← Analysis page  ·  Avi Salzman hub  ·  Research hub

Actionable insights — When a Permitting Freeze Marks Up the Incumbent's Inventory

Not "the backlash is bearish for AI power," but how to identify the asset a supply restriction silently revalues — separating the installed base from the pipeline, and finding the owners who were repriced by a correlation rather than an exposure.
2026-AUG-24 · Barron's · Avi Salzman · Read ↗ · full analysis · transcript
How to read this page: each insight is a method — the decomposition, the screen, the milestone to watch — distilled from the article so it can be rerun the next time a regulator restricts new supply in a booming industry. The boxed line shows how it played out here.

1. A restriction on new supply is a mark-up on the existing stock — find who is already inside the fence

The repeatable method
  1. Write the rule down as precisely as the regulator did. Note whether it acts on new applications or on operating assets. Almost all political interventions in an unpopular boom act on the former, because it is cheaper and less litigable.
  2. If the rule acts on new entry only, the licensed/connected installed base is now a closed set. Its scarcity value rises by exactly the amount of entry that has been blocked.
  3. Screen for owners of that installed base who are not classified in the affected sector — the licence sitting inside a company the market files under a different theme is the one nobody has repriced.
  4. Verify the asset is transferable to the blocked use (can the incumbent actually rent, resell, or convert it?), otherwise the scarcity is real but uncapturable.
Here: governors are pausing approvals for new data centers looking to hook into the grid — not switching off existing interconnections. The owners already inside the fence are the Bitcoin miners, who queued for those connections years ago for an unrelated reason. Byrd: "most of the miners already have access to the electric grid at their existing sites… being connected to the grid is worth more money now than ever before." Names: CIFR, HUT, GLXY, MARA, RIOT.
Watch for

2. Split every "affected" company into installed base and pipeline, and value them separately

The repeatable method
  1. For each name exposed to the rule, put its capacity into two buckets: energised / permitted and announced but still requiring approval.
  2. Apply the regulatory discount only to the second bucket. Applying it to the whole company is the standard market error during a sector-wide selloff.
  3. Rank the sector by the ratio of bucket one to bucket two. High ratio = protected and revalued; low ratio = genuinely impaired despite the same headline.
  4. Check geographic overlap between bucket two and the states that have actually moved — a pipeline in an unaffected state is not discounted at all.
Here: the article keeps the caveat rather than burying it — "while some new Bitcoin-to-AI projects in states like Texas may see delays, most Bitcoin miners are still in a strong position to convert their warehouses for AI use." That single sentence is the split: the conversion of existing warehouses is safe, the new Texas projects are not. RIOT's Texas concentration is where the distinction matters most.
Watch for

3. When a group rerates into a theme, ask whether it inherited the theme's risk or only its beta

The repeatable method
  1. Identify groups whose market classification changed recently (a miner that became an "AI power" name, a utility that became a "data-center" name). The reclassification is usually done by flows, not by analysis.
  2. When the theme sells off, test the specific risk against the specific company: does the shock hit a line item this business actually has?
  3. Where the shock's transmission channel is absent, you have a correlation-driven drawdown — a dislocation rather than a repricing.
  4. Size to the gap between the drawdown and the fraction of value genuinely at risk, and set a horizon over which the market must re-separate them.
Here: "As miners shift to the data center model, their stocks have traded like AI names" — so when AI power stocks fell, they fell too: the WGMI ETF down 16% over the past month. Byrd's whole call is that this is inherited beta, not inherited exposure: "investors are misunderstanding the risks."
Watch for

4. For every rule, name the loser and the transferee — value is redistributed, not destroyed

The repeatable method
  1. Write the rule as a transfer: it takes economics from party A and gives them to party B. Refuse to stop at party A, which is where the headlines stop.
  2. Party A is normally whoever needed the blocked approval to monetise an asset; party B is whoever already holds a substitute the blocked party must now buy or rent.
  3. Confirm B can charge for it — a substitute with no market (or a regulated price) captures nothing.
  4. Express the view as the pair, not the single leg; the pair is insulated from being wrong about the underlying theme's direction.
Here: losers are the power-plant owners whose data-center contracts depend on the projects being approved — CEG and NRG, "hurt" by the backlash. Transferees are the owners of connected sites who can rent "the power or the warehouses themselves" to tech companies. Same rule, opposite signs — the identical structure to Aug 19's TLN/VST/PEG versus PPL split.
Watch for

5. Re-underwrite a cyclical when its revenue converts to contracted rent

The repeatable method
  1. When a commodity-exposed company begins selling a contracted service off the same asset base, treat it as a change in the quality of cash flow, not just the amount.
  2. Ask what fraction of gross profit is now contracted, at what tenor, with what counterparty credit. That fraction deserves a different multiple from the commodity remainder.
  3. Beware the trap the reclassification creates: a higher multiple also means a higher-beta shareholder base, so drawdowns get larger even as the business gets steadier.
  4. Use the mismatch — steadier fundamentals plus a jumpier register — as the source of entry points.
Here: "The AI business model can generate steadier returns than Bitcoin, insulating the miners from the ups and downs in crypto prices." The business got less cyclical, and the stock got more volatile relative to a different index — which is precisely the setup that produced a 16% monthly drawdown on a political story with no direct transmission channel.
Watch for

6. Prefer the beneficiary whose advantage survives the rule being reversed

The repeatable method
  1. Ask how the restriction was created — moratorium, audit, executive order — and how easily it can be lifted after an election.
  2. Then ask what the beneficiary would still own if it were lifted tomorrow. A permit, an interconnection, a reserved turbine survives repeal; a pure price umbrella does not.
  3. Prefer beneficiaries whose asset was scarce for structural reasons before the rule (multi-year queues, physical constraints) and which the rule merely made scarcer.
  4. Then the policy is the catalyst, not the thesis — and the reversal risk is capped.
Here: interconnection queues were multi-year before any governor acted; the moratoria only make the wait longer. So a miner's connection is worth more with the freeze and still worth a great deal without it. Byrd's phrasing points at the durable version: cancellations and delays "just makes these companies' sites more valuable" — the site is the asset, and no election takes it back.
Watch for

Methods distilled from the public Barron's article (full text in transcript.txt) for personal study. Not investment advice. © Barron's / Dow Jones for source material.