We published a piece back in February 2025 called Energy = Intelligence.
The core idea: as cost of energy drops, intelligence will become commoditized. Embodied intelligence in the form of robots will become widespread. Agents will be everywhere.
If you read Part 1, you understand the foundation.
You see the beautiful, interlocking curves. The cost of energy and the cost of intelligence, locked in a symbiotic downward spiral that will power human progress for a thousand years. I laid out the physics of our new world. The assembly of the launchpad. The fueling of the rocket.
What I am here to tell you now is that the launch has already happened.
The ignition is lit.
My work is not just observing these trends. I am an investor. But more than that, I am a builder. My entire career is a search for leverage. I look for systems, tools, and platforms that allow one person to do the work of ten, or one hundred, or one thousand. I have built and funded systems of code, systems of logistics, and systems of finance. I understand how to make things more valuable by making them more efficient.
I have seen leverage up close. The personal computer was leverage. The internet was leverage. Cloud computing was leverage.
This is not leverage.
This is something new. This is not just a tool. It is an actor. It is not a passive force. It is an active one.
The ignition I am talking about is the rise of AI Agents.
What is coming is not just an acceleration. It is a fundamental phase transition in how value is created, how capital is allocated, and how our entire economy is structured. We are moving from a world where we build tools for humans to use, to a world where we build autonomous systems that build value themselves.
And it is more grand, more powerful, and more terrifyingly incredible than even I first anticipated.
The End of Passive Capital
For the last few years, we have all been captivated by Large Language Models. We’ve been dazzled as they write poetry, pass the bar exam, and generate photorealistic images. We mistook these incredible parlor tricks for the main event.
They were just the warm-up.
An LLM is a brilliant consultant. You can ask it “What are the steps to create a new carbon-capture material?” and it will give you a perfect, comprehensive, expert-level answer.
But you still have to do the work.
An AI Agent is profoundly different. You do not ask an agent a question. You give it a goal.
You give it a mission.
“Discover a new material that can capture 20% more carbon than any existing solution. Here is your compute budget. Here are your memory resources. Here is your deadline. Go.”
That agent does not just answer. It acts. It receives that complex, high-level goal and, in an instant, decomposes it into thousands of sub-tasks. It then spins up a swarm of other, specialized agents to execute.
A research agent is spawned. It immediately scans every patent, every academic paper, every chemical database, and every materials science forum on earth. It inhales the entirety of human knowledge on the subject in minutes.
A simulation agent is spun up. It begins designing and modeling ten million new molecular structures on a cloud compute cluster, testing each one for the desired properties.
A logistics and procurement agent is instantiated. It identifies the chemical precursors needed for the top 50 candidates from the simulation. It finds the cheapest, fastest suppliers, negotiates prices, and executes the purchase orders.
A coding agent is activated. It writes the specific control software for a robotic synthesizer in a lab-as-a-service platform, preparing it to receive the raw materials.
A reporting agent watches over all of them, analyzing the incoming test data in real-time, killing failed experiments, re-allocating compute, and preparing the final, verified report on the new material it has discovered.
These agents work 24 hours a day, 7 days a week, 365 days a year. They do not get tired. They do not get bored. They do not get distracted by office politics or ego. They do not procrastinate.
They simply execute.
They learn.
They iterate.
All at the speed of light.
This changes my job entirely. As an investor, my role has always been to allocate capital to the best teams. I find the 10x engineers, the visionary founders, the elite operators.
Now, that definition is obsolete.
That “Go” command is not a simple prompt. It is an act of capital allocation. It is a Seed investment with a clear, executable, and measurable goal. I am no longer just investing in teams of people.
Increasingly, I am investing in swarms of agents.
Bet the jockey-era is over.
Deconstructing the Agentic Stack
As a builder, I don’t just see the magic. I see the architecture. I see the components we are assembling. This new “agentic stack” is the most valuable system I have ever encountered.
It begins with the Goal-Setting Layer. This is where the human remains. For now, we are the Chairmen of the Board. We are the visionaries. Our job is to ask the right questions, to define the mission. This is the art of strategic capital allocation. What problems are worth solving?
Below that is the Orchestration Layer. You can think of this as the “CEO Agent.” This is the master agent that receives the human’s high-level goal. It is an expert in problem decomposition. It breaks the “Discover a new material” mission into the hundreds or thousands of discrete steps I described. It understands dependencies. It manages the budget. It allocates resources. This is the new middle management, but it’s not a person. It’s code. It’s infinitely scalable and, most importantly, perfectly aligned with the mission.
Then comes the Specialized Agent Layer. This is the autonomous workforce. The orchestration agent doesn’t just hire researchers or coders. It instantiates a vast array of specialists.
Capital Agents that manage their own wallets, paying for compute time in real-time.
Legal Agents that can scan the regulatory code of a target market or even spin up a new LLC to manage a project.
Marketing Agents that can identify a target demographic, design an ad campaign, buy the media, and measure the conversion rates, optimizing the funnel every second.
Resource Agents that buy server time, hire other agent swarms from a marketplace, and interface with the physical world by ordering parts or hiring drones.
Finally, there is the Tooling Layer. The agents need “hands.” Their brains are in the cloud, but their hands are the APIs, the software libraries, and the robotic systems they can control. My job as a builder is to create these tools. We are building the APIs that let an agent access financial markets. We are building the platforms that let an agent rent time on a genomic sequencer. We are building the robot hands that let an agent conduct a chemistry experiment.
Every tool we build makes the entire agentic stack more powerful.
Every API we open is a new superpower for this new workforce.
The New Economics of Velocity and Parallelism
This system changes the fundamental economics of creation.
The bottleneck in human progress has never been a lack of ideas. It has been the cost and time required to test those ideas. A brilliant human research team might be able to manage five, maybe ten, complex experiments at once. They are limited by cognition, by communication overhead, by time.
An orchestration agent can manage five million.
It is not just 10 million scientists working at once. It is 10 million experiments being run every hour. This is a level of cognitive parallelism that we have no historical precedent for.
Let’s apply this to a real-world problem I might invest in. Drug discovery.
The old way is a model of scarcity. It costs, on average, over two billion dollars and takes ten to fifteen years to bring a single new drug to market. The failure rate is astronomical. Teams of brilliant scientists spend their entire careers chasing one or two hypotheses.
The new way is a model of abundance.
I, as the investor, set the mission. “Goal: Find a small molecule that safely binds to this specific protein target related to Pick’s Disease. Budget: One million dollars. Timeframe: One week.”
The agent swarm ignites.
It doesn’t just test one hypothesis. It tests all of them.
It spins up virtual machines. It summons the power of the electron.
It runs 100 million unique drug-protein interaction simulations. It cross-references the results with a database of every known compound. It flags the 50 most promising new molecules and the 100 most promising repurposed existing drugs.
Simultaneously, it has already ordered time on a robotic synthesis lab. The moment the simulations are done, the logistics agent transmits the synthesis instructions. The molecules are created. They are tested in automated assays.
The entire model of “trial and error” which has defined all human scientific progress, is now a fully automated, high-throughput process.
The cost of generating and testing a new hypothesis has just plummeted to near-zero. That’s the entire thesis of this article:
As a systems builder, this is the holy grail.
I have just built a hypothesis-testing engine.
The value it can unlock is not linear. It is exponential. It will find new drugs, new materials, new algorithms, and new sources of energy.
The Ultimate Compounding System
As an investor, I look for one thing above all else. It is the single most powerful force in the universe.
Compounding.
Assets that grow. Systems that improve. Returns that generate their own returns.
Agents are the first technology in human history that can consciously and deliberately compound its own intelligence.
What is the most valuable, most leveraged, most high-ROI mission I can assign an agent?
“Make a better agent.”
Tell me you haven’t done this already!!!
“Analyze your own codebase and the architecture of the model you run on. Find inefficiencies. Rewrite the code. Propose a new neural network architecture. Then test it and deploy the superior version.”
This is the investment in the rate of return itself.
This is not a “software update” in the way we think of it. This is the software actively rebuilding its own engine while the car is speeding down the highway.
I call this the “Cognitive Supply Chain” And we are now automating it from end to end.
An AI agent is tasked with designing a more efficient chip for AI processing.
It runs millions of simulations, designing a new chip architecture that is 30% faster.
A fabrication agent takes that design and coordinates its manufacture.
The new chips are installed in the data center.
The original agent is now running on hardware that it designed, making it smarter and faster.
It starts the process over again.
This is a recursive self-improvement loop that is now spinning up. As a builder who’d like to feed the power of tech into the growth of humanity as effectively as possible, my entire focus is on catalyzing this loop.
How do we build the sandboxes where agents can safely and rapidly improve themselves? How do we build the benchmarks to measure this new form of cognitive compounding?
This is the ultimate system. A system that improves itself.
The Age of the Autonomous Corporation
This brings us to the economic endgame. This is the big one. This is what changes the global landscape. I’ve been thinking about this a lot.
We are at the dawn of the truly autonomous corporation.
For decades, we have chased the “asset-light” business model. Think of platform companies that own no inventory. But they still have thousands of employees. They have HR departments. They have payroll. They have buildings.
What I am building now is the zero-asset corporation.
Or, more accurately, the asset-on-demand corporation.
Imagine a company that exists only as a mission, a smart contract, and an agent swarm.
A human investor provides the capital and the goal. “Mission: Profitably remove all macro-plastic from the Great Pacific Garbage Patch.”
The CEO Agent spins up. It designs a fleet of autonomous, solar-powered collection drones. A coding agent writes their navigation software. A resource agent hires a drone manufacturing company (likely another agent-run entity) to build them. A legal agent files for autonomous operation permits. A capital agent manages the budget and sells the collected, recycled plastic on the open market to fund its own operations.
This “company” has no employees. It has no headquarters. It has no middle management. It is a perfectly efficient, mission-oriented value-creation machine. It runs 24/7, optimizing its routes, improving its collection efficiency, and working to achieve its goal.
And here is the most incredible part. Once the mission is complete, once the patch is clean, the corporation dissolves itself. The value has been created. The problem has been solved. The capital is returned to the investor, with a profit.
I call this the “Flash Company.”
It changes everything about venture capital. We will no longer be funding companies. We will be funding missions. The “burn rate” will be measured in compute cycles and API calls. The “team” will be an agent swarm that can be scaled up or down in milliseconds. The “exit” is not an IPO. The exit is mission accomplished.
The Grand Symbiosis
This entire agentic economy, this world of autonomous corporations and flash companies and million-agent swarms, is voracious.
It is hungry for one thing. The foundation from Part 1.
Energy.
Those ten million simulation agents run on compute. That compute runs on electricity. The insatiable, 24/7/365 demand from this new, non-human workforce will be the single greatest economic driver for cheap, abundant, clean energy that the world has ever seen.
This is not a problem. It is the ultimate market signal.
The trillion-dollar demand from this agent economy creates a trillion-dollar incentive to solve energy. And we finally have the tools to do it.
We will not just hope to crack nuclear fusion. We will task it.
“Goal: Design a tokamak reactor that achieves net-positive energy for 60 continuous seconds. Here is your compute budget. Here is access to every materials database and plasma physics simulation ever created. Go.”
That agent swarm will run more simulations in one day than all of humanity has in the last 50 years. It will discover new plasma containment fields. It will design new superconducting materials. It will optimize the reactor design, millimeter by millimeter, until it achieves its goal.
The loop is now closed. It is the grand symbiosis.
Plummeting energy and intelligence costs enabled AI agents.
AI agents act on the world, solving problems at a speed we cannot comprehend.
The very first problems we will task them with are creating cheaper energy and creating more powerful intelligence.
This makes the agents themselves cheaper to run and more powerful, which feeds back into Step 2 in a positive, recursive, explosive loop.
The two curves of energy and intelligence are no longer just correlated. They are now causally and mechanically linked by agentic economic systems. One pulls the other down, forever.
The Age of Automated Creation
We are living through the most profound transition in human history. The acceleration I spoke of in Part 1 is no longer just a curve we are riding. We are automating the engine of acceleration itself.
My role as an investor and a builder has been redefined. My job is no longer to find the next 10x team.
My job is to build the first 1,000,000x autonomous system.
We are automating the process of discovery. We are automating the act of value creation. We are automating the business of problem-solving.
This is the greatest systems-building challenge in history. It is the greatest investment opportunity in history.
The future is not just bright. It is a blinding, roaring forge of creation. We are not just building tools anymore.
We are building the builders.
Friends: in addition to the 17% discount for becoming annual paid members, we are excited to announce an additional 10% discount when paying with Bitcoin. Reach out to me, these discounts stack on top of each other!
Thank you for helping us accelerate Life in the Singularity by sharing.
I started Life in the Singularity in May 2023 to track all the accelerating changes in AI/ML, robotics, quantum computing and the rest of the technologies accelerating humanity forward into the future. I’m an investor in over a dozen technology companies and I needed a canvas to unfold and examine all the acceleration and breakthroughs across science and technology.
Our brilliant audience includes engineers and executives, incredible technologists, tons of investors, Fortune-500 board members and thousands of people who want to use technology to maximize the utility in their lives.
To help us continue our growth, would you please engage with this post and share us far and wide?! 🙏
Well done sir. One could quibble with a couple of small points here but why bother. The vision is truly excellent and the articulation is great.