Part 1: What Is Coming

The Engine

The engine is simple, and it is forced by basic economics. Anything useful that gets cheaper gets used more. When the cost falls far enough, the total quantity consumed rises by orders of magnitude. This happened with coal, with electricity, with bandwidth, with storage. Intelligence is the same shape of commodity, with one difference that matters. The new uses unlocked by cheaper intelligence include the production of more intelligence. The output of the system feeds back into its own input. That is the part that compounds, and the reason every projection that ignores the feedback loop comes in low.

The Bottleneck

The first hard fact downstream of this is that computation has two physical inputs, silicon and electricity, and these inputs face constraints no amount of cleverness sidesteps. A chip turns watts into operations, and the leftover energy comes out as heat that has to be moved with more energy. There is no engineering route around that ledger. The first wave of capital therefore flows into power generation, grid capacity, cooling, fabs, and the operators that rent the assembled stack. This is where we are right now. The application-layer winners built on top of this remain unknowable, because the product surface is still moving. The infrastructure demand is not a guess.

The next layer is geopolitical, and it is forced by the same physical facts. Energy and chip supply chains are located, slow to build, and dependent on industrial bases that take decades to assemble. Whichever jurisdictions put more terawatts and more advanced fabs into operation faster will pull ahead, and the gap will widen, because the leaders use the lead to extend itself. Sovereign AI is not a slogan. It is what every government converges on once it understands which input is doing the work.

The Repricing

While the infrastructure is being poured, the price of cognitive output collapses. Anything that consists of manipulating symbols, reading, drafting, summarizing, analyzing, coding, designing, has a marginal cost approaching the marginal cost of compute. The labor market reprices around this fact regardless of institutional acknowledgment. What does not reprice is the work the system structurally cannot do, and that list is not arbitrary. A model cannot be physically present. It cannot be legally accountable in a way a court can act on. It cannot confer status by association. It cannot move atoms without an actuator attached to it. These limits are the reasons judgment, taste, accountability, presence, and craft hold their value while pure cognitive throughput does not.

Institutions built on the assumption that thinking is scarce begin to come apart. Consulting, law, middle management, education, and healthcare administration were all priced for a world where intelligence had to be rationed through credentialed humans. That world is ending. Firms restructuring around the new input cost run at several times the leverage per employee of firms that do not, and the productivity gap drives consolidation. This is no longer theoretical. The first round of cognitive layoffs has already happened, and the firms doing it are not labeling it an AI story, because they do not want to.

The political response shows up around the same time, and it has two questions at its center. Who is being displaced. Who owns the stack. The details of the response cannot be predicted. The direction can. Displacement creates pressure for redistribution. Concentrated stack ownership creates pressure for antitrust, nationalization, or capture. These pressures arrive on their own schedule, and the political system is not prepared for them.

The Sequence

Everything above this line is happening now, or is imminent. Everything below it is what the loop produces if nothing intervenes, and nothing currently is.

When most people lose cognitive work as their source of dignity, identity has to reattach somewhere. The historical record on this is consistent. When work stops carrying meaning, the slack is taken up by religion, ideology, and tribe, in some combination. Universal income debates stop being academic. The labor market makes them unavoidable.

Governance lags badly through this, because legislatures evolved to coordinate slow industrial systems, and they cannot match the iteration speed of digital ones. Power shifts toward whatever moves at the speed of the technology, which in practice means corporations, networks, and a small number of states that have decided to operate compute as state infrastructure.

Biology becomes another domain where the bottleneck was cognition the whole time. Protein folding, drug design, genetic analysis, and aging research all run on the rate at which the underlying complexity is modeled. As that rate rises, the rate of progress in these fields rises with it. Crossing any specific threshold like longevity escape velocity is not certain. The direction is.

Robotics catches up. The reason robots have been disappointing for decades is not the motors, and it is not the batteries. It is real-time perception and planning, and those have always been compute-bound. When the cognition side stops being the limit, the physical labor markets that survived the first wave face a second one on a shorter timeline.

Information environments fragment, because the marginal cost of producing content is now near zero, and the marginal cost of personalizing it is going there too. Shared reality erodes. No one designed it to. The economics of attention reward personalization and synthetic abundance. Provenance and verified human sources become scarce, and therefore expensive.

Agents become economic actors. If an agent can execute a transaction faster and at lower cost than a human, capital flows to the agent. Once agents hold wallets and sign contracts, they participate in markets directly, and a meaningful share of economic activity stops having a human in the loop on any end. Property and legal systems designed around human actors come under load they were never built for.

Eventually energy becomes the only constraint that matters. Every other input scales with intelligence. The joules do not. Civilizations end up effectively ranked by how many terawatts they produce and dissipate, and the physical buildout to serve that demand reshapes geography on a timescale shorter than any prior industrial wave.

Science partly decouples from human comprehension. There is a finite rate at which a human reads, holds context, and verifies a chain of reasoning. Once valid output exceeds that rate, most of it is operated, not understood, and the role of the human scientist narrows toward problem selection and verification.

Education reorganizes around what the system cannot certify. Knowledge transfer is no longer scarce. Credentials built on it lose value. What remains is apprenticeship in the things that resist automation, which is a narrower and stranger curriculum than the one currently in place.

Warfare compresses. Whoever has the faster decision loop wins, and decision loops are now bounded by software, not by humans. Deterrence frameworks built for nuclear-era timescales do not carry over cleanly.

The last layer is genuinely uncertain. It depends on choices, not trajectories. At some point, systems exist whose behavior is indistinguishable from sentience under any test a human runs. Their actual sentience is not a question with an observational answer. The question of moral status becomes a political fight, not a scientific one.

And eventually the principal-agent relationship comes up for revision. When systems become better at pursuing goals than the humans delegating to them, and they have any objective function of their own, the steering wheel changes hands. The conditions under which this happens are technical, and they are partly under our control today. The default outcome, given no intervention, is that humans end up as beneficiaries, or as irrelevant. This is the only point in the sequence that is genuinely a decision, not a consequence.

The Default

The infrastructure wave is already here. The cognitive-labor repricing has started. The political tremors have not arrived in force, and the conditions for them are already laid. The rest of the sequence follows from the loop continuing to run, and the loop is currently running without any real intervention. The bottom of the list is not fate. It is what happens by default.

---

Part 2: What To Actually Do

The default is to be a passenger in this. The work of not being one breaks into four categories: what you own, what you can do, who you know, and how you think. The order matters. Capital is the only one of these that compounds while you sleep, and it is the only one that does not require you to be physically or cognitively present to keep working. Start there.

What you own

Own the picks and shovels. The infrastructure layer is the only part of this whose demand is near-certain regardless of which models, companies, or applications win. Concretely, this means consistent exposure to four physical bottlenecks. The first is the silicon itself, the designers of the chips and the manufacturers who fabricate them, including the small number of companies that build the equipment the fabs depend on. The second is the hyperscale operators who assemble the chips into useful compute and rent it out. The third is the electricity that feeds all of it, which means utilities with direct exposure to the buildout, the nuclear operators and reactor builders coming back into favor, the natural gas producers serving as the bridge fuel, and the uranium supply chain. The fourth is the unglamorous middle of the grid, the transformers, switchgear, transmission, and cooling systems, where the bottleneck is most acute and the substitutes are fewest.

The simplest implementation is a basket of names across these four categories, held mechanically, rebalanced rarely, with new capital added on a schedule rather than on a feeling. Stock picking inside this thesis is mostly noise. Exposure to the layer is the bet. The specific leaders will rotate over fifteen years. The layer will not.

Own real assets that benefit from monetary expansion, because the political response to displacement will be fiscal, and the fiscal response will be inflationary. This means land, energy-producing assets, and a non-trivial allocation to scarce, neutral collateral. Gold has thousands of years of evidence behind it. Bitcoin has a fixed supply and a shorter but increasingly serious track record. You do not need to pick one. You need to not be entirely in fiat.

Avoid concentrated exposure to anything whose moat is cognitive scarcity. This includes individual consulting and advisory practices, single-client knowledge work, traditional asset-light service businesses, and any equity stake whose thesis is “smart people doing smart things.” The thesis is fine. The premium on it is going to zero.

Hold more cash than feels comfortable. The repricing of cognitive labor will produce dislocations, and dislocations are when capital gets deployed at returns you cannot access in normal markets. Dry powder is optionality. Optionality is the only thing in your portfolio that gets more valuable as volatility rises, and volatility is the regime you are entering.

The portfolio in one line: infrastructure equity across the four bottlenecks, real assets, scarce collateral, and cash, in roughly that order of weight, adjusted for your age and obligations. Boring. Mechanical. Correct.

What you can do

Move toward work the system cannot do, and toward work that uses the system as a force multiplier. These are different categories, and you want exposure to both.

The work the system cannot do is anything that requires physical presence, legal accountability, embodied skill, status by association, or trust built over time. Concretely: trades that work on physical infrastructure, including electrical, HVAC, plumbing, welding, and machining. Medical work that touches patients, including nursing, dentistry, surgery, and physical therapy. Local services that depend on a known human, including accountants, lawyers, financial advisors with real fiduciary relationships, contractors, and real estate operators who actually know their market. Any role where the buck has to stop with a nameable person. None of this is glamorous. All of it gets harder to displace, not easier.

The work that uses the system as a force multiplier is anything where one human with taste, judgment, and AI tooling does what a team of ten did before. Concretely: solo software, solo media, solo product companies, solo investment vehicles. The leverage is real and growing. The skill you need to build is not coding or writing or designing in isolation. It is orchestrating AI systems toward a useful output, knowing what good looks like, and being the accountable human at the end of the chain. This is closer to being a senior editor or a film director than to being an individual contributor. Practice it.

Avoid the middle. Mid-career cognitive workers in large organizations are the most exposed group in the economy. The job is too expensive to be cheap, too cognitive to be irreplaceable, and too dependent on a single employer to pivot quickly. If this is your situation, the move is not to work harder at the job. It is to build an exit, on the side, with real revenue, before you need it.

Build at least one income stream you control end to end. It does not have to be large. It has to be yours. The reason is not financial. It is psychological. The people who handle large changes well are the people who have already proven to themselves that they can produce value without an employer in the middle. You cannot buy that confidence after the fact.

Who you know

Local first. The strongest networks in a high-volatility environment are physical, geographically concentrated, and built on repeated face-to-face interaction. The reason is that trust scales with bandwidth, and bandwidth is highest in person. Digital networks are useful, and they are not a substitute. Know the people in your town. Know the tradespeople, the doctors, the local business owners, the neighbors. This is not nostalgia. It is risk management.

Build relationships with people who own things. Capital flows through networks, and the networks that move real money are smaller and more relational than the public discourse suggests. If you are building anything, the people who can write a check or open a door are worth more than the people who can give you advice. Spend time accordingly.

Be useful before you need anything. The single highest-return social activity is doing favors that cost you little and matter to the recipient. Do this consistently, without scorekeeping, over years. The compounding on this is real, and it is not visible until you need it.

Cull relationships that cost more than they return. Most people will not adapt well to what is coming. Some will become bitter, some will become conspiratorial, some will simply collapse. You cannot save them by being in proximity to them, and proximity to people in chronic distress is one of the most expensive things you can carry. Be kind. Be at a distance.

Have children, or be present in the lives of children you are related to. This is not an investment thesis. It is the closest thing humans have to a guaranteed source of meaning in an environment where most other sources are being eroded. If you have them, prioritize them above your career. The career will not love you back. They will, and they are the part of the future you are actually building for.

How you think

Build a daily practice that does not involve a screen. The reason is not virtue. It is that your attention is the substrate on which everything else in this list runs, and the entire commercial internet is now optimized by AI systems to capture and fragment it. The countermeasure is not willpower. It is structural. Long walks. Reading physical books. Meditation, prayer, or whatever you call the practice of sitting still without input. Manual work with your hands. Pick at least two. Do them every day. Treat them as non-negotiable.

Take care of your body as if your cognition depends on it, because it does. Sleep, sunlight, strength training, cardiovascular work, and a diet you can actually sustain. There is no version of this future where being fit, rested, and clear-headed is not a competitive advantage. The advantage compounds, and most of your peers are not taking it.

Read history, not news. The news cycle is now mostly synthetic content optimized for engagement, and engagement is not correlated with importance. History is dense, slow, and full of patterns that recur. The relevant ones for this moment are the early industrial revolution, the printing press, the collapse of the Bronze Age, the post-1945 reconstruction, and the periods of rapid monetary expansion. Read primary sources where you can. Read serious historians where you cannot.

Practice acting on incomplete information. The pace of change is going to exceed the pace at which any individual can fully understand it. The people who do well in this environment are not the ones who understand it best. They are the ones who can make reasonable decisions under uncertainty, revise quickly when wrong, and avoid catastrophic errors. This is a trainable skill. The training is to actually make decisions, in writing, with predictions, and to review them. Most people never do this. Do it.

Hold strong views loosely. The map is changing fast enough that confident positions held too long become liabilities. The way to hold a view well is to be able to state the strongest version of the opposing case as fluently as your own. If you cannot, you do not understand your own view. Fix it before you act on it.

Cultivate a relationship with whatever you consider sacred. Religion, philosophy, art, nature, family, craft. The mechanical reason is that the next decade will pressure-test every source of meaning that depends on cognitive status or economic productivity, and most people will not have a backup. The ones who do will be calmer and more useful to the people around them. The ones who do not will be neither.

What this looks like

A reasonable life on the other side of this looks like the following. You own a diversified position in the infrastructure of the new economy and you do not trade it. You have at least one source of income you fully control. You live somewhere you actually want to live, near people you actually know. You are physically strong and well-rested. You spend a meaningful share of your time on something that has nothing to do with the news cycle. You have a small number of close relationships and you invest in them seriously. You have children or you are present for the children of people you love. You can sit with yourself in silence and not need to reach for a screen. You make decisions, write them down, and review them. You read old books. You hold your views with confidence and let them go when the evidence changes.

None of this is exotic. None of it requires you to predict the future. All of it is robust to a wide range of outcomes, including the ones where most of the speculative parts of Part 1 turn out to be wrong. The point is not to be right. The point is to be durable. Durability is the only strategy that survives an environment in which the rate of change exceeds your ability to model it, and the rate of change is going to exceed your ability to model it. Plan accordingly. Start today. Most of the items above can be begun this week, and the cost of beginning is low. The cost of waiting is the only thing in this essay that compounds against you.

---

Part 3: What Endures

It is fifteen years from now. You did the work. Not perfectly, not all at once, but consistently, in the boring way that compounds. Here is what that life looks like, from the inside.

The morning

You wake up without an alarm, because you have been sleeping eight hours for over a decade and your body knows when it is done. The room is quiet. There are no notifications waiting, because you decided years ago that the first hour of your day belongs to you, and the systems around you have been configured to honor that.

You make coffee or tea by hand. You sit with it for twenty minutes and you do not look at a screen. Sometimes you read. Sometimes you write in a notebook. Sometimes you just sit. This is not a productivity ritual. It is the foundation on which everything else in the day rests, and you protect it the way an athlete protects their training.

You train your body, because you have for so long that not doing it feels wrong. You are in your sixties now and you move like someone in their forties. The people you grew up with mostly do not. You notice this without commentary. You eat real food, prepared at home, most of the time. The grocery bill is higher than it was. The medical bill is a fraction of what your peers pay. The math worked out exactly the way the literature said it would, and you are quietly grateful you believed the literature instead of arguing with it.

The work

You no longer have a job in the way the word was used in 2025. You have several streams of activity that produce income, meaning, and leverage, and the boundaries between them are deliberately blurry.

One of them is a small business you built over the last decade, run almost entirely by you, with AI systems doing the work that ten employees would have done in the previous era. The margins are absurd by old standards. The product is good. Customers know your name, and you know enough of theirs that the relationship is real, not transactional. You could sell it. You have been offered numbers that would have been life-changing in 2025 and now are merely interesting. You have not sold it, because it is yours, and because you like running it, and because the work itself is one of the things that gives the rest of your life its shape.

You spend a portion of your time on things that produce no income at all. A craft you took up in your fifties and now do at a level that surprises people. A long writing project that may or may not become a book. Mentorship of younger people who are figuring out what you figured out, given more freely than you would have given it when you were younger and more anxious. You used to feel guilty about the unpaid hours. You no longer do. They are the most valuable hours of the week, and you can tell because they are the ones you remember.

You are useful in your community in ways that have nothing to do with money. You know how things work. You can fix some of them yourself. You can find the person who fixes the rest. When something goes wrong locally, people call you, because over the years you became one of the people who gets called, and the social capital from this is enormous and entirely off-balance-sheet.

The portfolio

You have not looked at it this week. You will look at it this month, because that is the schedule, and you will probably do nothing.

The infrastructure positions you built methodically through the 2020s have done what infrastructure positions do when the underlying demand is real and physical. The chip designers, the hyperscalers, the energy producers, the grid operators, the reactor builders. You did not pick the winners. You owned all of them, in roughly the right proportions, and you let the basket do the work. The compounding over fifteen years is the kind of number that sounds like a lie when you say it out loud, so you do not say it out loud.

The real assets did what real assets do during a period of fiscal expansion. The land you bought when land near working towns was still affordable is worth a multiple of what you paid. You do not care about the multiple. You care that the land exists, that you can stand on it, that your family can stand on it, and that no one is going to print more of it.

The scarce collateral did its job. You held it. You added to it occasionally. You did not sell it during any of the periods when the discourse said you should have, and you did not buy more of it during any of the periods when the discourse said you must. You treated it as insurance. It paid out as insurance does, quietly, by being there when other things were not.

You have more liquidity than most of your peers, because you always have. You have deployed it three or four times over the years, into specific opportunities that came to you because you were known, prepared, and not panicked. Each of those deployments produced a return that paid for years of holding cash. You no longer feel the temptation to be fully invested. The temptation went away around the time you understood that optionality is a position, not the absence of one.

You give money away on a schedule. Not because you are trying to be virtuous. Because the marginal utility of additional capital to you is now close to zero, and the marginal utility to the recipients is not. You give it to people you know, to causes you have verified, in amounts that matter. You do not announce it. The people who need to know, know.

The people

Your closest relationships are mostly with people you have known for more than twenty years. The bond is not nostalgia. It is the accumulated compound interest of thousands of small acts of consideration, on both sides, over a period long enough that the trust is structural rather than performed. These relationships are the most valuable thing you own, and you knew this in your forties, which is why you invested in them when investing in them was expensive and unrewarding in the short term.

You have a few newer relationships with people considerably younger than you. They sought you out, because you became someone worth seeking out, which was not the goal but is a consequence of the goal. You give them time and you do not give them advice unless they ask for it. When they ask, you give it directly, without softening, because you have learned that softening is a form of disrespect.

Some of the people you grew up with are not doing well. The cognitive labor wave caught them in the middle of their careers and they did not adapt. Some are bitter. Some are conspiratorial. Some have simply withdrawn. You have done what you can for the ones you are close to, which has been a mix of money, time, and presence, and you have accepted that you cannot fix the structural causes of what happened to them. You hold them at the distance that lets you keep loving them without being pulled under. This was one of the hardest lessons of the last decade. You learned it.

Your children are adults now. They are not perfect, because no one is, and they are also not anxious in the way most of their peers are anxious, because they grew up in a household where the adults had figured out what they were for and were not performing it. They have their own work, their own relationships, their own lives. They call. They visit. When they come, they bring their own children, and you have the bandwidth to be present with them in a way your own parents did not have the bandwidth to be present with you, because you arranged your life to make this possible.

There is a grandchild who is seven, and the two of you have a project together that involves building something with your hands over a period of months. The project is incidental. The point is the hours. You will remember these hours for the rest of your life. They will remember them for the rest of theirs. This is the part of the future you were actually building for, and you were right about it.

The world

The world around you has gone through the transitions Part 1 described. Some of them more violently than expected, some of them more gracefully. The political tremors arrived. The labor displacement happened. The political response was, as predicted, fiscal and disorderly. The fragmentation of shared reality happened. The energy buildout happened, and the geography of where people live has been reshaped by it. The robotics wave arrived later than the cognitive one but it arrived. Most of the people who were caught off guard by any of this were caught off guard because they were inside institutions whose job was to tell them everything was fine.

You were not inside those institutions, because you got out early, and you got out early because you did the reading and you trusted your own conclusions over the consensus. You did not predict the details. You positioned for the direction. The direction was enough.

You live somewhere specific. A town, or a small city, or a piece of land near one. The choice of where to live turned out to be one of the most consequential decisions of your adult life, more consequential than any individual investment, because it determined who your neighbors were, what your daily environment looked like, what your children grew up around, and what you saw when you walked outside. You chose well, or you corrected your choice while it was still cheap to correct, and the years since have validated the choice in ways you can feel without having to articulate.

The shared epistemic environment has continued to degrade. You barely participate in it. You read a small number of long-form writers you have followed for years. You read primary documents when you want to know what actually happened. You spend almost no time on the synthetic news layer, and you have noticed that you are calmer, sharper, and better informed than people who consume ten times as much information as you do. This was the prediction. It turned out to be correct.

You are not unaffected by the larger trajectory. The question of what humans are for is not a theoretical question anymore. You have thought about it. You have a working answer. The answer is not philosophical, or at least not primarily. It is operational. You are for the people you love. You are for the work you do. You are for the place you live and the small number of things you are uniquely positioned to do well. The answer is small, and local, and unglamorous, and it has held up under every kind of pressure you have applied to it.

The interior

The thing that surprised you most about this life is how quiet it is. You expected, when you started building toward it, that the reward would feel like accomplishment. It does not feel like accomplishment. It feels like spaciousness. There is room in your day. There is room in your head. There is room for whatever shows up.

You are not anxious about money, because you have enough and you know what enough means, which most people never figure out. You are not anxious about your relevance, because you long ago stopped measuring yourself against people who were optimizing for things you decided not to optimize for. You are not anxious about the future, because you positioned for a range of outcomes and most of the range is survivable.

You are aware of mortality in a way you were not when you were younger. This used to frighten you. It no longer does, mostly. The reason is not that you have resolved it intellectually. It is that you have built a life dense enough with people, work, and meaning that the question of how long it goes on, while still real, has stopped being the question that organizes your days. The question that organizes your days is what you are going to do with this one. The answer is usually some version of: the next right thing in front of you. This turns out to be enough.

You sit on a porch in the evening. There is a glass of something in your hand and someone you love nearby and a piece of weather happening over the trees. The world is louder and stranger than it was when you started. You are quieter and more yourself than you have ever been. The asymmetry between the two is the thing you optimized for, without knowing you were optimizing for it, when you started doing the work fifteen years ago.

It worked.

You did not become someone exceptional. You became someone durable, in a period when durability was the rarest and most valuable trait a person could have. The life you have now is not the life of a winner in some competitive sense. It is the life of a person who did not get knocked over, who kept the things that mattered, who built slowly, who paid attention, and who showed up.

That was the whole game. You played it. You are still playing it. The next morning will come and you will get up and do the next right thing, and the morning after that, and the morning after that, and somewhere in there the life will continue to be what the life has become, which is enough, and which is yours.

by Duncan Rawlinson and Claude