Actionable insights — Portfolio Update: Double Follow-Up

1. Grade a big customer contract by its incentive structure, not its headline size

1. When a supplier announces a large customer deal, read the 8-K for what the customer gets, not just the dollar ceiling.
2. Is the customer given equity or warrants that vest in proportion to its purchases? If so, it profits from the supplier's stock and has a reason to keep ordering.
3. Check whether any of it vested at signing. That separates a committed relationship from a letter of intent or a "preferred vendor" status.
4. Split the ceiling into near-term committed deliveries and the long tail, and model only the committed part.

• Later filings showing further vesting, which means orders are actually flowing, and the Oct 27 Q3 call's color on the Amazon ramp.

2. Anchor today's multiple to what the market was paying for at the old peak

1. Pull the 10-year forward P/E and P/S chart. Note the peak and what drove earnings at the time.
2. If the peak priced temporary (cyclical or pandemic) demand as if it were permanent, treat it as a mistake, not a benchmark.
3. Compare today's multiple with the long-run range excluding that episode, and ask whether today's earnings are more durable (contracted, recurring) than the earnings at the peak.
4. A multiple near the historical norm on more durable earnings is under-priced even if it looks "fair."

• How much of revenue is contracted or data-center (C&I share of sales). If the mix keeps shifting toward it, the case for re-rating strengthens.

3. Mark analyst targets as stale after a material event

1. Note the date of the latest consensus targets against the date of the material filing (contract, guidance change).
2. If the targets predate the event, treat the average as a floor, and the highest pre-event targets as the likely new center.
3. Expect revisions over the following weeks, with the next earnings call as the formal catalyst.

• Target raises in the weeks after the 8-K, and whether the $450M data-center guide is kept or raised on Oct 27.

4. Buy the stock's own pull-back rhythm when the breakout is incomplete

1. Where a stock has broken out but not cleared the long-term range, look for a series of higher highs.
2. Measure the pull-back after each high. If past pull-backs were later rewarded, treat the next one as the add point.
3. Build or add in the pull-back rather than chasing the high.

• A higher low holding on the next pull-back. A lower low would break the pattern.

5. Own the input to an expensive theme instead of the theme's leaders

1. For a hot theme (AI), list its physical inputs (power, then fuel for that power).
2. Find the largest low-cost producer of that input trading at a discount to the market.
3. Quantify incoming demand in physical units (GW of plants under construction or planned) and compare local prices with world prices to size the upside.
4. Look for committed third-party capital (e.g., foreign-funded plants) near the producer's footprint as the catalyst.

• Henry Hub moving off ~$3, final investment decisions on the Japanese-funded plants, and gas-turbine delivery slots (turbines are "in extremely short supply").