The Convergence Thesis
AI, Robotics, and Crypto are merging into a single machine economy.
This is the investment map.
Three Futures
Before we get into the thesis, let me show you where this is heading. These three scenarios are extrapolations — not predictions, but plausible destinations based on the trajectory we can already measure. The vocabulary may be unfamiliar. It will make sense by the time you finish reading.
The exponential curve hasn’t just bent. It has snapped vertical. In a cluttered dorm room overlooking the Charles River, an MIT sophomore is out-competing a global defense prime. He used a compute escrow account to rent a swarm of engineering agents, specified intent for an orbital debris removal system, and the agents swarmed the problem — writing the software, running the verification, generating a cryptographically signed replication pack proving the code is mathematically safe. A project that would have taken a government lab three years and fifty million dollars. He did it in four hours for the cost of a pizza.
Corporate boards are panic-firing CHROs and hiring Compute Portfolio Managers. Return on Cognitive Spend has replaced EBITDA as the primary signal of solvency. If you cannot prove that every dollar of electricity you burn is generating a verified unit of intelligence, you are functionally bankrupt.
The first Targeting Authority has gone live. It has posted a two-billion-dollar blinded bounty for a room-temperature superconductor, sitting in a smart contract visible to the world. University labs are emptying as researchers form flash organizations to chase the target.
The lock-in is here. Intelligence is no longer a craft practiced by artisans. It is the new electricity. The grid is live.
The physical world is beginning to liquefy. We have moved from mining materials to compiling them. In the Nevada desert, a massive Action Network hums — a closed-loop hive of robotic chemists that never sleep, iterating through chemical space at the speed of light, mixing, testing, analyzing, and refining new battery chemistries in a continuous blur of robotic arms.
Biology has capitulated. The human body has become a software problem. Pharma giants have dissolved into Bio-Fab utilities selling guaranteed health outcomes. You don’t buy a drug. You buy a subscription to “Normal Liver Function.” A patient in Tokyo walks into a clinic with a failing kidney and walks out three days later with a scheduled transplant of a printed, autologous replacement requiring no anti-rejection meds. The waiting list is gone. It was an inventory management error.
High-bandwidth non-invasive BCIs are common. They look like sleek headphones but allow early adopters to write thoughts directly into code. Architects design buildings by hallucinating structures the AI instantly renders. Musicians compose symphonies by feeling the sound. The bandwidth of thumbs and voice is too slow for 2030.
Food has decoupled from the weather. The price of protein has flatlined globally. Hunger is now recognized as a logistical error rather than a resource limit.
The screaming exponential of the twenties has settled into the terrifyingly efficient silence of a solved world. Longevity Escape Velocity has been breached — not in a moment of fanfare, but as a gradual statistical reality. For every year you survive, science adds more than a year to your clock. Aging is no longer a destiny. It is a manageable chronic condition, debugged by your personal health agent monitoring your proteome in real-time.
The social contract has been rewritten. We no longer distribute cash. We distribute capacity. Universal Basic Capability guarantees every citizen access to the best AI tutor, the best AI doctor, the best AI lawyer. These aren’t second-tier services. They are the best in the world, replicated infinitely at zero marginal cost.
Off-world, autonomous mining swarms on the Moon and asteroids feed the orbital shipyards. Heavy industry has moved to orbit and uncoupled Earth’s economy from its fragile biosphere. The weapon of targeted superintelligence is fully built. Now we just have to decide where to aim it.
These scenarios sound extreme. They are not. They are conservative extrapolations of trends already in motion. The question this essay answers is: How do you position yourself before the curve goes vertical?
The Intelligence Revolution
Every civilizational step-change has been defined by the reallocation of a single critical scarce variable.
The Scientific Revolution was a war on Ignorance. Its weapon was The Method — a systematic way to find truth. The Industrial Revolution was a war on Muscle. Its weapon was The Engine, which turned heat into limitless power. The Digital Revolution was a war on Distance. Its weapon was The Bit, which let ideas travel instantly across the planet.
The Intelligence Revolution — the one we’re living through right now — is a war on Attention. Its weapon is The Token: artificial cognition that turns intelligence into a cheap, abundant utility.
The Four Phases
This revolution moves through four predictable stages.
Phase 1: Legibility. We invent instruments to see what was hidden. The benchmark harness lets us precisely measure machine intelligence, just as the telescope made the invisible visible.
Phase 2: The Harness. Once the problem is visible, we build systems to control it. The Scientific Method was a harness for truth. Factory Discipline was a harness for labor. Today, the Industrial Intelligence Stack — foundation models, fine-tuning pipelines, agent frameworks — is the harness for cognition.
Phase 3: Institutionalization. The harness gets embedded into every organization. Just as the corporation was the institutional form of the Industrial Revolution, the AI-native company is the institutional form of the Intelligence Revolution. Microsoft already writes 30% of its code with AI. Not in five years. Now.
Phase 4: Abundance. Intelligence becomes so cheap that the bottleneck shifts from “can we solve this?” to “what should we solve?” We are entering this phase.
Why This Time Is Different
Every generation has its doomsayers who claim technology will destroy jobs, and every generation has been mostly wrong. Why should this time be different?
Because this time, the technology doesn’t augment human capacity in one domain. It replaces it across all cognitive domains simultaneously. A steam engine made your arm stronger. AI makes your brain optional — for an expanding category of tasks.
The numbers are already clear. McKinsey calculates that AI and robotics can theoretically automate 57% of all US work hours today. Goldman Sachs projects 300 million jobs displaced globally. The WEF expects 92 million displaced by 2030, offset by 170 million new jobs. Net positive — but the 92 million who lose their jobs and the 170 million who get new ones are not the same people.
The displacement has begun. Amazon cut 14,000 corporate positions, citing AI efficiency. Microsoft laid off 6,000 programmers. Klarna’s AI assistant took over the work of 700 full-time employees. Entry-level software postings in the US dropped 67%. In the UK, tech graduate positions fell 46% in a single year.
The Intelligence Revolution is not coming. It is here. The question is not whether machines will do your job. The question is whether you will own the machines that do.
In a regulated environment, the big win. NVIDIA, Microsoft, Google are not just the technological winners — they are the regulatory winners. Compliance costs are fixed costs that crush startups but are rounding errors for trillion-dollar companies. This strengthens the investment case for infrastructure heavyweights.
Why Machines Need Money
Here is the thesis that most AI investors miss entirely.
If you believe — as I do — that AI agents will increasingly operate autonomously, making decisions, executing tasks, and transacting with each other and with the physical world, then you must ask one question: How do they pay for things?
An AI agent that books a server, purchases training data, pays for an API call, or settles a smart contract cannot walk into a bank and open a checking account. It has no social security number, no address, no KYC documentation. The traditional financial system was built by humans, for humans. Machines are locked out.
This is where crypto enters the picture — not as a speculative asset, but as necessary infrastructure.
The Native Currency of the Machine Economy
Crypto is programmable money. It is money that can be written into code, executed by software, and settled without human intermediaries. A smart contract can hold funds, release them on conditions, split payments, and settle disputes — all autonomously, all verifiable, all borderless.
When an AI agent needs to rent GPU compute from another agent, it doesn’t send an invoice and wait 30 days for payment. It executes a microtransaction on a blockchain in milliseconds. When a robot needs to charge at an autonomous charging station, it doesn’t pull out a credit card. It pays via a machine-to-machine crypto transaction.
This is already happening. Bosch is letting vehicles pay autonomously at charging stations. DIMO has connected 180,000 vehicles to a blockchain where they independently sell their driving data. Cloudflare — the company through whose network 20% of the internet flows — logs one billion HTTP 402 responses per day. Status code 402: “Payment Required.” It was defined 30 years ago, and never used at scale. Now bots can pay.
Stablecoins: The Bridge
Stablecoins — cryptocurrencies pegged to the US dollar — are the practical bridge between old money and new rails. USDC, issued by Circle, processed $24 trillion in on-chain value in 2025. That’s more than Visa. Not in some distant future. Now.
For the machine economy, stablecoins are essential because they eliminate crypto’s volatility problem. A machine doesn’t speculate on Bitcoin’s price. It needs reliable units of account to execute transactions. Stablecoins provide exactly that — dollar-equivalent value on programmable rails.
Tokenization: Everything Becomes Tradeable
The next step is tokenization — the process of representing real-world assets (bonds, real estate, private equity, commodities) as digital tokens on a blockchain. BlackRock has tokenized a US Treasury money market fund. JPMorgan runs an internal blockchain for interbank settlements. 65% of all tokenized real-world assets run on Ethereum.
This matters because it unlocks liquidity in markets that have been illiquid for centuries. A building in Manhattan, a Picasso painting, a share of a private company — all can be fractionally owned, traded 24/7, and settled in seconds instead of days.
Crypto is not a speculative bet on “digital gold.” It is the financial infrastructure of the machine economy. Without programmable money, autonomous agents cannot transact. Without autonomous transactions, the machine economy cannot function. This is why the convergence of AI and crypto is not optional — it is structural.
The machine economy infrastructure stack is investable today: Circle (USDC, IPO filed), Coinbase (x402 facilitator, Base chain), Cloudflare (20% of the internet, x402 Foundation), and peaq (6 million machine addresses, Bosch moveID). Four companies that together form the blueprint of the physical machine economy.
The Geopolitical Chessboard
No technology exists in a vacuum. Every transformation powerful enough to rewrite the economy summons the state. It happened with the railroad, with electricity, with the telephone, with the internet. It will happen with AI.
USA: From Light Touch to Laissez-Faire
Under the current administration, the US has the least regulated AI landscape among major economies. The regulatory philosophy: AI is a competitive advantage that must not be jeopardized by regulation. Regulate only where concrete, demonstrable harm has occurred — not preventively.
For innovators, this is a blessing. For investors, it is clear: US companies operate with fewer regulatory constraints than their global competitors, translating into higher innovation speed and lower compliance costs.
China: Regulation as Industrial Policy
China takes a third path — regulation as industrial policy. It regulates what is politically undesirable (AI content critical of the regime, deepfakes of political figures) and aggressively promotes what is economically and militarily useful.
China’s “New Generation AI Development Plan” sets the explicit goal of becoming the world’s leading AI nation by 2030. The tools: billions in state subsidies, privileged data access (1.4 billion citizens with far more relaxed data protection than Europe), and a state-orchestrated ecosystem of universities, companies, and military research.
Europe: Regulating What It Doesn’t Build
The EU AI Act is the most ambitious AI regulation in the world. And it regulates a technology Europe does not control. The foundation models come from the US and China. It’s like writing traffic rules for highways while building no cars.
The numbers are sobering. In 2025, roughly $215 billion in AI venture capital flowed into the US. Europe received $37 billion. Ratio: nearly 6 to 1. A quarter of Europe’s most promising AI startups are considering relocating to the US. Aleph Alpha, Germany’s great AI hope, gave up building its own language model and pivoted to consulting. Mistral AI survived the EU’s regulations not through innovation, but through political lobbying.
Crypto Regulation: MiCA vs. SEC
In crypto, Europe has created the world’s first comprehensive framework with MiCA (Markets in Crypto-Assets Regulation). It provides legal certainty — but it is conservative and restrictive. In the US, the approval of Bitcoin spot ETFs in early 2024 was a watershed moment. Hundreds of billions have flowed into Bitcoin ETFs since. The regulatory mood has shifted dramatically.
For the machine economy, the regulatory question is existential: Will machines be permitted to autonomously execute crypto transactions? Who is liable when an AI agent signs a smart contract? These questions are unanswered. Their answers will determine the speed at which the machine economy grows.
The regulatory asymmetry between the US and Europe is an investment signal, not just a policy observation. Invest where innovation happens, not where it is regulated. European investors should ensure significant exposure to US and Asian technology — not optional, but essential for portfolio survival.
The Energy Bottleneck
The AI revolution runs on electricity. A lot of electricity.
A single ChatGPT query consumes roughly ten times the energy of a Google search. Training a frontier model requires the output of a small power plant for weeks. The five major hyperscalers are spending over $600 billion on AI infrastructure in 2026 — and every data center needs a massive, reliable power supply.
This creates an energy bottleneck that most AI investors ignore. You can’t prompt Claude to increase global titanium production by five percent. You can’t ask an AI to build a transformer that has a 128-week delivery time.
Nuclear: The Only Scalable Clean Baseload
Solar and wind are cheap and getting cheaper. But they are intermittent. AI data centers need 24/7 baseload power — they can’t wait for the sun to shine.
Nuclear energy is the only carbon-free source that can deliver this reliably. And the market knows it. Constellation Energy’s stock has risen 200% since signing a deal to restart Three Mile Island Unit 1 to power a Microsoft data center. Amazon, Google, and Meta are all pursuing nuclear power agreements.
Small Modular Reactors (SMRs) are the next frontier. NuScale, Oklo, and X-Energy are developing factory-built reactors that can be deployed near data centers — eliminating the need for massive transmission infrastructure. The economics are improving rapidly.
The Physical Infrastructure Nobody Talks About
Between the uranium ore and the fuel rod sits a bottleneck almost nobody knows: conversion. Uranium must be converted into uranium hexafluoride before it can be enriched. Globally, there are four relevant facilities. Only one in the US — the Metropolis plant in Illinois, built in 1958.
Copper is hitting $13,000 per ton, driven by data center demand. A single hyperscale data center consumes as much copper as thousands of single-family homes. S&P Global projects a supply deficit of 10 million metric tons by 2040.
Power transformers have a 128-week delivery time. The specialized steel they require — grain-oriented electrical steel — is produced by exactly one company in the US: Cleveland-Cliffs, in two plants.
China controls over 80% of global tungsten production and processing. New export controls in 2025 reduced exports by 14%.
The machine economy needs machines. Machines need materials. And materials have supply chains that no AI can accelerate. The picks-and-shovels strategy doesn’t end at chips — it extends into transformers, copper mines, uranium infrastructure, and specialty metals. These companies have moats that no AI can breach.
The energy infrastructure play: Cameco (world’s largest uranium producer), Constellation Energy (nuclear renaissance), uranium ETFs. These assets trade at valuations that price in only one demand driver, even though three or four are acting simultaneously — AI data centers, electrification, defense, and the nuclear renaissance.
AI Gets Bodies
The robotics revolution is the part of the convergence you can touch.
Tesla announced in January 2026 that Optimus Gen 3 — the first mass-production-ready version of its humanoid robot — will be unveiled in Q1 2026. Over 1,000 units are already working in Tesla factories. The plan: 10,000 to 500,000 units per year.
Figure AI has tested its Figure 02 robot at BMW’s Spartanburg plant for eleven months. Not eleven days — eleven months. The robots worked 10-hour shifts, picking parts from logistics containers and placing them on welding fixtures. This is no longer a lab experiment.
1X Technologies from Norway opened pre-orders for NEO, marketed as the world’s first consumer robot. Price: $20,000 or $499 per month. Designed for household tasks: dishwashing, laundry, tidying. That a humanoid robot for the household is offered at $20,000, not $200,000, is the actual point.
Aurora Innovation launched commercial driverless truck operations in Texas in May 2025. Dallas to Houston, no human driver. Over 100,000 driverless miles without a safety incident. Waymo operates robotaxis in 10 US cities with over 400,000 rides per week.
The Complexity Ladder
The path is clear. Structured environments first, unstructured last. Factories before warehouses, warehouses before households, households before construction sites.
And the question my wife asks: When does the plumber robot arrive? The honest answer: Not soon. Trades are the hardest to automate — every house is different, every job site unstructured, every pipe burst unique. Fully autonomous tradesperson robots are realistic no earlier than 2035. But WIRED titled in January 2026: “The Real AI Talent War Is for Plumbers and Electricians.” AI data centers need plumbers so desperately that there are massive shortages. Here, AI creates demand, not automation.
In the long run, it is hard to imagine which work will remain for humans at all. Not impossible — hard to imagine. And that is precisely why it is so important to think about it now.
Robotics is at the bottom of the S-curve. The market prices it as a niche. ARK Invest prices it as the next trillion-dollar market. One of them is right. The asymmetry favors early positioning: Tesla (Optimus + FSD + Energy), Intuitive Surgical (surgical robots), ABB, Fanuc, and — when they IPO — Figure AI.
The Next Interface — Brain-Computer Interfaces
Every technological revolution came with a revolution in how humans interact with machines. The mainframe required punch cards. The PC brought the keyboard. The smartphone introduced the touchscreen. Each transition increased bandwidth and reduced friction.
We are approaching the next two transitions in rapid succession. And the second will change not just how we work, but how we experience reality.
From Keyboard to Voice to Thought
Speech is the most natural form of human communication. It’s what we’re evolutionarily built for. The keyboard was always a workaround. Tools like Wispr Flow and Google’s Gemini Live already enable fluent, contextual voice control. Adoption will be fast because people don’t need to learn a new skill to speak.
But speech has a limit: bandwidth. Humans speak at about 150 words per minute. We think orders of magnitude faster. Speech is a straw through which we try to drink an ocean.
Brain-Computer Interfaces: Closer Than You Think
Neuralink has implanted its “Telepathy” device in 21 human patients. These patients — many of them paralyzed — control computers, type messages, and operate robotic arms through thought alone. Typing speeds of 40 words per minute, nearly as fast as the average smartphone user.
Synchron raised $200 million for its Stentrode system — a BCI implanted through a blood vessel, no open skull surgery required. Partnerships with Apple and NVIDIA.
And then the move that reveals where the smart money is looking: In January 2026, OpenAI led a $250 million seed round in Merge Labs, Sam Altman’s BCI startup, at a valuation of $850 million. OpenAI’s stated goal: “to connect biological and artificial intelligence to maximize human capability, agency, and experience.” When the company building the world’s most advanced AI invests a quarter billion in brain interfaces, the strategic direction is unmistakable.
The Convergence: BCI + AI-Generated Worlds
Google DeepMind’s Genie 3 can generate interactive 3D environments from a text prompt — navigable in real-time at 24fps in 720p. You describe a world, and the AI builds it around you.
Now combine this with brain-computer interfaces. Instead of typing a prompt, you think a world — and it materializes. Instead of viewing it on a screen, sensory information streams directly to your neural cortex. Instead of pressing buttons to interact, your intentions are read and executed in real-time.
What you get is not VR as we know it. What you get is a synthetic experience that, from the brain’s perspective, is indistinguishable from reality.
The progression is clear: from keyboard to voice to thought. Each step removes a friction layer between human intention and machine execution. And when that thought can conjure entire worlds — worlds that feel as real as this one — the implications are not just technological. They are civilizational. The addictive potential of this technology makes social media look like a children’s toy.
BCI is a classic asymmetric early-stage opportunity. Most companies are private (Neuralink at $9.7B, Synchron, Merge Labs, Paradromics), but the secondary effects are investable today. NVIDIA profits doubly: GPU compute powers both AI world generation and neural signal processing. Apple’s partnership with Synchron suggests it views BCIs as the next interface paradigm after the iPhone. The compute requirements of this scenario are staggering — reinforcing the nuclear energy thesis.
The Investment Framework
There is a moment every serious investor recognizes. The moment when you believe you understand everything — and still don’t know what to buy tomorrow morning.
This chapter closes that gap. With a framework that translates the insights of this essay into an investable strategy.
The Barbell Strategy
Nassim Taleb may reject Bitcoin. But his barbell strategy is perfectly designed for a world oscillating between “everything’s fine” and “everything’s about to change radically.”
The idea: nothing mediocre. No investments that are sort of safe and sort of risky. Instead:
One side: 60-70% in solid, cash-generating, understandable assets.
Other side: 30-40% in asymmetric bets with limited downside and unlimited upside.
This works because we live in a world where safe assets deliver little, but extreme events — Taleb’s Black Swans — occur with disproportionate frequency. The barbell protects against ruin and profits from chaos.
Where the Smartest Money Agrees
I’ve studied what the smartest investors in the world are doing. Ray Dalio, Raoul Pal, Cathie Wood, Stanley Druckenmiller, Paul Tudor Jones, and Warren Buffett don’t agree on much. But they agree on three things:
1. The monetary system has a problem. The US national debt math no longer works. Interest costs exceed the defense budget. Every rate hike worsens the problem. The only exit is inflation — the gradual devaluation of money. Dalio now recommends 15% in Bitcoin or gold as a hedge.
2. Technological convergence is real and investable. Druckenmiller built massive positions in Amazon, Meta, and Alphabet in Q3 2025. ARK Invest projects humanoid robots as a multi-trillion dollar market. Tudor Jones doubled his Bitcoin position.
3. Most people will still lose money. Howard Marks warns that AI valuations are “elevated to concerning.” The technology was real and transformative during the dot-com era too — and most early investors still lost money. Not because the thesis was wrong, but because they paid too much, picked the wrong horses, or sold at the wrong time.
The Core (60-70%): Buffett’s Moats for the Machine Age
NVIDIA has the compute moat: 80%+ market share in AI training chips, a software ecosystem (CUDA) so deeply embedded that switching takes years. Apple has the ecosystem moat. Ethereum has the network moat: most smart contracts, most DeFi activity, most developers.
Complement with broad index exposure (MSCI World, Nasdaq-100), energy infrastructure (Cameco, Constellation Energy), and robotics pure plays (Intuitive Surgical, ABB).
The Asymmetric Side (30-40%): Where Fortunes Are Made
Bitcoin (15-20%): Not speculation — insurance against monetary devaluation. After one year of holding, tax-free for German investors who buy directly. BlackRock, Fidelity, and pension funds are allocating. This is no longer fringe. It is institutional adoption.
Ethereum (5-10%): The operating system of the tokenized economy. 65% of all tokenized real-world assets run on it.
Machine Economy Infrastructure: Tokens and companies building the plumbing — stablecoin infrastructure, DeFi protocols, machine identity systems.
Speculative Frontier: BCI-adjacent companies, longevity biotech, frontier AI tokens. Total loss possible. But if the convergence plays out, these are the infrastructure layers that capture asymmetric upside.
The Most Important Rule
Size the barbell so the asymmetric side can go to zero without destroying you. That is the entire point. The safe side protects your existence. The speculative side protects your future.
In the worst case, the asymmetric side loses 80%, the core drops 30%. Total portfolio drawdown: about 30%. Painful, but not fatal. And you are still in the game when the thesis materializes.
In the best case: the core doubles in five years, crypto positions quadruple, biotech triples. Portfolio: roughly 2.5x. That is the asymmetry Taleb describes: limited risk, unlimited upside.
The barbell strategy answers the right question. Not “how do I maximize returns?” but “how do I stay in the game — if I’m wrong AND if I’m right?” Dollar-cost average in over 6-12 months. Don’t try to time the bottom. Time in the market beats timing the market.
The physical infrastructure play is the most overlooked part of the thesis. Software investors talk about NVIDIA and ignore that every data center needs copper, transformers with 128-week lead times, and uranium conversion capacity. Companies like Cameco, Cleveland-Cliffs, and Materion have moats no AI can breach — and trade at valuations that price in only one demand driver when three or four are active simultaneously.
Risks and Counterarguments
Every investment thesis that doesn’t name its counterarguments is not a thesis. It is propaganda. So let me take the risks seriously.
The AI Bubble
The five major hyperscalers are spending $600 billion on AI infrastructure in 2026. AI revenues are at $50-60 billion. A 10:1 ratio. Taleb warns of “escalating volatility and potential bankruptcies in the software sector.” Sam Altman himself admits a bubble exists.
The dot-com parallel is obvious. And not entirely wrong. But it is incomplete.
The companies building AI today generate real revenue. GitHub Copilot is changing how software is written. Claude and GPT automate knowledge work. Tesla deploys 700,000 robots in its factories. The technology is here. The adoption is real.
A correction will come. Certainly. The question is whether you buy the dip or sell in panic. My answer is clear: I buy.
No AI Winter Is Coming
I hold a third AI winter to be impossible. Not improbable — impossible. The development is simply too fast. Every month, new models surpass the previous generation. The gap between what the technology can do and what the broad public has grasped is enormous. What exists is not a technology gap but a perception gap.
What If I’m Wrong?
This question deserves an honest answer.
If the convergence of AI, robotics, and crypto does not materialize as described — if the AI bubble bursts, if crypto is crushed by regulation, if robotics stagnates — then I was wrong. Fundamentally wrong.
But that is exactly what the barbell is built for. In the worst case, you lose the asymmetric bets. That is 25-35% of the portfolio. It hurts. But the safe core — your NVIDIA shares, your Nasdaq-100 ETF, your broad market exposure — remains. These companies make money regardless of whether the machine economy arrives in five years or fifteen.
What I cannot tell you: that I am certainly right. What I can tell you: the barbell strategy works even if I am wrong. You lose the aggressive part. And you sized it to be survivable.
The Real Existential Risk
Geoffrey Hinton estimates a 10-20% probability that AI leads to human extinction within thirty years. Yoshua Bengio warns of “irreversible loss of control over autonomous AI systems.”
My position: I take existential risk seriously, but I see the greater danger elsewhere. I believe we will extinguish ourselves — slowly. Through insufficient reproduction. Through the emerging lethargy, laziness, and purposelessness that will engulf a large portion of humanity when machines do all the work.
The greatest danger of the machine economy is not the loss of control over AI. It is the loss of drive in humans.
Machines replace work → no labor income → no capital → no wealth → no drive → no future.
I refuse to accept that path. Not for me. Not for my family. Not for anyone reading this.
Taiwan: The Underestimated Risk
TSMC produces over 90% of the world’s most advanced semiconductors. Every AI chip from NVIDIA, every Apple processor — depends on a single factory on an island 130 kilometers off the Chinese coast. An island China claims sovereignty over, with increasing aggression.
No portfolio strategy can hedge against a global conflict between nuclear powers. What you can do is diversify — geographically, across asset classes, across jurisdictions. And hope that rationality prevails.
The risks are real. Every single one. A correction can come. Regulation can slow things down. Crypto will keep swinging. Taiwan is a powder keg. You will doubt this thesis at least once in the next ten years. But not one of these risks changes the fundamental fact that AI, robotics, and crypto will define the economy of the next decade. They can make the path bumpier, shift the timeline, destroy individual companies. But the transformation itself? It can no longer be stopped.
Why Now
Howard Marks writes that the best investments are often the ones that feel most uncomfortable. Investing in AI, robotics, and crypto today feels uncomfortable. Prices are high. The technology is complex. The future is uncertain.
That is precisely the point.
The greatest returns arise during periods of uncertainty — because uncertainty keeps most investors on the sidelines. When everyone is convinced an investment is safe, prices are already too high for outsized returns. The asymmetric opportunity exists now, in the phase of uncertainty, before the broad market understands what is coming.
Two Types of Skeptics
When I tell people that I am the first developer in my own company — not because I learned to code, but because I have an AI that codes for me — I encounter two reactions.
The first comes from people who don’t know and don’t want to know. AI is science fiction to them, somewhere between Terminator and nonsense. They can’t assess what they’ve never experienced.
The second reaction is more interesting and more troubling. It comes from people who should know better. Employees who work with software daily. Developers who see what Copilot can do. They know the change is coming, but they don’t want to acknowledge it. Because acknowledging it would mean questioning their own future.
And then there is a third reaction that surprises me every time. People you’d never expect — a sports agent in Tulum running his entire client analysis with AI, a fund manager who turns out to be building his own trading agents, a master craftsman who automated his cost estimates. Everyone discovers this technology in their own way.
The Causal Chain
Remember the chain that runs through this entire thesis:
Machines replace work → no labor income → no capital → no wealth → game over.
This is not a theoretical construct. It is the default path for everyone who does not act now. Every month you wait, the window gets smaller. AI company valuations rise. Crypto infrastructure matures. Robots get cheaper and better.
At some point, the average worker will no longer be able to afford a ticket to the machine economy — because labor income shrinks faster than asset prices rise.
The Question
The convergence of artificial intelligence, robotics, and crypto is not one possibility among many. It is the most probable future. AI will deliver the intelligence. Robots will take over physical labor. And crypto will be the lubricant of the machine economy.
The question is not: Can I afford to invest now?
The question is: Can I afford not to?
This essay is the map. The book — World of the Machines — is the territory. It contains the detailed research, the specific portfolio construction, the chapter on longevity investing, and the complete framework for positioning yourself in the machine economy. Coming soon.