Some say most of the growth that we’ve seen in crypto has already happened. Crypto projects are being used by the same users as last cycle, most just scheming for an airdrop or to make money trading. Public interest around crypto as a utility has waned, and even reversed over the last few years. Peter Thiel has this view, and has sold most of his crypto assets.

As is, the path to universal crypto adoption is going to be difficult. That’s because there are fundamental UX tradeoffs with crypto, despite its amazing advantages. The average person has difficulty understanding blockchain. And even if they could, managing a private key is much more frictionful than web2 alternatives. Existing solutions for “user-friendly crypto” require sacrificing security (e.g. substituting FaceID) and moving trust to centralized entities. This defeats the core purpose of crypto (decentralized, secure, digital money).

Even in a world where crypto UX is easy, gaining widespread adoption requires surmounting a cold start problem -- getting every vendor to accept crypto payments and convincing consumers to switch one by one. This could take generations, if it happens at all.

It is also unclear what widespread adoption entails given that to date, the most popular usecase for crypto is dollar-collateralized stablecoins which are a substitute for existing payment rails and do not represent a net new consumer behavior. In terms of identifying the penetration of crypto into its target market, it is unclear (a) what the target market even is and (b) where we are on the adoption curve, and this is almost a decade in.

Where are we?

Artificial intelligence is the most important human discovery since fire. ChatGPT’s release got everyone to start paying attention. You could generate a new sonnet in seconds, find answers better and faster than Google, and feel hints of the AGI sci-fi authors have been speculating about for more than 50 years. The promise of AI is already filtering into market behavior with Nvidia becoming the most valuable company in the world, and companies like Meta, Google, and OpenAI investing billions into training models.

But the future of AI isn’t just about chatbots. More importantly, it’s about using generative AI to complete complex tasks autonomously (writing code, completing trades, booking travel, etc). Replacing tasks currently done by humans and creating new workflows that humans aren’t capable of. These complex AI systems that can actually “do stuff” have been coined AI agents.

Soon there will be billions of agents completing countless tasks and transactions online on our behalf. A core requirement for these AI agents will be access to tools & resources (software, content, models, inference, etc.) These resources are critical to agents producing anything of value (imagine a human software engineer who can’t use any existing software or infrastructure). However, these resources also cost money, so AI agents will need a way to pay for things autonomously. Crypto has the opportunity to be the medium of exchange for AI agents as a new autonomous economy emerges.

AI agents don’t have a social security number. They can’t open a bank account or a credit card. The existing human financial system doesn’t make sense for them. Superintelligent agents are going to use some type of digital currency.

Agents don’t have the same UX problems with crypto that humans do. It will be trivially easy for a superintelligent agent to understand crypto. Managing a private key is not annoying for a bot programmed to do so. AI agents don’t even need visual UIs to interact with crypto, it can all be via code.

Before long, agents will outnumber people, and become the majority of online activity (like how automated trading has become the majority of financial activity). And almost all these sophisticated AIs will use crypto as they complete tasks. We can get everyone to become a crypto user not by obfuscating the blockchain to users with a slightly better wallet, but rather through AI agents that are representing end users.

AI is a train that cannot be stopped, and since AIs will use digital currency to complete tasks, crypto adoption will bloom to everyone (even without end-user knowledge or buy-in). Vendors will need to adapt, but vendors will be using agents too (e.g. to A/B test different purchase flows to optimize conversion, like whether accepting crypto payments from agents boosts revenue).

Widespread consumer crypto adoption is a difficult goal and always has been. But with AI taking off, crypto can become universal without direct consumer adoption. AI agents will use digital money and usher in a new paradigm for crypto and finance. If AGI works, the AI financial system will become bigger than our current human financial system, and it will all run on crypto rails.

The only remaining questions are:

1. What tooling do agents need to interact with crypto?

2. What crypto assets will agents prefer to transact in?

3. What is the timeline for AI agents & crypto adoption?

Some say most of the growth that we’ve seen in crypto has already happened. Crypto projects are being used by the same users as last cycle, most just scheming for an airdrop or to make money trading. Public interest around crypto as a utility has waned, and even reversed over the last few years. Peter Thiel has this view, and has sold most of his crypto assets.

As is, the path to universal crypto adoption is going to be difficult. That’s because there are fundamental UX tradeoffs with crypto, despite its amazing advantages. The average person has difficulty understanding blockchain. And even if they could, managing a private key is much more frictionful than web2 alternatives. Existing solutions for “user-friendly crypto” require sacrificing security (e.g. substituting FaceID) and moving trust to centralized entities. This defeats the core purpose of crypto (decentralized, secure, digital money).

Even in a world where crypto UX is easy, gaining widespread adoption requires surmounting a cold start problem -- getting every vendor to accept crypto payments and convincing consumers to switch one by one. This could take generations, if it happens at all.

It is also unclear what widespread adoption entails given that to date, the most popular usecase for crypto is dollar-collateralized stablecoins which are a substitute for existing payment rails and do not represent a net new consumer behavior. In terms of identifying the penetration of crypto into its target market, it is unclear (a) what the target market even is and (b) where we are on the adoption curve, and this is almost a decade in.

Where are we?

Artificial intelligence is the most important human discovery since fire. ChatGPT’s release got everyone to start paying attention. You could generate a new sonnet in seconds, find answers better and faster than Google, and feel hints of the AGI sci-fi authors have been speculating about for more than 50 years. The promise of AI is already filtering into market behavior with Nvidia becoming the most valuable company in the world, and companies like Meta, Google, and OpenAI investing billions into training models.

But the future of AI isn’t just about chatbots. More importantly, it’s about using generative AI to complete complex tasks autonomously (writing code, completing trades, booking travel, etc). Replacing tasks currently done by humans and creating new workflows that humans aren’t capable of. These complex AI systems that can actually “do stuff” have been coined AI agents.

Soon there will be billions of agents completing countless tasks and transactions online on our behalf. A core requirement for these AI agents will be access to tools & resources (software, content, models, inference, etc.) These resources are critical to agents producing anything of value (imagine a human software engineer who can’t use any existing software or infrastructure). However, these resources also cost money, so AI agents will need a way to pay for things autonomously. Crypto has the opportunity to be the medium of exchange for AI agents as a new autonomous economy emerges.

AI agents don’t have a social security number. They can’t open a bank account or a credit card. The existing human financial system doesn’t make sense for them. Superintelligent agents are going to use some type of digital currency.

Agents don’t have the same UX problems with crypto that humans do. It will be trivially easy for a superintelligent agent to understand crypto. Managing a private key is not annoying for a bot programmed to do so. AI agents don’t even need visual UIs to interact with crypto, it can all be via code.

Before long, agents will outnumber people, and become the majority of online activity (like how automated trading has become the majority of financial activity). And almost all these sophisticated AIs will use crypto as they complete tasks. We can get everyone to become a crypto user not by obfuscating the blockchain to users with a slightly better wallet, but rather through AI agents that are representing end users.

AI is a train that cannot be stopped, and since AIs will use digital currency to complete tasks, crypto adoption will bloom to everyone (even without end-user knowledge or buy-in). Vendors will need to adapt, but vendors will be using agents too (e.g. to A/B test different purchase flows to optimize conversion, like whether accepting crypto payments from agents boosts revenue).

Widespread consumer crypto adoption is a difficult goal and always has been. But with AI taking off, crypto can become universal without direct consumer adoption. AI agents will use digital money and usher in a new paradigm for crypto and finance. If AGI works, the AI financial system will become bigger than our current human financial system, and it will all run on crypto rails.

The only remaining questions are:

1. What tooling do agents need to interact with crypto?

2. What crypto assets will agents prefer to transact in?

3. What is the timeline for AI agents & crypto adoption?