Actionable insights — Building North America's #1 Independent Gas Storage Company

44:35 1. The four-part barrier-to-entry test — and the fourth part is the one nobody models

1. Geology. Ask whether the physical substrate is genuinely rare. Usually it isn't: "in Alberta, for example, there's probably another dozen reservoirs that have the quality of our reservoirs." Geology alone is never the moat — strike it off the list first.
2. Location. Is it near a market that needs it, and how far is it from a major transmission line? Distance converts directly into the next cost.
3. Pipeline cost + permitting. Add the connecting pipe to the reservoir/drilling cost. Cost overruns on pipelines have been running "at a high pace" for 5–8 years, and the regulatory process to build one is itself a hurdle. Still surmountable.
4. Interconnect availability — the real barrier. Ask the question nobody asks: can the project interruptibly get molecules onto and off the pipeline? End users have contracted the delivery white space; producers have contracted the receipt white space. If neither is free, the project cannot guarantee performance, so no proponent will backstop it and no board will sanction a multi-hundred-million spend.
5. Generalise: for any infrastructure asset, separate the buildable constraints (geology, capital, permits) from the contractual constraint (someone else already owns the throughput). Only the last one is durable, because it can't be solved with money.
6. Test the moat's honesty by looking at where new supply is being built. If it's a different, lower-quality format serving a different need, the incumbent's claim survives.

• Any asset whose competitors' expansion announcements are in a different format or geography than the incumbent's — that's the tell the moat is real, not rhetoric.
• Conversely: the first credible greenfield depleted-reservoir sanction in the WCSB, or a pipeline expansion that frees receipt/delivery white space, would start eroding the fourth barrier.

1:05:58 2. Separate the commodity from the service — when a low price is a tailwind

1. For any "commodity-adjacent" business, write down what the customer is actually buying. Here it isn't gas — it's access and timing. Revenue comes from seasonal spreads and an insurance premium, not from price level.
2. Ask what a low price does to inventory behaviour, not to revenue. Low prices mean gas stays in the ground; gas in the ground can be "transact[ed] on that same molecule over and over and over again with no risk."
3. Ask what a high price does. Withdrawal creates an obligation to refill — and if you can't guarantee interruptible injection space, that obligation is an uncovered risk. "We don't go and forward hedge that risk. We'll only inject when we know we physically have that opportunity."
4. Follow the chain forward one full cycle: full storage this year → low prices into winter → low prices into next summer → wider intrinsic (summer/winter) value as you exit the winter.
5. Then check the customer side. Low prices let customers buy "below variable cost" molecules — a line item nobody budgeted five years ago — so the customer's economics improve at the same time, which is what makes the demand durable rather than opportunistic.

• Any business the market prices as a commodity proxy that is actually paid for optionality — screen for correlation between the share price and the underlying commodity that shouldn't exist.
• The tell that the framing is genuine: management refuses to hedge, because they have no obligation to deliver a price.

40:55 3. Decompose the book into intrinsic value vs insurance (non-intrinsic) value

1. Split the asset's earning power in two. Intrinsic value = the observable summer/winter spread — the part any analyst can compute off the forward curve. Insurance (non-intrinsic) value = what a customer pays above that spread for the right to act when something goes wrong: the value of the call and the value of the put.
2. Track which half is growing. Insurance value expands with realised volatility, not with the level of the curve — so a flat curve with fat tails is a rising-value environment even though the spread looks unchanged.
3. Price the tails off actual prints, not models: Chicago at $70 last winter, $1,000 in Oklahoma and Texas, $60 in California. If a well-piped market with every economic incentive still couldn't attract molecules, the whole continent is vulnerable and the insurance bid is structural.
4. Use the split to time contracting. If intrinsic is fine but insurance value is still below where you think it's going, refuse the long-term contract — sign short and re-price later. If insurance value is already rich, lock it in.
5. Accept the cost of that choice: staying short-dated is exactly why the market pays you a lower multiple. Decide which you want.

• Management that can quantify why its multiple is depressed and names the contract-mix fix with a date — then check the take-or-pay percentage each quarter against that glide path.
• Winter price prints in well-supplied hubs; each one raises the insurance premium the whole industry can charge.

49:57 4. Read the revenue split as three different risk businesses — and test the third for optionality vs obligation

1. Bucket the revenue by what creates it, not by segment name. Here: take-or-pay (paid regardless of use — de-risks the operator, gives the customer flexibility), short-term storage (a bank leaves its own physically-collateralised gas in the ground; the operator captures 90–95% of the calendar spread — "effectively a financing"), and optimization (~15%).
2. Interrogate the third bucket, because that's where a trading loss hides. The question to ask: is the position an option or an obligation?
3. Compare against the analogue everyone distrusts — a midstream marketing wedge. A midstreamer's inlet substances never match its outlet substances, so it is forced into longs and shorts (long transport, long rail cars, short butane). Involuntary exposure — which is why the market discounts it.
4. Contrast the structure here: reserved operational space that 99% of the time isn't needed becomes a daily choice to inject or withdraw. "We never have an obligation to do anything with that business… every single day we're long the put and every single day we're long the call." No open position carried from one period to the next.
5. Verify with realised results across a window containing no tail events: if the option book earned an equivalency with the contracted terms without a black swan, the income quality is high and the tail is free upside, not the whole story.

• Any "optimization" or "marketing" line in an infrastructure P&L: ask whether the imbalance is structural (obligation, discount it) or reserved capacity (option, credit it).
• A quarter where the optimization line spikes — that's a tail event, not a run-rate; and a quarter where it goes negative would falsify the no-open-positions claim.

39:22 5. The read-across template — find a market that already ran the experiment ten years ago

1. Identify the structural shock arriving in your market (here: LNG export demand connecting a domestic basin to international price signals).
2. Find a market where the same shock already landed, and date it precisely. Gulf of Mexico: first LNG cargo 2014.
3. Check the lag before repricing. By 2015 there was "no real movement in the storage values in that market" — the re-rating is not a trade, it's a decade. Anyone who sized the response off year one saw nothing.
4. Measure the terminal move: Gulf storage values today are 300% of what they were — a tripling over 10–12 years.
5. Decompose why, so you can test whether the analogue transfers. Two mechanisms: (a) new demand inelastic to price in both seasons because it's protecting a $10–20 export value chain; (b) capacity cannibalisation — reserving injection capability for an LNG-linked player takes roughly three times the space a conventional utility user consumes, shrinking conventional share while volatility rises.
6. Apply to the target market only if both mechanisms are present, and state it as a hypothesis, not a forecast.

• AECO storage rate prints and Alberta long-term contract awards over the next several years — the Gulf pattern says the first two years show nothing.
• LNG Canada's first Pacific winter and the end of commissioning flaring as the concrete milestones that start the clock.

31:26 6. Find the missing lever — ask what used to cap the spread, and whether it still works

1. For any spread business, name the mechanism that historically capped the spread. Here it was shale's price elasticity: at $2–3 producers turned on, at $1 they shut in, so no spread ever ran.
2. Test whether that mechanism is still operative, on economics and on physics:
   • Economics — what are producers actually solving for? If they're targeting liquids-rich and oil-associated gas, the dry-gas price barely enters the decision, and they'll produce below variable cost on the dry leg for extended periods.
   • Physics — can the wells even be cycled? Expensive horizontals "can't turn on and off at will" without damaging long-term productivity.
3. If both legs are broken, the historical spread cap is gone and every backward-looking valuation of the asset is anchored to a regime that no longer exists.
4. Add the corroborating datapoint from outside the basin: the attack on Iran "exposed how much associated gas is coming from oil" — i.e. how much supply is now a by-product, indifferent to gas price.
5. Sanity-check the reverse case before committing: if the constraint were removed, does the thesis break? Here he argues it doesn't — "too much egress is as good for us as not enough egress," because oversupply in a basin with no home is just volatility on the other side (1:17:38). Treat that symmetry claim sceptically; it is the most self-serving argument in the interview.

• Any return of price-responsive dry-gas shut-ins — that would restore the cap and re-run the Niska outcome.
• Sustained $100 oil: he concedes it means "way too much gas looking for a home," which he frames as bullish but is really a test of the symmetry claim.

26:27 7. The accumulated-conservatism audit — 10 cautious decisions compound into a 20–30% inefficiency

1. Target assets that have changed hands several times and been run cautiously by each owner — here: AEC → EnCana → Carlyle/Riverstone (Niska) → Brookfield.
2. Walk the whole operating chain and, at each link, ask what limit is in place and who set it and why. Distinguish limits that were reasoned from limits that were inherited.
3. Recognise the compounding: each link's conservatism is individually defensible, but "if everyone's been conservative at every move of the entire chain and over the years nothing's been really investigated or challenged, you get 10 conservative decisions that can lead to a 20 or 30% inefficiency."
4. Attack it from two directions: risk-management limits (where is the framework too restrictive — or not restrictive enough?) and physical/commercial restrictions self-imposed on the assets.
5. Import an outside benchmark deliberately — he used Castleton's bottom-up, fundamentals-led risk discipline as the yardstick for what "normal" should look like.
6. The payoff is free capacity: the same reasoning produces the brownfield programme (compression, adjacent reservoirs, new wells, debottlenecking) at a 4–6× build multiple rather than greenfield risk.

• Serially-acquired infrastructure with a long-tenured workforce and no operational challenge in a decade — the classic setup for this audit.
• Underutilised ancillary infrastructure (power interconnects, land, staffed sites, load factor well under 100%) as the cheapest expansion capital there is.

1:14:48 8. Underwrite the announcement discount — separate FID from narrative

1. When a headline demand number lands (a $13bn data-centre announcement, a gigawatt queue), immediately ask how much of it has reached FID. "There's more discussion about the demand than there has actually been FID projects."
2. Build the case on the conservative forecast instead, and check whether it still works: LNG 20 → 30 Bcf; Alberta power doubling in a decade. If the thesis survives at that level, the announcements are optionality, not the base case.
3. Look past the named proponent to the structural requirement the demand class imposes. Data centres want 99.9% redundancy, which means gas peakers backstop the load even when renewables supply it — "the battery backup or the storage backup to back up the real battery which is the power peaker."
4. Prefer the demand that doesn't depend on picking the winner: "it doesn't have to be one specific user or one specific segment."
5. Note the second-order geography: where basins are gas-rich but pipeline-constrained, new demand co-located with supply can bypass the full transmission system with billion-dollar-but-short pipelines — which changes where, not whether, the build happens.

• FID announcements versus MOU/announcement counts in a power queue — the ratio is the real growth rate.
• Peaker-plus-storage backup requirements in data-centre power contracts as the mechanism that converts AI demand into gas-infrastructure demand.