The frame

Our standing view of what matters for people who own stocks: what we think is going on, what we are watching, and what would change our mind. Free, permanent, revised when the evidence asks. Every revision is dated in the log at the bottom. Nothing is quietly deleted.

Not advice. General commentary for a general audience.

What changed in this version

The core argument: debt finds its way out through inflation

One claim, with a long lineage: governments that owe more than they can comfortably repay find their way out through inflation, because inflation is the path of least political resistance.

The mechanism is political, not statistical. Debt builds through peacetime and faster through crises. In time, servicing it crowds out the choices governments want to make. When the choices narrow, inflation is the exit that hurts least today: softer than default, quieter than austerity, faster than growing out of it. Reinhart and Rogoff catalogued the pattern across eight centuries of financial crises ¹. Dalio's long-term debt cycle ² and Napier's work on financial repression ³ make the same argument from other desks. We anchor on the mechanism, not on any chart that seems to rhyme with it.

The case against, which we keep alive. Whether debt is sustainable may depend less on its size than on the gap between growth and interest rates. Inflation may be a monetary event, not a fiscal choice. And a government that issues the world's reserve currency has room the historical cases never had. Serious people hold all three views. If growth keeps outrunning interest rates and inflation falls without a fight, the spine is wrong for this cycle, and we will say so.

What makes this cycle different. Global debt has held just above 235 percent of world output while government debt has risen to nearly 93 percent ⁴. That debt is arriving alongside a productivity event of genuinely uncertain size. Inflation may still be the political way out. If the productivity arrives, real growth may carry part of the load that inflation carried in past cycles. The spine does not change. The texture of this rhyme might.

What sits under the argument

Computing power, the resource the others now depend on

Oil organized the twentieth century. Compute is organizing the twenty-first.

That one shift explains much of what we watch. Fiscal strain, because the build's capital needs are larger than the old system absorbs smoothly. Geopolitics, because access to chips, the machines that make them and the power that runs them is now rationed and contested, as access to oil once was. Capital, because capital follows the productive base, and the base is moving from energy-intensive to compute-intensive.

The loud story is substitution: machines doing work people did. The strategic story is constraint relaxation: compute applied at scale moves limits that were treated as fixed, in materials science, drug discovery, power engineering, industrial design and agriculture. The first story makes headlines. The second decides who compounds.

The build has moved its own bottleneck. The scarce input is no longer the chip; it is the power to run it and the grid connection to deliver it, which is why the frame reads power, not chips, as the constraint. Over a longer horizon, where that power comes from is itself changing, and we track it.

What spills out of the build

Compute is the resource; the waterfall is everything that spills out of the build and reshapes activity downstream. We hold this part more loosely, and if it is right, it matters most. Three currents.

The productivity lift, and its timing. The lift arrives in two stages. First come fast gains in a handful of tasks where the tools drop straight into existing work: writing code, answering customers, searching documents. Then a long stretch in which those early gains level off while the rest of the economy reorganizes around the tools: new processes, new skills, new kinds of firm. Electricity took decades to show up in factory productivity, because factories had to be rebuilt around the motor before the motor paid ⁵. Economists call the measured version a J-curve: firms spend on the change before the change pays, so measured productivity can lag, or dip, before it lifts ⁶. The frame expects the broad lift. It does not expect it on the build's timetable.

Programmable settlement. Stablecoins, tokenized funds and machine-to-machine payments are becoming plumbing underneath the compute layer. The GENIUS Act, signed in July 2025, brought that plumbing inside the federal regulatory perimeter; its rules take effect by January 2027 ⁷. In June we recorded both halves of the story: the rails matured while the assets riding on them fell ⁸. We weigh the plumbing, and we say plainly when the water level drops.

Hyperdiversification and the self-operated institution. Tokenized rails plus machine-built portfolios point toward individuals holding thousands of positions, continuously balanced to a stated tolerance for risk. Legal and administrative machinery that once needed a law firm or a bank is being compressed into tools a person can run. It reads like fiction. We name it because each piece already exists somewhere in working form, and we hold it as tracked, not forecast.

How sure we are, in three layers

Three layers, kept apart, because mixing them is the failure we guard against.

Layer 1: dated and falsifiable. Inflation prints, policy rates, the yield curve, recession signals, company reports. Each has a threshold, a release date and a reading that can prove us wrong.

Layer 2: structural, multi-year. Export policy, sovereign capital, where power and compute get built, defense reshoring, the minerals underneath all of it, and how fast the productivity lift broadens. These move over years and rarely reverse. We watch them through dated disclosures and review them on a longer clock.

Layer 3: tracked, never forecast. Where the build leads in five or ten years, what settlement looks like at maturity, the endpoint of the self-operated institution. We follow the live evidence and refuse to pretend we know the endpoint.

This year gave us the cautionary case. A war pushed up an inflation print, which is a Layer 1 event. Reading that print as proof of the spine would have been exactly the mixing error the layers exist to prevent.

Any transformative technology works in three orders, in the same direction. Substitution: the new thing does the old job, visibly and loudly. Constraint relaxation: limits that were fixed start to move, less visibly and decisively. New categories: things nobody could have described from inside the old regime; the suburb was a third-order effect of oil. We report the first, cover the second, and name candidates for the third without forecasting them.

Where things stand now

Since late February there has been a war with Iran, and it changed the kind of inflation we have. Hormuz, which carried about a fifth of global petroleum liquids consumption, has been mostly shut since late February. ⁹¹⁰ Washington says oil is still getting out under escort; independent trackers say less. ¹¹ In its August report the IEA counted 8.3 million barrels a day of Gulf output still shut in. ¹⁰ Brent's May high was $118.26 a barrel, on 1 May; it was $105.32 on 23 July and $96.92 on 21 August. ¹² Diesel set a record price in September. ¹³

Inflation comes in two kinds and only one of them is kind to equities. Demand-pull — too much money chasing goods — lets companies raise prices with everything else, and stocks keep up. Supply-shock — a blocked strait, an embargo — raises input costs that cannot always be passed through, and the 1970s say stocks do not keep up. We are in the second kind.

So the spine holds, but the hedge we drew from it — own productive businesses and you keep pace — is conditional on the kind of inflation. In a supply shock it holds only for the businesses that are the supply.

What is moving

What would change our mind

Each watchpoint is a dated claim that can be wrong. Layer 1 claims resolve on a release date. Layer 2 claims resolve on a longer clock, with a progress note at each review.

The same rows render on the claims board — one copy, two surfaces.

Claims made

7

Resolved

0

Confirmed

0

Refuted

0

Withdrawn

0

As of

10 Nov 2026

Dated and falsifiable

Structural, over several years

What the frame implies

A frame is not a list of names. It says which parts of the economy matter most to how this regime plays out, and that is where our coverage goes. None of what follows is a view on any company or security.

The substrate. The chips, the machines that make them, the data centers that house them, and above all the power and grid equipment that now set the pace of the build.

The defense industrial base. Geopolitical realignment runs on a longer clock than any earnings season, and the businesses tied to it run on that clock too.

Owners of proprietary data. Businesses sitting on data the public web cannot reach — legal, medical, financial, information services — stand to gain from the build rather than lose to it, because the tools extract more from data only they hold. The substitution story misses this. The constraint-relaxation story does not.

The settlement rails. Payment networks, tokenized funds and identity layers sit in the path of whatever shape the waterfall takes. We watch the plumbing, not the price of what flows through it.

Five areas the frame reaches.

• Space: the launch and orbital layer under compute's long-run power supply.
• Metabolic medicine: laboratory throughput is the productivity lift's nearest proof point, and it reports in pipeline data on a clock of quarters.
• Robotics and embodied systems: constraint relaxation crossing from software into the physical economy. The labor-substitution story will dominate the headlines; the capability story matters more.
• Critical minerals: the contest over compute is also a contest over gallium, rare earths, copper and the feedstocks under power itself — the substrate of the substrate. As of 14 October 2026, China's suspension of six of its 2025 export-control announcements runs until 10 November 2026 ²⁹.
• Monetary hedge and hard assets: the one area anchored to the spine itself. If inflation is the way out, official reserve behavior is where the mechanism shows first. The dollar's share of the world's official reserves has been falling for years ³⁰.

Where this leads

Three currents push in the same direction. More money in the system: the spine's own mechanism. More real output: the productivity lift, if it arrives at anything near the low end of serious estimates. More people reaching the system more cheaply: the settlement rails and the tools that compress what institutions used to do.

Together they mean more demand for what the economy is built on: productive businesses, the infrastructure beneath them, and scarce real assets. Over the regime horizon, that is a world in which asset prices rise. In money terms it is close to arithmetic: when inflation is the way out, the price of everything real goes up with it.

The harder question, and the one this frame exists to answer, is whether owning those assets keeps an owner ahead after inflation. That depends on which kind of inflation is running. When demand pulls prices up, businesses raise prices with everything else and broad ownership keeps pace. When supply breaks, as it has this year, input costs outrun what most businesses can pass on, and only the businesses that are the supply keep up. The long-run conclusion stands. The road to it runs through the current regime, and the current regime is the hard kind.

The conclusion holds on the regime's horizon, not the quarter's. On the near clock, spending on the build runs ahead of the lift it is meant to produce. On the long clock, the lift catches up.

This is a view about the economy over years, not a call on any security or on any quarter. What would prove it wrong: a full cycle in which inflation runs hot and real asset values fall anyway, as they did through most of the 1970s, which would mean inflation had become a tax on ownership rather than a tailwind for it. Or growth that keeps outrunning interest rates while inflation recedes without strain, which would mean the spine itself was wrong for this cycle.

What this page is not

Not personal advice. Not a buy list. Not a model portfolio. The publisher's holdings are disclosed on every piece. Every claim above is dated and names what would prove it wrong.

Changelog (newest first)

The changelog runs in the margin, beside the page.