← Analysis page  ·  Avi Salzman hub  ·  Research hub

Actionable insights — A New Stock to Play the AI Power Boom (ERock)

The repeatable analysis behind sizing up a hot IPO: not "buy ERock," but how to value a story stock against the demand that's pulling it.
2026-JUN-10 · Barron's · Avi Salzman (featuring ERock CEO John Carrington) · Read ↗ · full analysis · transcript
How to read this page: each insight is a method — the screen, the diagnostic, the milestone to watch — distilled from the article so it can be rerun on the next AI-power IPO. The boxed line shows how it played out here.

1. Price a story-stock IPO on sales against an established comp

The repeatable method
  1. For a money-losing newly-listed company, P/E is meaningless — anchor on price-to-sales instead, using the IPO price × share count (or the stated valuation) over latest annual revenue.
  2. Find the closest profitable, public comp that makes similar gear and compute its P/S on the same basis.
  3. The ratio of the two multiples tells you how much growth the market is pre-paying. A gap of 5×+ means the stock is priced for a steep, specific ramp — not a value entry.
Here: EROC at ~$5.9B / $183M 2025 sales ≈ 32× sales, vs CAT at ~6× — so the market is paying ~5× the peer multiple on the promise of growth. The 5% debut dip barely dents that.
Watch for

2. Buy the seller of the workaround to a hard bottleneck

The repeatable method
  1. Identify a binding physical constraint in a boom (here: multi-year grid-interconnection queues for data centers in Texas/Pennsylvania).
  2. Find the company selling the bridge around it — on-site/behind-the-meter generation that delivers "speed-to-power" while customers wait for the grid.
  3. The longer the queue, the deeper the demand: the constraint itself is the moat. Note any product edge that widens adoption (quieter, lower-emission → easier community/permitting acceptance).
Here: EROC's "rockblock" gas gensets are bought as backup/temporary power until grid hookup; the yearslong wait is exactly why adoption is rising. Competitors doing the same: CAT, GNRC, CMI.
Watch for

3. Validate the demand with customers + backlog, not the pitch

The repeatable method
  1. Check who is buying — marquee, credit-worthy customers are third-party proof the product works at scale.
  2. Weigh the backlog (contracted future revenue) against current revenue to gauge the forward ramp, and against current losses to gauge the gap that must close.
  3. Confirm the company can physically deliver — capacity additions must be able to convert the backlog.
Here: customers META, MSFT, HNHPF (Foxconn); $1.3B backlog vs $183M 2025 revenue; capacity ramping to 1.2 GW/yr by year-end (> the ~1 GW installed base). The proof points are real — the question is conversion.
Watch for

4. For a "priced-for-ramp" name, name the milestone that breaks it

The repeatable method
  1. When a stock trades at a rich multiple on losses, write down the specific operational milestone the price assumes (capacity online, backlog billed, path to profitability).
  2. Decide in advance what evidence would confirm or kill the thesis — a missed capacity date or a stalled backlog is the tell, not the share price.
  3. Size the position to the binary nature of the bet; a 32× name re-rates hard if the ramp slips.
Here: the bull case rests on "growth will ramp soon" — 1.2 GW/yr capacity by year-end and conversion of the $1.3B backlog. If those slip, ~32× sales on a $59M loss has a long way to fall.
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.