customer is always right” implies merchants need to please consumers. So how do you please agents? Jeff’s answer was: think of agents as the best programmers you know. They want perfect information, structured formats, quick readability, and all the context needed for decision-making. Human consumers like beautiful images and smooth animations; agents want raw structured data, precise logistics info, and the ability to complete transactions in as few steps as possible.

In another conversation, Meta VP Ginger Baker summarized this shift more aggressively: payments will go from “instant” to “strategic.” Human purchases are discrete: you go to checkout, pull out your wallet, swipe, and the transaction is done. Agent consumption is continuous: you set rules like “grocery spending this week not to exceed $50,” “always prioritize this card,” or “transactions over $500 require manual approval.” Then, within your set authorization framework, the agent autonomously continues to spend.

Security: computing power as the new cash

If agents truly become a new kind of consumer, new risks will emerge. These risks differ fundamentally from traditional SaaS transaction risks and human consumer risks.

During the conference, I paid special attention to this topic and discussed it with several Stripe executives.

Stripe’s data and AI lead, Emily Glassberg Sands, described three rapidly growing fraud patterns. The first is multi-account abuse: the same person repeatedly registers different accounts, claiming free credits. According to Stripe network data, one in six AI company registrations involves this abuse. The second is malicious consumption during free trials: especially deadly for AI companies, as each trial burns real inference costs. She gave an example: a partner company’s token cost per paying customer exceeds $500, because converting a customer requires 25 free trials, 19 of which are fraudulent. The third pattern she called “free-riding”: customers consume tokens massively and refuse to pay at the end of the month. Emily also quoted a saying: “Compute power is the new cash.” When SaaS is abused, marginal costs are nearly zero; but each inference call in AI costs real money. Stolen tokens are stolen money.

However, there is a dilemma that troubles me greatly. Many AI founders respond to abuse by shutting down free trials.

Emily said she asked everyone claiming to have “solved” this problem, and found their solution was simply to turn off free tiers. But Jeff believes this creates another problem. Agents are becoming the main way to discover new services. If agents can’t try services themselves, they will jump directly to another URL. Emily added that if the call to action presented to agents is “join the waitlist” or “contact sales,” they will leave immediately. Shutting down self-serve registration to prevent fraud might mean handing over the most important growth channel to competitors.

Stripe’s answer to this dilemma is its fraud prevention system, Radar. Radar’s logic is simple: every transaction completed on Stripe teaches it something. Transaction data from 5 million businesses flows into a shared risk recognition network. If a company encounters a certain fraud pattern, all companies benefit. Last month, Radar blocked over 3.3 million high-risk free trial signups among eight high-growth AI companies.

Jeff also proposed an counterintuitive view: agent shopping might ultimately be safer than human web shopping. Trust validation in human web shopping relies on inference: how long the user stays on a site, whether click paths are normal, etc. But agent transactions can be programmatically verified. Stripe’s Shared Payment Tokens turn payment credentials into tokens; agents never see the raw credit card number. Users authenticate via biometrics and can set transaction limits, time windows, and merchant whitelists. When trust mechanisms shift from inference to confirmation, the security baseline may actually improve.

Ecosystem, protocols, and a bit of history

So far, it’s clear: without a well-functioning ecosystem, agentic commerce cannot be realized. At Stripe Sessions 2026, I met a food industry professional who said he attended to see if agentic commerce could become a new opportunity for his company — from a seller’s perspective.

Therefore, it cannot be achieved solely by Stripe; it requires an ecosystem.

Walking through the conference hall for two days, I saw many booths from companies in the financial supply chain. Stripe has also launched or joined a series of protocols with upstream and downstream partners, connecting different parts of the ecosystem: buyers and sellers, humans and machines, machine and machine. The Machine Payments Protocol (MPP) enables agents to discover and complete payments via HTTP. The Agentic Commerce Suite allows consumers to buy directly within AI apps from Google, Meta, OpenAI, and Microsoft. The Universal Commerce Protocol (UCP), initiated by Shopify and joined by Meta, Amazon, Salesforce, and Microsoft, is a cross-platform commercial protocol. Stripe has joined the UCP steering committee. A group of partners and competitors agree to develop a shared protocol because fragmentation makes cross-platform agent consumption difficult, which benefits no one.

On the topic of protocols, I noticed a particularly interesting Stripe partner: Visa. To me, Visa is essentially a protocol platform.

Seeing Visa immediately reminded me of a book I like very much: One from Many, by Visa founder Dee Hock. A core theme in the book is how banks, currency, and credit cards should be redefined in the electronic age. Money no longer has to be coins and paper; it can also be institution-backed, recorded on networks, and flowing globally as data. In the late 1960s, Bank of America issued BankAmericard, which expanded nationwide, bringing in many cross-state consumers and causing the old system to collapse. Hock realized the problem was organizational. Dozens of competing banks needed shared infrastructure, but existing organizational forms couldn’t allow cooperation and competition simultaneously. He applied decentralization principles to make all banks equal members of a new organization, and Bank of America relinquished exclusive control. This organization was later renamed Visa.

So, two different eras, two different companies, doing similar things — is there a certain inheritance between them?

Any agent can easily find the answer. Patrick Collison has publicly paid tribute to Hock. After Hock’s passing in 2022, Patrick called him “a severely underrated innovator,” profoundly influential to him and his brother. A clearer signal is in hiring decisions: Visa’s authoritative historian, David Stearns, later joined Stripe.

And a detail that will make those familiar with payment history smile: on stage, Tempo blockchain CTO Georgios Konstantopoulos showed a list of validators, one of which is Visa. Visa, founded by Hock, is now a participating node in the blockchain network incubated by Stripe. Students built the new network; the teacher became one of its nodes.

When Patrick traced Stripe’s origins at the conference opening, he said he was originally a Lisp programmer. The core idea of Lisp is “code is data.” He translated this idea into Stripe’s own language: “Stripe’s fundamental philosophy is that money is data. When we launched Stripe in 2011, this was not yet the industry orthodoxy.” Hock’s organizational theory approaches the essence of money as “a guarantee of value exchange,” which can be embodied in anything. Collison, from programming languages, directly equated money with data: data that can be programmed, called via APIs, and operated by agents. Both spoke the same thing in different words. On stage that day, Ginger Baker put it more plainly: “Money is just another form of digital content.”

If money is data, then data consumers will naturally become money consumers.

Side note: Stripe’s content gene

At this point, the story of the AI economy is nearing its end. But let’s take a small detour: Stripe can almost be seen as a peer of content creators.

This company is not only good at financial services but also excels in content products. Its publishing imprint, Stripe Press, has excellent taste; many know it for publishing Poor Charlie’s Almanack. Its podcast, A Cheeky Pint, is also distinctive, with a large audience. Google CEO Sundar Pichai, Anthropic CEO Dario Amodei, and a16z co-founder Marc Andreessen have all appeared on it.

During the conference, I met Stripe Press senior editor Tammy Winter and designer Pablo Delcan. Tammy joked, “Stripe is a publisher with a multi-billion-dollar company attached.” Pablo shared his understanding of taste: it’s the result of long-term accumulation, requiring time to settle. Regarding design trends, he believes that without sacrificing simplicity and clarity, the new challenge is how to add a certain level of complexity through details and precision.

Talking about books, Tammy told me that within Stripe Press, there’s a series called Turpentine. These books focus on how-to knowledge, tools, techniques, maintenance, and practical operations that keep work running. They are not abstract theories but aim to help readers solve specific operational problems.

The name comes from a story about Picasso: art critics gather to discuss form, structure, and meaning; artists gather to discuss where to buy cheap turpentine. The series aims to be the “cheap turpentine” for founders. If you think about it, for outbound AI companies, Stripe’s financial services are another kind of turpentine. You don’t need to worry about payments, compliance, or foreign exchange; you can focus on building your product.

This side thread may seem unrelated to the main story, but there’s an underlying connection. Stripe also has a magazine called Works in Progress, whose core question is how the economy grows. It features interviews with AI economy leaders. Sessions itself is somewhat like an economics lecture. The next morning, John Collison spent a whole session discussing economic data, Coase’s theory of the firm, and the Solow paradox. I suspect that a financial services company so concerned with economics does so because understanding the structural changes in the economy is how it finds the next product opportunity.

As a podcast enthusiast, when I saw John Collison on the first day, my first question wasn’t about finance but about podcasts. I asked him whether, after interviewing so many different people, there’s a core question underlying all conversations. He thought for a moment and said he’s really interested in how these people’s companies operate, what kind of competitive equilibrium they’re in, and how they understand their business.

Coincidentally, at the end of the first day, there was a small twist. The scheduled final fireside chat was supposed to be Patrick interviewing OpenAI co-founder Greg Brockman, but just before going on stage, the guest was replaced by Sam Altman. Patrick explained that “AI is a rapidly changing field.”

This turned surprise into joy. The audience cheered.

They have known each other for nearly 19 years. Altman was one of Stripe’s earliest angel investors, when the Collison brothers were under 20. Because of this, Altman appeared very relaxed throughout the conversation.

Near the end, Patrick asked a personal question: why did he invest in two teenagers back then? Altman said he remembers they wanted to build products that solved problems they personally encountered, and he saw the opportunity to scale because many others needed the same thing.

I think his answers about podcasts and investments point to the same core: find real needs, solve real problems. In the conversation, Altman divided OpenAI’s transformation into three stages: from research lab, to product company, to a “token factory” supplying intelligence to the world. Each stage has a different mission. Stripe is very similar. In 2010, the problem Irish young men solved was “online payments are too hard.” Along the way, they solved the same problem for 5 million users. By 2026, they discovered a new problem: their clients’ customers might soon no longer be human.

Holding a podcast in one hand and a publisher in the other, discussing Coase’s theory and the Solow paradox on stage, and laying out protocols and APIs in the exhibition hall, Stripe is not only creating the AI economy but also documenting it. During the conference, I had a somewhat crazy thought: Stripe has transaction data amounting to nearly 2% of global GDP. It can see where every dollar of AI income comes from, where it goes, and how fast it grows. If Solow had such a heartbeat monitor back then, he might not have had to wait ten years to find computers in the statistics.

Maybe one day, Stripe could provide a model for the AI economy. Not a large language model, but a Nobel-level economic model. Who says that’s impossible? Just a few years before Demis Hassabis, founder of DeepMind, received a Nobel Prize, who could have imagined it?