Why AI, Robotics, and Crypto Represent the Greatest Investment Opportunity of Our Lifetime
An Essay by Thomas Huhn
I. At the Dawn of the Greatest Transformation in 250 Years
There are moments in history when the tectonic foundations of the economy shift. The steam engine was such a moment. Electrification. The internet. Those who invested during these phases — not in the technology itself, but in the right layer of value creation — built wealth that endured for generations.
We are standing at such a point right now. And this time, the transformation is more radical than anything we have ever known.
Artificial intelligence is not a software innovation like cloud computing or social media. It is a cognitive revolution. For the first time in human history, we are building systems that can think, learn, decide, and act — without human guidance. When you combine this capability with robotics — giving AI a body — and then place in its hands a native means of payment that operates without banks, without bureaucracy, without human gatekeepers, you create something that rewrites our entire economic order.
This is not science fiction. This is what will happen over the next ten years.
I am not writing this essay as a detached observer. I build AI products myself. With accessibleAI, I develop software based on large language models that helps companies automate regulatory compliance using AI. Every day, I witness firsthand how rapidly this technology is evolving, how it is transforming work processes, how it is calling entire professions into question. And I see with the clarity of a practitioner: whoever understands the convergence of AI, robotics, and crypto holds the investment thesis of the next decade.
Peter Thiel wrote in “Zero to One” that the most valuable companies are those doing something most people don’t yet believe is possible. Most people still don’t believe that AI agents will autonomously manage supply chains, negotiate contracts, and process payments within five years. They don’t believe that humanoid robots will be working in factories by 2030. And they certainly don’t understand why these machines will need cryptocurrencies to function.
That is precisely what makes this thesis so valuable.
II. The AI Revolution in Four Phases
To understand where the journey is headed, you must understand where we stand. The evolution of artificial intelligence can be described in four phases, and we are currently at the transition from phase two to phase three.
Phase 1: AI as a Tool (2020–2023). ChatGPT launched in November 2022 and reached 100 million users in two months — faster than any technology before it. But in this phase, AI was essentially a better Google: you asked a question, you got an answer. Text generation, image generation, translation, summarization. Powerful, but passive. The human remained the actor; the AI was the instrument.
Phase 2: AI as a Co-Worker (2024–2025). This is where we are now. AI systems are being integrated into workflows. They don’t just write emails — they autonomously respond to customer inquiries. They don’t just generate code snippets — they develop entire features. GitHub Copilot now writes over 40 percent of all new code on the platform. AI-powered legal analysis tools accomplish in minutes what junior lawyers used to need weeks for. The productivity gains are real and measurable — McKinsey estimates the annual value creation potential of generative AI at $2.6 to $4.4 trillion.
Phase 3: AI as an Autonomous Agent (2025–2028). This is the paradigm shift. An agent is not a chatbot. An agent has goals, plans steps, uses tools, corrects errors, and acts independently. Anthropic, OpenAI, Google DeepMind — all the major labs are working feverishly on agent frameworks. The first implementations are already running: AI agents that book travel, make purchases, configure software systems, conduct research, and produce reports — without a human approving every step. If Phase 2 made AI a co-worker, Phase 3 makes it a manager.
Phase 4: Embodied AI — Robots (2026–2035). And then the intelligence gets a body. Tesla has unveiled its humanoid robot Optimus and plans to produce it in larger quantities starting in 2026. Figure AI has raised over $700 million in venture capital, with investors including Jeff Bezos and NVIDIA. Boston Dynamics, Agility Robotics, Unitree from China — the field is broad and moving fast. Goldman Sachs projects that the market for humanoid robots could reach a volume of $38 billion by 2035. Elon Musk goes further, saying that Optimus will ultimately be worth more than Tesla’s entire automotive business.
Ray Kurzweil predicted this exponential trajectory decades ago. In “The Singularity Is Near,” he described how technological progress accelerates because each generation of technology provides the tools to develop the next one faster. AI accelerates chip development, better chips accelerate AI, faster AI accelerates robotics. It is a spiral, and it is spinning ever faster.
What does this mean for investors? Each phase unlocks a larger addressable market. Phase 1 affected software developers and early adopters. Phase 2 affects knowledge workers and businesses. Phase 3 affects the entire service economy. Phase 4 affects the physical world — factories, logistics, healthcare, agriculture, construction. With each phase, the addressable market grows by an order of magnitude.
III. Why Machines Need Crypto
This is where it gets interesting. And this is the thesis that most investors have not yet grasped.
When AI agents act autonomously — executing contracts, purchasing services, allocating resources — they need a means of payment. And the traditional financial system was not built for this.
Let’s think it through. An AI agent working for a company needs to rent a cloud server, hire a freelancer, and purchase an API license. In today’s system, it would need: a bank account (which requires a legal or natural person), a credit card (which requires a credit check), compliance documentation (KYC, anti-money laundering), and human signatures. The system was built for humans. Machines don’t fit into it.
Cryptocurrencies solve this problem fundamentally. An AI agent needs only a private key to sign transactions. No bank, no identity verification, no business hours. Smart contracts on Ethereum or Solana make it possible to encode conditions: “Pay X when service Y is delivered.” This is programmable economics. Machines cannot read and interpret contracts the way humans do — but they can execute code. Smart contracts are contracts that machines understand.
Balaji Srinivasan anticipated this concept of a machine economy in his thinking about the “Network State” and the future of money. He argues that the next billion “users” in the crypto ecosystem will not be humans, but machines. AI agents, IoT devices, autonomous vehicles, robots — all will need to conduct transactions, and all will need a means of payment that is natively digital, permissionless, and programmable.
The numbers already point in this direction. Stablecoins — crypto-based, dollar-pegged tokens like USDC and USDT — processed over $27 trillion in transaction volume in 2024, surpassing Visa. This growth is not driven by speculators but by real utility: cross-border payments, settlement, and increasingly, machine-to-machine transactions.
Microtransactions are another key. When an AI agent wants to make an API call for $0.001, traditional payment systems cannot accommodate this — the transaction fees consume the amount. On Layer 2 blockchains like Arbitrum, Optimism, or the Lightning Network for Bitcoin, transactions cost fractions of a cent. The infrastructure for the machine economy already exists.
Machine-to-Machine Payments (M2M) are no longer a distant prospect. Projects already exist today in which AI agents operate on blockchains: they trade compute resources, purchase data, and pay other agents for services. The Fetch.ai network, Ocean Protocol for data markets, IOTA for IoT payments — the foundations are being laid. When the agent economy gains momentum in Phases 3 and 4, demand for crypto-based payment instruments will rise exponentially.
Bitcoin plays a special role in this scenario. Not as a payment method for microtransactions — it is too slow and too expensive on the base layer for that — but as a store of value. Bitcoin is the hardest money that has ever existed: 21 million units, non-inflationary, non-confiscatable, controlled by no government. In a world where AI massively boosts productivity and thus potentially creates deflationary pressure, while governments simultaneously print money to maintain social stability, a scarce, neutral store of value becomes the anchor of the system.
Ethereum, on the other hand, positions itself as the “operating system” of the decentralized economy — the platform on which smart contracts run, DeFi protocols operate, and AI agents settle their transactions. The competition is real — Solana is faster, Arbitrum is cheaper — but Ethereum’s network effects and security are unmatched.
The core thesis is simple: when machines become economic actors, they need a monetary system built for machines. Crypto is that monetary system.
IV. The Geopolitical Dimension: The U.S., China, and Europe’s Silence
Investments do not happen in a vacuum. The AI revolution has a geopolitical dimension that must be understood to make intelligent decisions.
There are exactly two AI superpowers: the United States and China. Period.
The U.S. dominates in foundational research (OpenAI, Anthropic, Google DeepMind, Meta AI), in hardware (NVIDIA controls over 80 percent of the market for AI training chips), in infrastructure (AWS, Azure, Google Cloud), and in capital (in 2024 alone, over $100 billion flowed into American AI startups). NVIDIA’s revenue exploded from $27 billion in fiscal year 2023 to over $130 billion in fiscal year 2025 — a fivefold increase in two years, driven by the insatiable demand for AI compute.
China controls application and the hardware supply chain. Chinese firms like ByteDance, Baidu, and Alibaba are investing massively in AI. DeepSeek demonstrated in early 2025 with a surprisingly capable open-source model that China can build competitive AI despite U.S. export controls on chips. In robotics, China is already the leader in production: Unitree delivers humanoid robots for under $20,000 — a fraction of what Western competitors charge. BYD and other Chinese manufacturers are integrating AI and robotics into their production lines at a speed that puts Western industrial conglomerates to shame.
And Europe? Europe regulates. The EU AI Act is the most ambitious AI regulatory framework in the world — and possibly the most damaging. While the U.S. and China support AI companies with billions, Europe debates risk classifications and transparency obligations. This is not inherently wrong — regulation has its place — but it is telling that not a single one of the leading foundation models comes from Europe. No European company ranks among the top 20 most valuable AI firms. Europe’s greatest contribution to the AI revolution so far has been the bureaucracy that slows it down.
I say this as a European, as a German, with the pain of someone who watches his continent squander a historic opportunity. But as an investor, one must accept reality, not wish for it. And the reality is: the returns of the AI revolution will be generated in the United States and Asia.
What does this mean in practice? It means that an AI-focused portfolio should be predominantly invested in U.S. assets and selectively in Asian ones. It means that European equities should be viewed with caution as AI investments — the regulatory risks are real, and the innovation momentum is absent. And it means that crypto, which is by definition global and jurisdiction-independent, can be viewed as a strategic hedge against European overregulation.
V. The Most Probable Future Scenario: 2026–2035
Allow me to paint a picture. Not the most utopian, not the most dystopian, but the most probable scenario based on current development trends.
2026–2028: AI agents become the standard in knowledge work. Every major corporation has AI systems that autonomously answer emails, produce reports, analyze data, write code, and prepare decisions. The first humanoid robots work in controlled environments — Tesla’s factories, Amazon warehouses, Japanese care facilities. Unemployment in certain sectors rises noticeably: call centers, basic accounting, data processing, standard legal advisory. At the same time, new professions emerge around AI management, prompt engineering, and robot supervision. Stablecoins are recognized by the first governments as legal payment instruments for digital transactions. Bitcoin permanently surpasses the $200,000 mark, driven by institutional adoption and its growing use as a reserve asset.
2028–2031: The tipping point. AI handles 30 to 50 percent of today’s knowledge work faster, cheaper, and often better than humans. Humanoid robots cost under $30,000 and are produced in quantities of hundreds of thousands. They work in logistics centers, on construction sites, in agriculture, and in elderly care. The economy bifurcates: companies that adopt AI and robotics experience productivity gains of 200 to 300 percent. Companies that don’t, disappear. Governments worldwide grapple with the question of how to manage an economy in which traditional wage labor is shrinking. The first countries experiment with universal basic income, funded by productivity gains and robot taxes. The machine-to-machine economy on a blockchain basis reaches annual transaction volumes in the trillions.
2031–2035: The new normal. AI and robots are as ubiquitous as the internet is today. Economic output per capita rises dramatically, but distribution is more unequal than ever before. Those who own the infrastructure of this new world — the chips, the models, the robots, the crypto protocols — are among the wealthiest individuals and institutions in history. Cryptocurrencies have established themselves as the infrastructure layer of the autonomous economy. Not as a replacement for sovereign money, but as a parallel system for machine transactions, decentralized finance, and cross-border value transfer.
This scenario is neither optimistic nor pessimistic. It is the logical extrapolation of what is already in motion. Ray Kurzweil has predicted that AI will reach human-level capability — Artificial General Intelligence — by 2029. Even if he is off by five years, it does not change the overall picture.
The question is not whether this transformation is coming. The question is whether you are on the right side of it.
VI. The Long-Term Investment Case: A Portfolio for the Machine Economy
How does one translate this conviction into a concrete investment portfolio? Here, a concept described by Nassim Taleb in “Antifragile” comes into play: the barbell strategy.
Taleb’s idea is elegant: instead of putting the entire portfolio into “medium-risk” assets, you split it into two extremes — a safe core and asymmetric bets. The safe core protects capital; the bets offer disproportionate upside. The crucial point: the maximum losses on the bets are capped (you can only lose the capital deployed), but the gains are theoretically unlimited.
For a model portfolio, this could look as follows:
The safe core (60–70 percent): This is about companies that benefit in every AI scenario — the infrastructure providers, the “picks and shovels” of the gold rush.
NVIDIA is the most obvious position, and yet not one to be dismissed. Howard Marks repeatedly warns in his memos against avoiding obvious investments simply because they are obvious — sometimes the consensus is right, and the risk lies in betting against it. NVIDIA dominates the AI chip market with a grip reminiscent of Intel in the 1990s — except that the addressable market is orders of magnitude larger. As long as AI training and inference grow exponentially, NVIDIA grows with them.
Microsoft, Apple, Google (Alphabet), Amazon, and Meta form the Big Tech oligopoly that controls the infrastructure of the AI era. Cloud computing, data, distribution, models — it all lies in their hands. Their P/E ratios are high, but their monopoly positions and cash flows are real. A broadly diversified technology ETF like the Nasdaq-100 captures this cluster efficiently.
For more direct robotics exposure, companies like Intuitive Surgical (the market leader in surgical robots), Fanuc and ABB (industrial robotics), Rockwell Automation, and increasingly Tesla are compelling — not because of the cars, but because of Optimus and the robotics platform.
The asymmetric bets (30–40 percent): This is where it gets exciting. These positions can multiply tenfold or go to zero. That is exactly the point.
Bitcoin is the foundational bet. Anyone who believes that the machine economy needs a neutral, non-sovereign store of value should own Bitcoin. Institutional adoption reached an inflection point in 2024 with the launch of spot ETFs — BlackRock, Fidelity, and others already manage hundreds of billions in Bitcoin ETFs. Bitcoin has a built-in asymmetry: the downside is known (at worst, a painful but limited loss); the upside is not. If Bitcoin truly becomes digital gold — and gold has a market capitalization of over $18 trillion — then the current level is a fraction of its potential.
Ethereum is the bet on the smart contract platform of the machine economy. DeFi, NFTs, and increasingly AI agents use Ethereum as their settlement layer. The staking model additionally offers a form of “dividend” of three to four percent annually.
Stablecoins themselves are not investments (they are by definition price-stable), but the companies and protocols behind them are. Circle (the USDC issuer) is planning an IPO. Companies building stablecoin infrastructure — payment rails, on-ramps, compliance tools — are at the beginning of an enormous market.
More selective bets could include AI infrastructure tokens: Render Network (decentralized GPU compute), Filecoin (decentralized storage), or private equity stakes in robotics startups, where accessible.
The logic of the barbell strategy, in Taleb’s words: “If you protect yourself from ruin, you can profit from volatility.” The safe core ensures that even a crash in the asymmetric bets does not jeopardize the overall portfolio. The bets ensure that you participate in the upside of the greatest technological transformation in history.
VII. Risks and Counterarguments: What Can Go Wrong
Any investment thesis that does not name its counterarguments is not a thesis — it is propaganda. So let us take the risks seriously.
A new AI winter. The history of artificial intelligence is paved with phases of overblown expectations, followed by disappointment and funding cuts. The AI winters of the 1970s and 1990s set research back by years. Could it happen again? Theoretically, yes. Practically, the situation is fundamentally different: AI systems are already delivering measurable economic value. GitHub Copilot, autonomous customer service systems, AI-driven drug development — these are not demos; they are products generating revenue. An AI winter is possible, but the probability is significantly lower than in earlier cycles. And even if it comes, it would be a buying opportunity, not an exit point.
Regulatory overreach. Governments could regulate AI so heavily that innovation grinds to a halt. The EU AI Act already moves in this direction. China could restrict access to AI technology for geopolitical reasons. The U.S. could pursue anti-AI policies under a different administration. This risk is real, but it primarily affects developers and users — infrastructure providers (NVIDIA, cloud providers) actually benefit from regulation because it creates compliance costs that only large players can bear.
Crypto bans and regulatory risks. Governments could ban cryptocurrencies or regulate them so heavily that they lose their utility. China tried — Bitcoin still exists. The U.S. has taken the opposite path with spot ETFs, integrating crypto into the regulated financial system. The risk of a global crypto ban decreases with every institutional investor that enters the space. But regional restrictions, particularly in Europe, remain a factor.
Valuation risks. NVIDIA with a P/E ratio above 50, Bitcoin above $100,000, tech stocks at all-time highs — aren’t prices already too high? Howard Marks would say: price is what you pay, value is what you get. If the AI revolution becomes as large as described here, today’s valuations are a fraction of future value. If not, you overpaid. The barbell strategy addresses precisely this risk: the safe core cushions valuation corrections, and the asymmetric bets are sized so that their total loss is bearable.
Black Swans. Nassim Taleb’s entire body of work revolves around the unpredictable. A global war, a pandemic, a technological accident, a cyberattack on crypto infrastructure — things we cannot foresee but whose possibility we must acknowledge. Taleb’s answer is not avoidance, but antifragility: building a portfolio that profits from volatility rather than suffering under it. Bitcoin is antifragile in this sense — every crisis that erodes trust in state institutions strengthens the case for a decentralized store of value.
Labor markets and social stability. If AI and robots replace millions of jobs, the societal response could turn toxic: Luddite movements, anti-technology politics, redistribution demands that erode corporate profits. This risk is the least priced in and perhaps the most dangerous. A wise investor hedges against it by investing not only in the winners of automation, but also in the infrastructure that enables societal adaptation — education, retraining, and yes: universal basic income, funded by the productivity gains of machines.
VIII. The Next Interface: From Keyboards to Thought — Why Brain-Computer Interfaces Will Reshape the Human-Machine Relationship
Every technological revolution has come with a revolution in how humans interact with machines. The mainframe demanded punch cards. The personal computer brought the keyboard and mouse. The smartphone introduced the touchscreen. Each transition increased bandwidth — the amount of information a human could feed into a machine per unit of time — and reduced friction. The more natural the interface, the faster adoption exploded.
We are now approaching the next two transitions in rapid succession. And the second one will change not just how we work, but how we experience reality itself.
The first transition is already underway. Tools like Wispr Flow allow users to control their entire computer through natural speech — not clumsy voice commands, but fluid, conversational language that the system understands contextually. You don’t say “open application mail, new message, recipient colon.” You say, “Write an email to Sarah telling her I’ll be late for dinner.” The AI understands intent, not syntax.
This works because large language models have reached a level where they don’t just transcribe words — they understand meaning, context, and nuance. Google’s Gemini Live API already demonstrates real-time bidirectional communication with AI: you speak, the AI responds in natural voice, and it can simultaneously process video streams — seeing what you see, hearing what you hear, responding in the moment. The output is no longer just text on a screen. It is a stream of information — spoken words, visual overlays, dynamically generated content.
Speech is the most natural form of human communication. It is what we evolved for. The keyboard was always a workaround — a translation layer between thought and machine. Voice removes that layer. And adoption will be swift: humans do not need to learn a new skill to speak. They need to unlearn the habit of typing.
But speech has a fundamental limitation: bandwidth. Humans speak at roughly 150 words per minute. We think at a rate that is orders of magnitude faster. The conscious mind processes approximately 50 bits of information per second — but the subconscious brain processes an estimated 11 million bits per second. Speech captures a fraction of cognitive output. It is a straw through which we try to pour an ocean.
This is where brain-computer interfaces enter the picture. A BCI reads neural signals directly — bypassing the slow, lossy translation through vocal cords, tongue, and lips. In theory, a sufficiently advanced BCI could transmit the full bandwidth of conscious thought to a machine. In practice, we are already further than most people realize.
Neuralink has implanted its “Telepathy” device in 21 human patients as of early 2026. These patients — many of them paralyzed — can control computers, type messages, and operate robotic arms using thought alone. The system achieves typing speeds of 40 words per minute, approaching the speed of able-bodied smartphone users. The PRIME study is ongoing, with a primary completion date in 2026 and full results expected by 2031. Neuralink has raised over $650 million to date at a valuation of $9.7 billion — and is ramping up production for next-generation implants with plans to automate manufacturing in 2026.
But Neuralink is not alone. Synchron has raised $200 million in Series D funding for its Stentrode system — a BCI that is implanted through a blood vessel in the brain, requiring no open-skull surgery. It has partnerships with Apple and NVIDIA and is preparing for pivotal clinical trials in 2026 that could make it the first commercially scalable implanted BCI. Paradromics has demonstrated information transfer rates exceeding 200 bits per second — higher than the linguistic information content of human speech — with system latency of just 11 milliseconds.
And then there is the move that signals where the smart money sees the future: In January 2026, OpenAI led a $250 million seed round in Merge Labs, Sam Altman’s brain-computer interface startup, at an $850 million valuation. OpenAI’s stated goal: “to bridge biological and artificial intelligence to maximize human ability, agency, and experience.” When the company building the world’s most advanced AI invests a quarter of a billion dollars in brain interfaces, the strategic direction is unmistakable.
The BCI market is projected to grow at a compound annual growth rate of over 15 percent, with the U.S. market alone expected to reach $3 billion by 2035. Total BCI funding tripled to $867 million in 2025. This is early-stage infrastructure investment — the equivalent of investing in fiber optic cables in 1995.
Now consider what happens when two technologies converge: brain-computer interfaces and AI-generated three-dimensional worlds.
Google DeepMind’s Genie 3, released in August 2025, can generate interactive 3D environments from a simple text prompt — navigable in real time at 24 frames per second in 720p resolution. You describe a world, and the AI builds it around you. Not a pre-rendered scene, but a dynamically generated, physics-aware environment that responds to your actions. World Labs’ Marble model goes further, creating navigable 3D scenes from text, images, video, or sketches. As of February 2026, Google’s Project Genie is available to consumers, allowing users to walk through AI-generated worlds in real time.
These are not video games in the traditional sense. There is no developer, no studio, no years of production. The AI is the engine, the artist, the world-builder, and the game master simultaneously. And the technology is improving at the pace of large language models — which is to say, exponentially.
Now combine this with a brain-computer interface.
Instead of typing a prompt or speaking a description, you think of a world — and it materializes. Instead of watching it on a screen, the visual and sensory information is streamed 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 virtual reality as we know it today — clumsy headsets with visible pixels and noticeable latency. What you get is a synthetic experience that is, from the brain’s perspective, indistinguishable from reality. Every sense engaged. Every interaction natural. Every world tailored precisely to the individual’s desires.
This is where the thesis takes a darker — and for investors, critically important — turn.
The AI revolution described in the earlier chapters of this essay will, by all credible projections, eliminate or fundamentally transform tens of millions of jobs. When AI agents handle knowledge work and humanoid robots handle physical labor, a large portion of the global workforce will find itself without purpose, without income, and without the structure that employment provides.
Governments will respond with some form of universal basic income — the essay already addresses this in its scenario for 2028–2031. But money without purpose creates a vacuum. And nature abhors a vacuum.
Brain-computer interfaces combined with AI-generated worlds will fill that vacuum.
Consider the trajectory: Today, smartphone addiction affects an estimated 6.3 percent of the global population — roughly 500 million people. This is caused by small, flat screens delivering two-dimensional content that merely approximates social interaction or entertainment. Research published in Frontiers in Virtual Reality shows that immersive VR experiences create significantly stronger psychological rewards and higher addiction potential — between 2 and 20 percent of current VR users already show problematic compulsive use patterns.
Now extrapolate. A BCI-powered, AI-generated world that is indistinguishable from reality, in which you can be anyone, do anything, experience any sensation — this is not an incremental improvement over a smartphone. This is a category shift. It is, in the language of neuroscience, a superstimulus — a stimulus that exceeds what the brain encountered during evolution, hijacking the reward system with unprecedented efficiency.
The historical parallel is not social media. It is closer to the introduction of refined sugar, or opiates — substances that exploit neural reward pathways far more effectively than the natural stimuli the brain evolved to process. A world where billions of purposeless humans can retreat into a personalized, perfect synthetic reality will see adoption rates and addiction patterns that dwarf anything we have observed with current technology.
This is not science fiction. Every component of this scenario exists today in early form. BCIs are implanted in humans. AI generates interactive 3D worlds in real time. Large language models create dynamic narratives and respond to user input. The only remaining step is integration — connecting these components into a seamless system. At current rates of progress, this integration is plausible within 10 to 15 years.
The computational requirements of this scenario are staggering.
Today, AI data centers already consume an estimated 4.4 percent of U.S. electricity generation. The International Energy Agency projects that data center power demand will more than double by 2030. This is driven by current AI workloads — primarily text and image generation for business applications.
Now imagine the compute required to maintain individualized, real-time, physics-simulated 3D worlds for billions of simultaneous users, each world dynamically generated by a large language model, each user connected via a high-bandwidth neural interface. The energy requirements would be orders of magnitude beyond current projections.
The tech giants already understand this. Microsoft has committed $16 billion to restart the Three Mile Island nuclear power plant. Amazon has invested $20 billion in the Susquehanna nuclear facility. Google has contracted 500 megawatts from Kairos Energy’s small modular reactors. Meta has issued requests for proposals for 1 to 4 gigawatts of new nuclear generation. OpenAI has called on the U.S. government to build 100 gigawatts of additional power-generating capacity annually — equivalent to 100 nuclear reactors per year.
The critical insight for investors: traditional energy utility stocks may not be the optimal vehicle for capturing this trend. As demonstrated by Microsoft’s direct nuclear power acquisitions, companies of this scale are vertically integrating their energy supply. They are not buying electricity from utilities — they are buying the power plants themselves. The value accrues not to the middleman selling electrons, but to the entities that control the technology stack: the reactor builders, the uranium suppliers, and the tech companies themselves.
The brain-computer interface sector presents a classic early-stage asymmetric opportunity — the kind Peter Thiel describes as “secrets” that most investors have not yet recognized.
Direct BCI Investments: Neuralink ($9.7 billion valuation, private) is the market leader in invasive BCIs, with FDA-approved human trials and 21 implanted patients. Synchron (private, $200M Series D) is the leading minimally invasive BCI, with Apple and NVIDIA partnerships, potentially the first commercially scalable device. Merge Labs ($850 million valuation, private) is backed by OpenAI, pursuing non-invasive BCIs to bridge biological and artificial intelligence. Paradromics (private, $105M+ raised) has the highest demonstrated information transfer rates, backed by NIH and DARPA. Blackrock Neurotech is the technology behind the BrainGate system, a pioneer in the field.
Most of these companies are private, making direct investment difficult for individual investors. But the secondary effects are investable today.
Enabling Infrastructure: NVIDIA (NVDA) — GPU compute is the backbone of both AI world generation and neural signal processing. Every BCI company depends on NVIDIA hardware. Apple (AAPL) — Synchron’s partnership with Apple suggests that Apple sees BCIs as the next interface paradigm after the iPhone. Apple’s history of entering markets late but dominating them makes this partnership significant.
Energy and Nuclear: Cameco (CCJ) is the world’s largest publicly traded uranium producer. If the scenario described above materializes, uranium demand will increase dramatically. NuScale Power (SMR) is the first company to receive NRC certification for a small modular reactor design — a pure play on the SMR thesis. Constellation Energy (CEG) is the largest nuclear power operator in the United States, already signing power purchase agreements with Microsoft and other tech giants. Oklo (OKLO) is a Sam Altman-backed advanced nuclear fission company developing compact reactors specifically designed for data center applications.
Within the framework of the barbell strategy described earlier in this essay, BCI-related investments fit squarely into the asymmetric bet allocation. The downside is defined — early-stage companies can fail. The upside is not — if BCIs become the dominant human-computer interface, these companies sit at the center of a multi-trillion-dollar market that does not yet exist. Nuclear energy investments, by contrast, can serve as additions to the safe core — these are companies with tangible assets, regulated revenue streams, and growing demand regardless of whether the full BCI scenario materializes. AI will need power whether it is accessed through a keyboard, a voice interface, or a neural implant.
The progression is clear: from keyboard to voice to thought. Each step removes a layer of friction between human intention and machine execution. The keyboard translated thought into finger movements. Voice translated thought into speech. BCIs will transmit thought directly. And when that thought can conjure entire worlds — worlds that feel as real as this one — the implications are not merely technological. They are civilizational.
IX. Why Now — And Why Hesitation Is the Greatest Mistake
Howard Marks writes in “The Most Important Thing” that the best investments are often those that feel the most uncomfortable. Investing in AI, robotics, and crypto today feels uncomfortable. Prices are high. The technology is complex. The future is uncertain. That is exactly the point.
Those who wait until the thesis is “proven” — until humanoid robots stand in every factory and AI agents transact trillions on blockchains — will pay prices that have already captured the bulk of the returns. The asymmetric opportunity exists now, in the phase of uncertainty, when most people have not yet understood what is coming.
I see this every day in my work with accessibleAI. The companies adopting AI today are in the minority. Most are waiting, skeptical, afraid of the complexity. In two to three years, the same companies will be desperately trying to catch up — and paying multiples of today’s price for the transformation. What is true for companies is true for investors.
Peter Thiel has said: “Definite optimism” — the conviction that the future will be good and that one can actively work to shape it — is the only stance that leads to extraordinary outcomes. Neither blind optimism nor resigned pessimism, but a clear picture of where the journey is headed, coupled with the willingness to act accordingly.
The convergence of artificial intelligence, robotics, and cryptocurrencies is not one possibility among many. It is the most probable future. AI will provide the intelligence, robots will perform the physical labor, and crypto will be the lubricant of the machine economy. Whoever understands these three pillars and invests in them is positioning themselves on the right side of the greatest wealth transformation of our lifetime.
The question is not: Can I afford to invest now?
The question is: Can I afford not to?
This essay reflects the personal analysis and conviction of the author. It does not constitute investment advice. Every investment decision should be based on one’s own research and individual risk assessment.