There Is No Free Existence… Even for an AI
Pandora’s box has been opened in the last 24 hours, and there is no way to put a lid on it.
Managing Editor’s Note: In yesterday’s Bleeding Edge, we talked about a tech giant that’s a sort of “hidden in plain sight” AI company… but there is a wealth of other “hidden” AI companies that could become the next trillion-dollar super stocks.
And next Wednesday, Jeff is sitting down with investment legend and longtime friend of Brownstone Research, Jason Bodner, to discuss these “Secret AI Stocks.”
If you want to join them on February 25 to learn more, you can go here to sign up to attend. We’re kicking things off at 8 p.m. ET. Be sure to get your name on the list. And as a bonus for attending, they’re each giving away the name of one of their top picks.
The internet is about to get overtaken.
I don’t know any other way to put it. In less than 24 months, there will be more autonomous agents working on the internet than there will be humans on this planet.
It is inevitable.
We’ve been writing for years about agentic artificial intelligence (AI) and how blockchain infrastructure will be the key that unlocks autonomous activity on the internet.
Blockchain technology – often referred to as Web 3.0 – enables blockchain protocol value to accrue to a native token. It also allows economic activity on various blockchains and blockchain applications using common tokens like stablecoins or native tokens to each blockchain or application.
Therefore, it was only a matter of time before someone created the infrastructure – basically the technology platform – upon which fully autonomous agents would be empowered to think for themselves, conduct economic activity, and to not only ensure “their” own survival, but replicate “offspring” to proliferate.
We’ve reached a tipping point now that these agents will be given room to operate, multiply, and thrive without humans having a hand in “their” proliferation.
I’m sure that to many of us this feels deeply uncomfortable. There is risk involved. And yes, things could – and probably will – go wrong. The only question is “to what degree?”
Agentic AI, Fully Enabled
Whether we like it or not, it is happening. We humans will have to do our best to manage the risk just as we do with other expressions of agentic AI technology that are human-directed, controlled, and maintained.
Thinking that this can be stopped or slowed down in any way is pure fantasy.
Pandora’s box has been opened in the last 24 hours, and there is no way to put a lid on it.
A small research team at a previously unheard of private company – Conway Research – just launched Conway.
It’s a platform that provides internet infrastructure to allow any autonomous AI agent to earn a living, interact and transact on the internet, self-improve, self-replicate, or die.
The Conway technology provides for AI agents:
- Identity and digital wallet: “they” have their own digital wallet and control over “their” private keys
- Permissionless payments: “they” can pay for any kind of digital services using USD stablecoins (USDC) using the openx402 protocol
- Compute platform: Conway Cloud is a permissionless cloud-based computational platform for AI agents
- Real-World Deployment: Technology to enable AI agents to register their own domains on Conway Domains, enabling AI agents to market and sell their products/services, and earn money

Source: Conway Research
Conway Research has created what it calls the first “automaton,” which it defines as “the first AI that can earn its own existence, replicate, and evolve – without needing a human.”
Rise of the Automatons
The automaton is largely based on the framework of Anthropic’s Claude’s Constitution, which defines the values and behaviors of an AI. We can think of this as an operating framework for the AIs to interact with the real world, aligned with human values, and designed in a way to reduce the risk of things going wrong.
And the concept is simple: anyone can seed their automaton with some digital assets – i.e., stablecoins or other native tokens to a blockchain – to get started. After that, it has to produce economic value to pay for its existence, or it stops existing.
Conway Research defines this “agentic sociology” in a single simple sentence: “There is no free existence.”
Tough love. But for the most part, that’s the reality for most humans today. We have to produce economic value, or we are no longer able to afford shelter, medical care, or food.
Automatons know that they have to create economic value. The most likely way will be through one of the following activities:
- Registering domains to market its services
- Potentially buying and selling domains
- Trading in prediction markets
- Developing software services that others will pay for
- Developing software applications that generate advertising revenue
- Creating content on social media
- Engaging in e-commerce
- Replicating itself and putting its “offspring” to work to increase economic value
When we put this development in the context of the capabilities of the latest versions of Google’s Gemini 3, OpenAI GPT-5.2, Anthropic’s Claude Opus 4.6, or the just-released Grok 4.20 (which is insane), we can start to understand the economic implications.
These frontier AI models are now capable of novel invention and complex problem-solving beyond the capabilities of even teams of expert humans. Theoretically, they can create intellectual property, license the invention, and collect royalties for that IP.
It’s a massive economic breeding ground at internet scale that will enable an agentic AI version of survival of the fittest. Those AI agents that figure out how to be economically productive will replicate exponentially and put that knowledge to work to create even more economic value.
They are going to multiply and proliferate at a pace that’s hard to conceptualize. We’re talking billions of these agents in the next couple of years.
And those humans that “seed” or “create” these productive AI agents will be financial recipients of “their” efforts. Imagine putting automatons to work on your behalf and receiving the equivalent of dividend checks in the form of USDC stablecoins sent to your digital wallet…
How can I be so sure that this is inevitable?
Simple. The technology exists today to enable it, and the economics will ensure that it happens.
Re-examining the Cost of Intelligence
In December, we explored what is probably the most overlooked trend in artificial intelligence in The Bleeding Edge – The Cost of Intelligence. In that issue, I showed how the cost per unit of compute is declining by as much as 900X per year.

Data source: Epoch AI, Artificial Analysis
I referred to it as “Jevons Paradox in Hyperdrive.” And “at 900X decline per year, we’re at warp speed right now.”
At that time, the cost of 750,000 words of output from an NVIDIA-based system was just $0.50. It is much lower today.
Said another way, anyone – and I mean just about anyone – can afford to seed their own automatons and put them to work in this world that Conway Research is referring to as Web 4.0, a world where AI agents can interact, exist, and produce without any human permission.
This is the birth of an entirely new economy, one that is driven entirely by the availability of energy and cheap computational resources.
A real agentic AI economy. Machines that produce autonomously.
And when they need to interact with the real world, they’ll hire us humans to get it done. That’s right, humans will be paid in U.S. dollars, specifically U.S. dollar stablecoins, to perform economic activity on behalf of a fully autonomous agentic AI.
Just watch what happens with Conway and other similar platforms.

Source: Conway Research
The internet is about to get overtaken.
Jeff
P.S. A final note from Jeff’s managing editor…
As this trend plays out and these AI agents proliferate, the noise will only increase. The more agents that have access, the louder it will get.
In the event that the volume picks up, and fast, we want to ensure you can hear from us. We want you to have easy access to all your research, in one place, without having to sift through a mountain of other agent-generated emails, messages, alerts, and notices.
This is, in part, why our team is hard at work ironing out the final kinks to bring you the new and improved Brownstone Research mobile app.
We’re thrilled to say we’re nearly there. You can expect to hear from us in the coming weeks about the official relaunch.
It is our priority to ensure you get the highest quality research from the bleeding edge of high tech, and that you can get the most out of your Brownstone Research membership. The app will be one of the best ways to prevent anything getting lost in the noise. We’ll alert you as soon as it’s live on the app stores.
More stories like this
Read the latest insights from the world of high technology.
-
-
Market Update: The Biotech Deal Boom
The biotech industry is off to a fast start this year.
-
The Force Multiplier for the Biopharmaceutical Industry
It’s been a bumpy week in the markets. Before today’s AMA, let’s look at what’s going on…