Actionable insights — AIA Free Weekly Email 8.13.26

1. Screen for a "permanent" asset: non-replaceable, contract-escalated, capex-free — and management that prices its own optionality

1. Start with the asset, not the operating business. Ask what the company owns that cannot be replaced or reproduced by capital — assembled contiguous land, water rights, pore space, a permitted position. If a competitor with money could recreate it, it is not a permanent asset.
2. Check that revenue is structured as a royalty-like fee on somebody else's activity (a cut of volumes across or beneath the asset) rather than as an operating margin the company has to earn each year with people and equipment.
3. Read the contract language for two growth levers before touching a growth forecast: (a) an inflation escalator (CPI-linked) on multi-year terms, which sets a floor on revenue growth requiring zero capital spending; (b) re-pricing headroom — whether renewals should be signed above the current unit rate as demand for the scarce thing rises.
4. Count the uses the same acre can be sold to. Stacked, non-competing revenue streams on one asset (water, disposal, roads, transmission, generation, carbon capture, a data-center pad) is optionality; a single tenant type is not.
5. Underwrite management on one criterion: does it demonstrably know what it owns and monetize the optionality, rather than run the operating business well? This is the jockey test applied to an asset owner.
6. Only then look at price. A permanent asset that fails the escalator/re-pricing test is a cyclical in disguise.

• The escalator's index and reset frequency (a CPI-linked fee is only capex-free growth while CPI runs hot); renewal rates printing above the current unit price as evidence the re-pricing lever is real; announced non-water tenants (a signed power or data-center lease is what converts "optionality" into cash); the through-cycle question of whether volumes — and so the royalty base — depend on drilling activity that can fall; and, for Polomny specifically, the actual publication of the Permanent Portfolio list ("More to come on this"), which is where the screen becomes a set of holdings.

2. Run the same screen across land banks — power hosting plus controlled water

1. Take the archetype's monetization routes and state them generically, then go looking for other owners of the same input. Polomny does this explicitly: "many of these land banks."
2. Route one — hosting generation: large contiguous acreage can site power plants of any kind (renewables, gas turbine, coal, nuclear) for a recurring fee, without the landowner taking construction or operating risk.
3. Route two — controlled water: check whether the same landholding carries large water resources, because thermal plants and data centers both need cooling water. Water rights attached to the land are a second royalty on the same acre.
4. Rank candidates by contiguity and control (fragmented parcels can't host a campus), by proximity to load and transmission, and by whether the water is owned or merely nearby.
5. Do the diligence on the counterparties, not the landowner: the landowner's revenue is only as durable as the tenants signing the leases.

• Interconnection queue positions and behind-the-meter generation deals on landholder acreage; water-rights disclosures (owned volumes and permits, not adjacency); the first hosting lease at any land bank being priced publicly, which sets the comparable for the rest; and local/state water politics, which is the most likely thing to break the second royalty.

3. Treat an extreme sector weight as the trigger, not a valuation screen

1. Watch a sector's share of the index rather than starting from single-name multiples. Weight is a measure of ownership and crowding; it compresses valuation, sentiment and flows into one number.
2. When the weight reaches a historic extreme, say so as a probability statement about reversion, not as a price target: "when things are this skewed, one must wonder if a reversion to the mean is coming sooner rather than later."
3. Enumerate both resolutions before acting — the weight can normalize because the small sector's earnings rise, or because the crowded end of the index falls. A position that only works under one of the two is not the sector trade.
4. Express it through the sector's own scarcity case (his standing preference: the asset owner and the toll collector), so the trade doesn't depend on which resolution happens.

• The sector weight itself, monthly; energy earnings share versus market-cap share (a wide gap is the mean-reversion fuel); passive/index flows and sell-side coverage counts as crowding proxies; and — the falsifier — the weight staying pinned at the extreme through a full oil-price upcycle, which would say the de-rating is structural rather than cyclical.

4. Buy the necessity, not the forecast — Robotti's narrative-versus-reality gap

1. For any technology boom, split the universe into recognized beneficiaries (priced for the outcome) and the physical economy that must be built for any version of that outcome to happen.
2. Refuse the forecasting problem entirely: "one does not need to predict which model wins, or whether today's spending proves too much or too little." Underwrite only what is true across all the scenarios.
3. What survives that filter is demand for machines, materials, energy and infrastructure — and, critically, supply of those things "that cannot quickly respond." Inelastic supply is the second half of the test; rising demand alone is not enough.
4. Locate the gap: "enthusiasm concentrates in one place while necessity accumulates in another." Buy where necessity accumulates and where a decade of indifference is still embedded in the price.
5. Keep the historical prior attached: railroads, radio, conglomerates, the internet, housing — each was a real and durable change, and each taught that "the significance of a technology and the returns to its most celebrated stocks are two very different things."

• Evidence that supply genuinely cannot respond (lead times, permitting, capacity utilization, backlogs) rather than merely has not yet; the moment the ignored side starts re-rating, which is when the gap begins to pay; and the disconfirming case — a capex air pocket at the spenders would hit the "necessity" names' order books before it hits the celebrated stocks' multiples.

5. Define the edges of the circle, not its size

1. Write down what you actually understand well enough to price, and treat the boundary — not the area — as the thing to work on: "it is not about how large your circle is, but how well you define its edges."
2. Apply it as a refusal rule during manias: a stock you cannot value is a pass, regardless of how much it has moved or how confident the consensus sounds.
3. Re-check the edges when a business changes character (an asset-light compounder turning capital-intensive leaves the circle even if the ticker doesn't change).
4. Use the circle to explain your omissions to yourself in advance, so a missed winner doesn't get retroactively reclassified as a mistake.

• Your own drift — positions taken in businesses you cannot model, sized as if you could; and the tell that the circle needs redrawing rather than widened: a thesis that rests on someone else's letter instead of your own valuation.