Inside the AI Economy: What Stripe's Data Reveals — Nilofer Rajpurkar, Stripe

Good morning everyone, Good morning everyone, thank you for joining thank you for joining thank you for joining us today. My us today. My us today. My name is Nilufar Raji name is Nilufar Raji name is Nilufar Raji Rokar, and I'm a Lead Rokar, and I'm a Lead Rokar, and I'm a Lead Product Manager Product Manager Product Manager at Stripe. I work in a at Stripe. I work in a at Stripe. I work in a team of agents and team of agents and team of agents and developers, and these developers, and these developers, and these areas are becoming areas are becoming areas are becoming increasingly inseparable. At increasingly inseparable. At increasingly inseparable. At Stripe, we partner with the Stripe, we partner with the Stripe, we partner with the fastest-growing AI fastest-growing AI companies, many of companies, many of companies, many of which are here today. We which are here today. We which are here today. We observe from the observe from the observe from the front lines what front lines what front lines what works, what doesn't, and how these works, what doesn't, and how these works, what doesn't, and how these companies companies companies scale. So scale. So scale. So today I'm excited today I'm excited today I'm excited to show you data on to show you data on to show you data on how companies how companies how companies are building their are building their are building their businesses. There are several businesses. There are several businesses. There are several patterns that patterns that patterns that we see in leading we see in leading we see in leading AI companies. First, AI companies. First, AI companies. First, they build faster. they build faster. they build faster. I think we all see I think we all see I think we all see this happening in this happening in this happening in real real real time. Second, they time. Second, they time. Second, they sell globally sell globally sell globally by default. From by default. From by default. From day one, the whole world day one, the whole world day one, the whole world becomes your market. becomes your market. Third, they adapt Third, they adapt Third, they adapt pricing to the AI pricing to the AI economy. 10 years economy. 10 years economy. 10 years ago, subscriptions were the ago, subscriptions were the ago, subscriptions were the new hot new hot new hot trend, but now it does trend, but now it does trend, but now it does n't work anymore. And n't work anymore. And n't work anymore. And lastly: they lastly: they lastly: they adapt very quickly adapt very quickly adapt very quickly to new to new to new go-to-market strategies. This is the go-to-market strategies. This is the go-to-market strategies. This is the general picture of what general picture of what general picture of what we will we will we will talk about today. And talk about today. And talk about today. And before we get started, I before we get started, I before we get started, I want to back it all up with a want to back it all up with a want to back it all up with a little bit of little bit of little bit of data. So, in 2025, data. So, in 2025, data. So, in 2025, we saw a we saw a 120% growth in AI revenue. A year later, we 120% growth in AI revenue. A year later, we see that this see that this see that this growth is growth is growth is already 175%. Almost already 175%. Almost already 175%. Almost tripled. So, tripled. So, tripled. So, revenue growth in revenue growth in revenue growth in AI is not AI is not AI is not slowing down.

slowing down. slowing down. In fact, it is In fact, it is In fact, it is growing rapidly, and this is what growing rapidly, and this is what growing rapidly, and this is what we we we will talk about in more will talk about in more will talk about in more detail today. And the people detail today. And the people detail today. And the people who lead these AI who lead these AI companies are even more companies are even more companies are even more outstanding. Take outstanding. Take outstanding. Take Lovable, for example. They Lovable, for example. They Lovable, for example. They reported 100 reported 100 reported 100 million in 8 million in 8 million in 8 months, and 400 million in another 8 months, and 400 million in another 8 months, and 400 million in another 8 months months months . Cursor's . Cursor's . Cursor's annual revenue reached $1 annual revenue reached $1 annual revenue reached $1 billion in 2 years, and billion in 2 years, and doubled to doubled to $2 billion just 3 months later. $2 billion just 3 months later. $2 billion just 3 months later. Anthropic shared that they Anthropic shared that they Anthropic shared that they went from zero went from zero went from zero in January 2023 to a in January 2023 to a in January 2023 to a billion in just billion in just billion in just 2 years, and now 2 years, and now 2 years, and now they are at 30 they are at 30 they are at 30 billion. So the billion. So the billion. So the growth rate here is growth rate here is growth rate here is really crazy. But really crazy. But really crazy. But we don't just see this in we don't just see this in we don't just see this in our B2B data. We our B2B data. We our B2B data. We see this in our see this in our see this in our consumer data consumer data consumer data with our consumer with our consumer with our consumer product Link. Our product Link. Our product Link. Our Link data shows that Link data shows that Link data shows that consumer consumer consumer adoption of AI adoption of AI products has doubled products has doubled products has doubled from just under 6 from just under 6 from just under 6 million to over 14 million to over 14 million to over 14 million in just million in just million in just one year. And all of these one year. And all of these one year. And all of these customers are also customers are also customers are also spending more. spending more. spending more. Our top Our top Our top Link buyers are spending Link buyers are spending Link buyers are spending nearly $360 per nearly $360 per nearly $360 per month on AI month on AI tools and tools and tools and services, up from services, up from services, up from $183 just 3 $183 just 3 $183 just 3 months ago. So that's months ago. So that's months ago. So that's almost a doubling in almost a doubling in almost a doubling in one quarter. That's one quarter. That's one quarter. That's more than more than more than the average the average the average American spends American spends American spends on their phone, internet, on their phone, internet, on their phone, internet, and streaming and streaming and streaming services combined.

services combined. services combined. But the difference is that But the difference is that But the difference is that users don't users don't users don't treat it treat it treat it like a streaming like a streaming like a streaming service that can be service that can be service that can be canceled tomorrow. AI canceled tomorrow. AI tools are becoming a tools are becoming a tools are becoming a critically critically critically important resource for them. important resource for them. important resource for them. So, we talked a little bit So, we talked a little bit So, we talked a little bit about about about revenue growth and revenue growth and revenue growth and doubling of doubling of doubling of adoption rates, but adoption rates, but adoption rates, but the question is, what's the question is, what's the question is, what's really behind really behind really behind that growth? I'd that growth? I'd that growth? I'd like to start with the like to start with the like to start with the first model first model first model I mentioned— I mentioned— faster development—and faster development—and faster development—and talk a little bit about talk a little bit about talk a little bit about speed. Just speed. Just speed. Just as you no longer as you no longer as you no longer need months of need months of need months of development to development to development to accept payments, accept payments, accept payments, you no longer need you no longer need you no longer need months of work to months of work to months of work to build and release build and release software. By software. By the end of 2024, the end of 2024, the end of 2024, the number of the number of the number of iOS app releases iOS app releases iOS app releases was declining. And was declining. And was declining. And then you see how then you see how then you see how agent-based agent-based programming tools came along, and that programming tools came along, and that growth rate growth rate growth rate increased significantly to increased significantly to increased significantly to 24% per month. We 24% per month. We 24% per month. We saw a very similar saw a very similar saw a very similar trend with trend with trend with new new new company registrations in Delaware as well. company registrations in Delaware as well. company registrations in Delaware as well. You can see where You can see where You can see where the graph starts to go up sharply the graph starts to go up sharply the graph starts to go up sharply , again , again , again coinciding with coinciding with coinciding with the release of these the release of these the release of these agent-based agent-based programming tools. So programming tools. So everyone's talking about these everyone's talking about these everyone's talking about these non-technical non-technical non-technical founders who are " founders who are " vibe coding" and vibe coding" and vibe coding" and creating new creating new creating new businesses. And yes, that is businesses. And yes, that is businesses. And yes, that is definitely true.

definitely true. definitely true. But what we found But what we found But what we found interesting is that interesting is that interesting is that the share of technical the share of technical the share of technical founders is also founders is also founders is also growing and growing and growing and has increased by 7% in the year. has increased by 7% in the year. has increased by 7% in the year. Many of you in this Many of you in this Many of you in this room are technical room are technical room are technical specialists, and you specialists, and you specialists, and you probably do in a probably do in a probably do in a week what week what week what used to take teams used to take teams used to take teams months. So months. So months. So the most important thing here is that the most important thing here is that the most important thing here is that speed is now speed is now speed is now available to everyone. Both available to everyone. Both available to everyone. Both for non-technical for non-technical for non-technical founders who founders who founders who host and build host and build host and build projects on projects on developer platforms, and developer platforms, and for technical for technical for technical founders who are founders who are founders who are used to used to used to using Cursor using Cursor using Cursor or Claude locally. It has never or Claude locally. It has never or Claude locally. It has never been so fast been so fast been so fast to create a product and to create a product and to create a product and start start start monetizing it. And we monetizing it. And we monetizing it. And we actually see this in the actually see this in the actually see this in the data from data from data from developer platforms that developer platforms that developer platforms that partner with Stripe. partner with Stripe. partner with Stripe. So for some of our So for some of our So for some of our non-technical non-technical non-technical users who users who users who build on Lovable, Vercel, build on Lovable, Vercel, build on Lovable, Vercel, or Replit, we see some or Replit, we see some or Replit, we see some interesting data. Each interesting data. Each interesting data. Each subsequent monthly subsequent monthly subsequent monthly cohort attracts cohort attracts cohort attracts paying customers paying customers paying customers faster and faster. faster and faster. faster and faster. In fact, our In fact, our In fact, our latest cohorts latest cohorts latest cohorts go from go from go from product to paying product to paying product to paying customer in less than 6 customer in less than 6 customer in less than 6 weeks. And this weeks. And this weeks. And this speed is speed is speed is also available to technical also available to technical also available to technical founders. This founders. This founders. This experience is made better experience is made better experience is made better by the work of my by the work of my by the work of my team at Stripe, team at Stripe, team at Stripe, releasing things releasing things releasing things like our MCP server, like our MCP server, like our MCP server, agent skills, and agent skills, and agent skills, and agent management. So I agent management. So I agent management. So I would like to would like to would like to demonstrate demonstrate demonstrate this to you with an example. Here this to you with an example. Here this to you with an example. Here I have a new business, Vibe I have a new business, Vibe I have a new business, Vibe Check, and I'm ready Check, and I'm ready Check, and I'm ready to add payments. So, to add payments. So, to add payments. So, I'll go over to Claude, I'll go over to Claude, I'll go over to Claude, talk a little bit about talk a little bit about talk a little bit about my business model, and my business model, and my business model, and let Claude know that I'm let Claude know that I'm let Claude know that I'm ready to add ready to add ready to add support for subscriptions and support for subscriptions and support for subscriptions and billing. Claude rightly billing. Claude rightly billing. Claude rightly recommends building a recommends building a recommends building a solution based on Stripe. And

solution based on Stripe. And solution based on Stripe. And it becomes even more powerful it becomes even more powerful when you add our when you add our when you add our MCP skills, our MCP skills, our MCP skills, our MCP server, and the skills of our MCP server, and the skills of our MCP server, and the skills of our agents. But this agents. But this agents. But this information allows information allows information allows Claude to provide a better Claude to provide a better Claude to provide a better recommendation on recommendation on recommendation on how to how to how to build an build an build an integration with Stripe. integration with Stripe. integration with Stripe. So he uses So he uses So he uses our MCP our MCP Stripe integration recommendation tool Stripe integration recommendation tool to ask me for to ask me for to ask me for a little more a little more a little more information to information to information to create the right create the right create the right integration. He integration. He integration. He asks where asks where asks where my clients are located my clients are located my clients are located to decide to decide whether to whether to whether to recommend Stripe Tax. recommend Stripe Tax. recommend Stripe Tax. He knows it's better He knows it's better He knows it's better to recommend Stripe to recommend Stripe to recommend Stripe Checkout over Checkout over Checkout over Card Elements Card Elements Card Elements because Checkout has because Checkout has because Checkout has better better better subscription integration. So I will tell subscription integration. So I will tell subscription integration. So I will tell Claude that my clients Claude that my clients Claude that my clients are in both the US and the are in both the US and the are in both the US and the EU. And he tells EU. And he tells EU. And he tells me that I'll probably me that I'll probably me that I'll probably want to use want to use want to use Stripe Tax, which he Stripe Tax, which he Stripe Tax, which he learned from learned from learned from our our our agents' skills, because in the agents' skills, because in the agents' skills, because in the EU my digital EU my digital EU my digital subscriptions are subject to subscriptions are subject to subscriptions are subject to VAT. And VAT. And VAT. And since I'm not a since I'm not a since I'm not a tax expert, I'll tax expert, I'll tax expert, I'll happily happily happily take that take that take that advice. You advice. You advice. You can also see Claude can also see Claude can also see Claude creating a new creating a new creating a new Stripe sandbox for me. This is Stripe sandbox for me. This is Stripe sandbox for me. This is done done done using a new using a new using a new CLI command that allows the CLI command that allows the CLI command that allows the agent to agent to agent to create a create a create a Stripe sandbox without requiring Stripe sandbox without requiring Stripe sandbox without requiring any any any prior prior prior Stripe credentials from me. It Stripe credentials from me. It Stripe credentials from me. It independently obtains independently obtains independently obtains API keys for API keys for API keys for the sandbox and the sandbox and the sandbox and stores them securely.

stores them securely. stores them securely. Now I can get a Now I can get a Now I can get a working end-to-end working end-to-end working end-to-end demo without leaving demo without leaving demo without leaving Claude and see that the Claude and see that the Claude and see that the checkout process is checkout process is checkout process is fully functional. So, Claude fully functional. So, Claude fully functional. So, Claude manages my keys manages my keys manages my keys and then reviews and then reviews and then reviews the other skills of our the other skills of our the other skills of our agents to agents to agents to make sure he's make sure he's make sure he's making the right making the right making the right recommendation. Now recommendation. Now recommendation. Now he's ready to ask he's ready to ask he's ready to ask me if I want to me if I want to me if I want to use use use MCP tools to MCP tools to MCP tools to create a create a create a product catalog, and I'll product catalog, and I'll product catalog, and I'll say yes. We'll say yes. We'll say yes. We'll skip ahead a bit skip ahead a bit skip ahead a bit to the to the to the finished product. finished product. finished product. Now there is a subscription button on my website Now there is a subscription button on my website . I'll click " subscribe" and subscribe" and subscribe" and see a fully see a fully see a fully working payment form, working payment form, working payment form, ready for testing. ready for testing. ready for testing. I will enter the test I will enter the test I will enter the test card details, click " card details, click " subscribe" and subscribe" and subscribe" and see how the see how the see how the test payment goes. That's test payment goes. That's test payment goes. That's all. Now I can see the all. Now I can see the all. Now I can see the full end-to-end full end-to-end full end-to-end process of what process of what process of what my customers will see when my customers will see when my customers will see when trying to trying to trying to subscribe on my subscribe on my subscribe on my site, and site, and site, and I didn't have to I didn't have to I didn't have to leave Claude to do it. So what does this leave Claude to do it. So what does this leave Claude to do it. So what does this mean for you in mean for you in mean for you in practice? First, you practice? First, you practice? First, you must be must be must be obsessed with obsessed with developer productivity. developer productivity. The best teams The best teams The best teams working on this working on this working on this are treating it are treating it are treating it as a priority, not as a priority, not as a priority, not as something as something as something secondary, because secondary, because secondary, because development times are getting shorter and development times are getting shorter and development times are getting shorter and shorter. The shorter. The shorter. The tools we tools we tools we build, such as the build, such as the build, such as the MCP server, agent skills and MCP server, agent skills and MCP server, agent skills and management, and management, and safeguards for safeguards for sensitive transactions sensitive transactions sensitive transactions like like like funds transfers, are designed to funds transfers, are designed to funds transfers, are designed to help you through this help you through this help you through this process. Next: you process. Next: you process. Next: you need to need to need to develop, develop, develop, sell, and sell, and sell, and improve improve improve the product simultaneously. The old the product simultaneously. The old the product simultaneously. The old approach—building something for a long time approach—building something for a long time approach—building something for a long time before before before finding your first finding your first finding your first customer—is simply customer—is simply customer—is simply too slow too slow too slow for the reality for the reality for the reality we live in.

we live in. we live in. The best teams The best teams The best teams do this do this do this in parallel. And finally in parallel. And finally : when the choice arises : when the choice arises : when the choice arises between “develop it between “develop it between “develop it yourself” or “buy it yourself” or “buy it yourself” or “buy it ready-made”, leading AI ready-made”, leading AI companies choose companies choose companies choose both. They both. They both. They create what they create what they create what they consider their consider their consider their competitive competitive competitive advantage and buy the advantage and buy the advantage and buy the basic components basic components basic components needed to do so. needed to do so. So, building quickly is one thing, but where is one thing, but where is one thing, but where to sell is just as to sell is just as to sell is just as important. So let's important. So let's important. So let's talk about what it talk about what it talk about what it means to sell in a means to sell in a means to sell in a global default market global default market global default market . . . Previously, SaaS companies Previously, SaaS companies Previously, SaaS companies focused on the focused on the focused on the domestic market, and domestic market, and domestic market, and as they became as they became as they became more confident, more confident, more confident, they began to expand into they began to expand into they began to expand into other markets. You could other markets. You could other markets. You could hire a general hire a general hire a general manager in London, manager in London, manager in London, open a new office— open a new office— and voila, you're an and voila, you're an and voila, you're an international company. international company. international company. But that doesn't But that doesn't But that doesn't work anymore and is too work anymore and is too work anymore and is too slow for what slow for what slow for what we see from leading we see from leading we see from leading AI companies. With the old AI companies. With the old AI companies. With the old SaaS model, you could SaaS model, you could SaaS model, you could reach 25 countries in the reach 25 countries in the reach 25 countries in the first year and 50 first year and 50 first year and 50 countries by the third countries by the third countries by the third year. But our year. But our year. But our leading AI companies leading AI companies leading AI companies exceed these exceed these exceed these indicators many times over. In the indicators many times over. In the indicators many times over. In the first year, they first year, they first year, they sell in 48 countries, sell in 48 countries, sell in 48 countries, and by the third year, they sell and by the third year, they sell in 120 different in 120 different in 120 different countries. And it's not countries. And it's not countries. And it's not just about just about just about presence in these presence in these presence in these countries, but about countries, but about countries, but about real real real solvent solvent solvent clients. Gamma is a clients. Gamma is a great example of great example of great example of this. Perhaps. That's right this. Perhaps. That's right this. Perhaps. That's right . Let me . Let me . Let me tell you about them.

tell you about them. tell you about them. It's an AI-based service It's an AI-based service It's an AI-based service for creating for creating for creating slides and websites slides and websites slides and websites that generated over that generated over that generated over $100 million in $100 million in $100 million in revenue in its first year. revenue in its first year. Although they are based in Although they are based in San Francisco, most of San Francisco, most of San Francisco, most of their their their revenue comes from revenue comes from revenue comes from outside the United outside the United outside the United States. So, we States. So, we States. So, we see that they are see that they are see that they are no longer the exception. no longer the exception. no longer the exception. In fact, this is the rule of In fact, this is the rule of In fact, this is the rule of how how how leading AI companies develop. leading AI companies develop. Maybe the slides Maybe the slides will come back. But I will come back. But I will come back. But I will continue to speak anyway will continue to speak anyway will continue to speak anyway . We . We . We see that in these see that in these see that in these leading AI leading AI companies, 48% of revenue companies, 48% of revenue companies, 48% of revenue comes from markets comes from markets comes from markets outside their outside their outside their country. That's country. That's country. That's almost half of almost half of almost half of every dollar that every dollar that every dollar that leading leading leading AI companies earn. So what AI companies earn. So what AI companies earn. So what we're really seeing we're really seeing we're really seeing now, with that now, with that now, with that number up from 33% number up from 33% three years ago, is that three years ago, is that three years ago, is that global revenue global revenue global revenue is no longer a bonus is no longer a bonus is no longer a bonus to your business to your business model. It actually model. It actually model. It actually becomes becomes becomes critical to how critical to how critical to how your business makes your business makes your business makes money. How are we doing money. How are we doing money. How are we doing with the slides? Perhaps. with the slides? Perhaps. OK. Good. So I OK. Good. So I will continue the will continue the will continue the story. Um, story. Um, story. Um, let's talk a little bit let's talk a little bit let's talk a little bit about the data about the data about the data we're seeing about we're seeing about we're seeing about where people where people where people are spending the most are spending the most are spending the most on AI at Stripe. I would on AI at Stripe. I would on AI at Stripe. I would show you a nice show you a nice show you a nice map, so just map, so just map, so just imagine it. Um, we imagine it. Um, we imagine it. Um, we see that the highest see that the highest see that the highest spending is spending is spending is in countries with the highest in countries with the highest in countries with the highest GDP, no surprise, GDP, no surprise, GDP, no surprise, right? These are countries like the right? These are countries like the United States, United States, United States, Japan, and Germany.

Japan, and Germany. Japan, and Germany. But what's more interesting is that But what's more interesting is that But what's more interesting is that new new new hotspots are emerging where people hotspots are emerging where people hotspots are emerging where people spend the most spend the most . These are countries like . These are countries like . These are countries like Brazil, India, and Brazil, India, and Brazil, India, and South Korea. So, South Korea. So, South Korea. So, it’s clear that it’s clear that it’s clear that there is a demand for AI there is a demand for AI services and services and services and tools in these countries, but how do you tools in these countries, but how do you tools in these countries, but how do you make sure you’re make sure you’re make sure you’re covering it? One covering it? One covering it? One way to do way to do way to do this is to meet this is to meet this is to meet your your your customers' expectations: sell in customers' expectations: sell in customers' expectations: sell in the currency they're the currency they're the currency they're used to and used to and used to and support the support the support the local payment local payment local payment methods they methods they methods they want to see. want to see. In fact, we saw that companies that that companies that that companies that list prices in list prices in list prices in local currency local currency local currency earn almost 18% earn almost 18% higher international higher international higher international revenue. And companies that revenue. And companies that revenue. And companies that offer and offer and offer and support support support at least one at least one at least one local payment local payment local payment method saw a method saw a method saw a conversion increase conversion increase conversion increase of over 7%. And agents are of over 7%. And agents are of over 7%. And agents are increasingly better at knowing increasingly better at knowing increasingly better at knowing how to do this with how to do this with how to do this with our Stripe our Stripe our Stripe skills. As you saw skills. As you saw skills. As you saw during the demo, the during the demo, the during the demo, the Stripe agent advised Stripe agent advised Stripe agent advised me to use me to use me to use Stripe Tax, but that's not Stripe Tax, but that's not Stripe Tax, but that's not all. Thanks to our all. Thanks to our all. Thanks to our skills, the agent knows to skills, the agent knows to support support local currency pricing local currency pricing local currency pricing and payment and payment and payment methods to help methods to help methods to help you optimize you optimize you optimize sales in these markets sales in these markets . You no longer . You no longer . You no longer need to be need to be need to be an expert in the markets an expert in the markets an expert in the markets where you sell, because Stripe where you sell, because Stripe where you sell, because Stripe already is and can help you with already is and can help you with already is and can help you with that. Here are that. Here are that. Here are a few things to a few things to a few things to think about to think about to think about to check your check your check your global readiness global readiness . First, . First, . First, localize your localize your localize your prices and payment prices and payment prices and payment methods to the markets methods to the markets methods to the markets you intend to you intend to you intend to sell in. Secondly, sell in. Secondly, sell in. Secondly, automate automate automate tax collection, because tax collection, because tax collection, because no one wants to have problems

no one wants to have problems no one wants to have problems due to ignorance of due to ignorance of due to ignorance of tax rules. tax rules. tax rules. Finally, track Finally, track Finally, track your revenue and your revenue and your revenue and conversion rates conversion rates conversion rates by country and by country and optimize them relentlessly. Okay , so let's , so let's , so let's talk a little bit about talk a little bit about talk a little bit about how how how pricing is changing for the pricing is changing for the pricing is changing for the AI ​​economy, because AI ​​economy, because AI ​​economy, because the models we're the models we're the models we're used to no longer used to no longer used to no longer work. We are used to work. We are used to work. We are used to thinking about the value of thinking about the value of software as a software as a tool. You can tool. You can tool. You can spend 10k on spend 10k on spend 10k on Zendesk and then hire Zendesk and then hire support reps to support reps to actually handle actually handle actually handle incoming incoming incoming customer requests. But AI is customer requests. But AI is customer requests. But AI is fundamentally changing fundamentally changing fundamentally changing our understanding of this our understanding of this our understanding of this value. As value. As value. As products acquire products acquire products acquire agentic capabilities agentic capabilities , we can see how , we can see how , we can see how they provide they provide they provide value and perceive value and perceive value and perceive this value as the this value as the this value as the result of their result of their result of their work. For example, Finn work. For example, Finn work. For example, Finn from Intercom personally from Intercom personally from Intercom personally manages and resolves manages and resolves manages and resolves customer inquiries for customer inquiries for customer inquiries for you. So when we you. So when we you. So when we start thinking about start thinking about start thinking about this new model, this new model, this new model, value becomes value becomes value becomes measurable: it can be measurable: it can be measurable: it can be counted through counted through counted through issues resolved, issues resolved, issues resolved, commits, or commits, or commits, or workflows completed. And workflows completed. And workflows completed. And then we are forced and then we are forced and then we are forced and encouraged to think about encouraged to think about encouraged to think about how to price our how to price our how to price our products to truly products to truly products to truly reflect that reflect that reflect that value. And thoughts value. And thoughts value. And thoughts about reassessment are not about reassessment are not about reassessment are not new. We did new. We did new. We did this every time there this every time there this every time there was a major was a major was a major technological shift.

technological shift. technological shift. We started with We started with We started with on-premises on-premises software, where everything software, where everything was pretty was pretty was pretty stable after stable after stable after installation, so a installation, so a installation, so a one-time license one-time license one-time license made sense. But then made sense. But then made sense. But then we moved to we moved to we moved to cloud solutions. And cloud solutions. And cloud solutions. And your software your software was was constantly updated. Therefore, the constantly updated. Therefore, the constantly updated. Therefore, the subscription model began subscription model began subscription model began to meet the demands of the to meet the demands of the to meet the demands of the time. And right now you time. And right now you time. And right now you expect me to tell expect me to tell expect me to tell you which model is you which model is you which model is right for AI. right for AI. right for AI. So I have good news and So I have good news and So I have good news and bad news. The bad bad news. The bad bad news. The bad news is that we are news is that we are news is that we are only at the beginning of the journey only at the beginning of the journey . We are still looking for . We are still looking for . We are still looking for answers. But the good answers. But the good answers. But the good news is that we news is that we news is that we see some see some see some patterns in patterns in patterns in our Stripe data. our Stripe data. our Stripe data. Take Take Take Replit, a Replit, a developer platform that has developer platform that has been around for almost a been around for almost a been around for almost a decade, for example. When decade, for example. When decade, for example. When agent agent agent programming began to take off programming began to take off programming began to take off , , , they changed their they changed their they changed their product focus, product focus, product focus, revised their revised their revised their pricing, and have pricing, and have been experiencing been experiencing explosive growth ever since. explosive growth ever since. They started with a They started with a fixed fixed fixed subscription model and then subscription model and then subscription model and then added credits as added credits as added credits as AI became the core of AI became the core of AI became the core of their product. their product. their product. Subscriptions are for Subscriptions are for Subscriptions are for predictability for predictability for predictability for users, and users, and users, and usage is for usage is for usage is for value. value. value. The result is that The result is that The result is that Replit is aiming to reach Replit is aiming to reach Replit is aiming to reach $1 billion in $1 billion in $1 billion in annual revenue this annual revenue this annual revenue this year. And they're not the only ones year. And they're not the only ones year. And they're not the only ones who think about who think about who think about pricing this pricing this pricing this way. Pay- way. Pay- way. Pay- per- per- per- use is becoming use is becoming use is becoming critically important critically important critically important for anyone for anyone for anyone selling AI services.

selling AI services. selling AI services. In fact, two out of three In fact, two out of three In fact, two out of three companies on the companies on the companies on the Forbes AI 50 list have some Forbes AI 50 list have some Forbes AI 50 list have some form of form of usage-based pricing, and usage-based pricing, and many offer the many offer the many offer the same hybrid same hybrid same hybrid model as Replit. The model as Replit. The model as Replit. The subscription component subscription component subscription component helps solidify the helps solidify the helps solidify the customer relationship customer relationship customer relationship and provide and provide and provide predictability, while predictability, while predictability, while pay-per- pay-per- pay-per- use use use helps helps helps scale scale scale with the customer's needs with the customer's needs with the customer's needs as they grow. as they grow. So, knowing that So, knowing that you need to switch you need to switch you need to switch to pay-per- to pay-per- to pay-per- use is use is use is one thing, but what does that one thing, but what does that one thing, but what does that mean in practice? mean in practice? mean in practice? Here are a few principles to Here are a few principles to consider. First , evaluate your , evaluate your , evaluate your product in your product in your product in your customer's language. Developers customer's language. Developers customer's language. Developers think in tokens, think in tokens, think in tokens, businesses think in businesses think in jobs, but jobs, but jobs, but more and more people more and more people more and more people are getting used to the are getting used to the are getting used to the consumption model. So consumption model. So consumption model. So think about how you think about how you think about how you can convey can convey can convey the value the value the value your your your users receive through the users receive through the users receive through the way you way you way you price. Next, you should price. Next, you should price. Next, you should give your give your give your customers the ability customers the ability customers the ability to see their costs in to see their costs in to see their costs in real time before they real time before they real time before they receive an invoice. No receive an invoice. No receive an invoice. No one has ever opened an one has ever opened an one has ever opened an electricity bill electricity bill electricity bill and thought, "Oh, that's and thought, "Oh, that's and thought, "Oh, that's cheaper than I cheaper than I cheaper than I expected." And you definitely expected." And you definitely expected." And you definitely don't want don't want your your product users to feel that way. So product users to feel that way. So product users to feel that way. So make sure you make sure you make sure you report report report their consumption in their consumption in their consumption in real time so real time so real time so they understand the they understand the they understand the final cost. This is final cost. This is final cost. This is n't just a nice n't just a nice n't just a nice addition, it's addition, it's addition, it's critical to critical to critical to preventing preventing preventing user churn.

user churn. user churn. Finally, think Finally, think Finally, think about selling credits, about selling credits, about selling credits, not spending them. not spending them. Offer Offer users to users to users to exchange exchange exchange money for credits in advance so that money for credits in advance so that money for credits in advance so that when working with when working with when working with the product, they the product, they the product, they think about the think about the think about the value received, not about value received, not about value received, not about every every every penny spent. Okay, penny spent. Okay, penny spent. Okay, let's look at another let's look at another let's look at another trend. This is how trend. This is how trend. This is how leading AI companies leading AI companies leading AI companies are adapting to new are adapting to new are adapting to new go-to- go-to- go-to- market strategies. The old model market strategies. The old model market strategies. The old model was that you was that you was that you built a product built a product built a product based on PLG and based on PLG and based on PLG and expected expected expected companies and startups companies and startups companies and startups to find it to find it to find it organically. Then, a organically. Then, a organically. Then, a few years later, few years later, few years later, after achieving after achieving product-market fit, you product-market fit, you would move into would move into would move into sales through the sales through the sales through the sales department. But we sales department. But we sales department. But we see that this process is see that this process is see that this process is too slow too slow too slow for the way for the way for the way new companies are built today new companies are built today new companies are built today . Many . Many . Many companies actually companies actually companies actually start selling start selling start selling to the corporate to the corporate to the corporate sector in their first sector in their first sector in their first year of existence. Cursor, year of existence. Cursor, year of existence. Cursor, for example, launched in for example, launched in for example, launched in 2023 as a 2023 as a 2023 as a self-service model self-service model self-service model for developers. And for developers. And for developers. And they quickly they quickly they quickly realized that they realized that they realized that they needed to needed to needed to implement direct implement direct implement direct sales to secure sales to secure sales to secure corporate corporate corporate contracts. Cursor contracts. Cursor contracts. Cursor has now built an impressive has now built an impressive has now built an impressive sales system that would have sales system that would have sales system that would have previously taken companies previously taken companies previously taken companies months, months, months, if not years, to build. And this is not the if not years, to build. And this is not the if not years, to build. And this is not the only shift we are only shift we are only shift we are seeing. Yes, seeing. Yes, seeing. Yes, you are starting direct you are starting direct you are starting direct sales much sales much sales much earlier, but we are also earlier, but we are also earlier, but we are also seeing the emergence of new seeing the emergence of new seeing the emergence of new strategies. For example, strategies. For example, strategies. For example, affiliate sales affiliate sales affiliate sales are becoming a much are becoming a much are becoming a much bigger part of bigger part of bigger part of the equation.

the equation. Users purchase Users purchase Anthropic and OpenAI services Anthropic and OpenAI services Anthropic and OpenAI services through their cloud through their cloud through their cloud providers, and many providers, and many providers, and many AI companies AI companies AI companies are implementing are implementing are implementing marketplaces so marketplaces so marketplaces so others can do the others can do the others can do the same. And there's another same. And there's another same. And there's another billion-dollar, maybe billion-dollar, maybe billion-dollar, maybe trillion-dollar question: trillion-dollar question: trillion-dollar question: how do we adapt how do we adapt how do we adapt to the new buyer- to the new buyer- agents? We agents? We have been studying human have been studying human psychology in psychology in product pricing for decades. For example, product pricing for decades. For example, prices of 9.99 instead of 10, or prices of 9.99 instead of 10, or prices of 9.99 instead of 10, or “good-better- “good-better- best” systems. But the thing is, best” systems. But the thing is, best” systems. But the thing is, the agents don't the agents don't the agents don't care about all these care about all these care about all these tricks. This graph tricks. This graph tricks. This graph probably probably probably illustrates this best. In 2025, we illustrates this best. In 2025, we illustrates this best. In 2025, we saw saw saw agent traffic to agent traffic to agent traffic to Stripe documentation Stripe documentation Stripe documentation increase 10x in one increase 10x in one increase 10x in one year. This graph year. This graph year. This graph shows shows shows human traffic and human traffic and human traffic and agent traffic to agent traffic to agent traffic to documentation, and as you documentation, and as you documentation, and as you can see, these lines are can see, these lines are can see, these lines are getting closer together. We getting closer together. We getting closer together. We expect that by the end of expect that by the end of expect that by the end of 2026, agents 2026, agents 2026, agents will be reading significantly will be reading significantly will be reading significantly more documentation more documentation more documentation than humans, than humans, than humans, using it as a using it as a using it as a discovery mechanism for discovery mechanism for discovery mechanism for our products. Here's the our products. Here's the our products. Here's the part that part that part that most people most people most people underestimate. When you underestimate. When you underestimate. When you start adapting start adapting start adapting new new new go-to-market strategies, go-to-market strategies, go-to-market strategies, monetization is just the monetization is just the monetization is just the tip of the iceberg. There is a tip of the iceberg. There is a tip of the iceberg. There is a whole layer that whole layer that whole layer that needs to be changed underneath needs to be changed underneath needs to be changed underneath it. Every time you it. Every time you it. Every time you implement a new implement a new implement a new approach, you have to approach, you have to approach, you have to adapt your adapt your adapt your product. You need to product. You need to product. You need to think about how think about how onboarding will now change for these onboarding will now change for these new channels. Your new channels. Your new channels. Your pricing model pricing model pricing model needs to be adjusted needs to be adjusted , and your organization , and your organization , and your organization needs to be ready for needs to be ready for needs to be ready for it. Supporting a it. Supporting a it. Supporting a corporate corporate corporate customer is significantly customer is significantly customer is significantly different from different from different from supporting the mass supporting the mass self-service segment.

self-service segment. So, a new approach to So, a new approach to So, a new approach to market entry market entry market entry really means really means really means changing your company's basic business changing your company's basic business model model model . Another . Another . Another difference is difference is difference is how companies how companies how companies actually build actually build revenue infrastructure. They used to revenue infrastructure. They used to just add just add just add individual elements, and individual elements, and individual elements, and when you moved when you moved when you moved slowly, that was slowly, that was slowly, that was acceptable. When you acceptable. When you acceptable. When you start compressing 10 start compressing 10 start compressing 10 years of years of years of infrastructure development into infrastructure development into infrastructure development into a few months, the seams a few months, the seams a few months, the seams between these between these between these tools become tools become tools become very visible and very visible and very visible and become a become a become a problem. When thinking about problem. When thinking about problem. When thinking about customers customers customers moving between moving between moving between different channels, there are different channels, there are a few things to consider for a few things to consider for your your your go-to-market strategies. You go-to-market strategies. You go-to-market strategies. You need to clearly need to clearly need to clearly understand how people understand how people understand how people move from move from move from self-service self-service self-service to enterprise to enterprise to enterprise level and how the level and how the level and how the pricing pricing pricing model changes as a result. You should model changes as a result. You should model changes as a result. You should build unified build unified build unified business systems. This business systems. This business systems. This means a single means a single means a single customer profile, a customer profile, a customer profile, a single single single product catalog, and product catalog, and product catalog, and unified unified unified data models for all these data models for all these data models for all these services. And finally, services. And finally, services. And finally, you need to you need to you need to think about how you think about how you think about how you design interactions design interactions design interactions with agents as with agents as with agents as buyers. The agent buyers. The agent buyers. The agent should be able to should be able to should be able to discover, evaluate, and discover, evaluate, and discover, evaluate, and activate your activate your activate your product without product without product without human intervention. So, we human intervention. So, we human intervention. So, we discussed how discussed how discussed how leading AI companies leading AI companies leading AI companies today are today are today are building, selling, building, selling, building, selling, pricing, and pricing, and pricing, and adapting adapting adapting go-to-market strategies differently.

go-to-market strategies differently. go-to-market strategies differently. They build faster, in They build faster, in They build faster, in hours or days, hours or days, hours or days, sell in 42 countries sell in 42 countries sell in 42 countries from day one, from day one, from day one, price for price for price for value, and start value, and start value, and start corporate sales corporate sales corporate sales before raising before raising before raising Series A investment. Series A investment. Series A investment. This is an incredibly This is an incredibly This is an incredibly exciting time for exciting time for exciting time for developers. My developers. My developers. My team at Stripe team at Stripe team at Stripe is focused on making sure is focused on making sure is focused on making sure we build the we build the we build the right right right tools to tools to tools to help you help you help you monetize and monetize and monetize and build faster. So build faster. So , if you have any feedback , if you have any feedback , if you have any feedback or questions about or questions about or questions about how we can how we can how we can better support you better support you better support you on this journey, we on this journey, we on this journey, we 'd love to 'd love to 'd love to chat. chat. chat. Thank you very much.