Stock Analysis · Bulls vs Bears

ERock, Inc. EROC

🐂 Bulls vs Bears 🐻Report #211 Oct 2026
★ Willem owns this stock · 3% of the portfolioOlder version — see the latest →
62
high risk▲ +1 vs last
🐂 7vs8 🐻
Bears lead
33%
▲ Chance of the biggest tailwind the forecasts miss
27%
▼ Chance of the biggest headwind the forecasts miss
The verdict

ERock is a small, sold-out assembler of fast-to-deploy gas power, riding a shortage that competitors confirm will last until at least 2028. Demand and supply look more secure than three days ago, but its home state has turned politically hostile to data centres right where its biggest projects sit. So the next 3 months (Q3 results, the TCEQ report, the lock-up and the PUCT vote) decide whether risk falls or rises.

🎙️ Bulls vs Bears podcast — ERock, Inc.
Alex and Sam debate every bull and bear argument · 9:52
0:009:52
⤓ Download

Written by Claude (Anthropic’s AI) from public sources · Willem owns EROC · AI can make mistakes — numbers in brackets point to the sources · Not financial advice

🐂 Bull arguments 7

  1. 1The speed-to-power shortage is real, documented by competitors, and lasts until at least 2028Strong
  2. 2Marquee customers, backed by cashStrong
  3. 3Engine supply secured for 2027–28 (new)Medium
  4. 4One asset, three revenue lives, plus a service tailMedium
  5. 5An asset-light model with a strong cash positionMedium
  6. 6Low emissions and a small footprint ease sitingWeak
  7. 7Texas rules require backup generation disclosure and emergency curtailmentWeak

🐻 Bear arguments 8

  1. 1Everything rides on an H2 2026 delivery cliffHigh
  2. 2Texas politics now hit the core projects (new)Medium
  3. 3Customer and geographic concentrationHigh
  4. 4Bigger competitors are adding capacity into 2027–28Medium
  5. 5Lock-up, short build-up and an insider-tilted structureMedium
  6. 6A thin-margin business priced for growthMedium
  7. 7Leadership track record and control weaknessesMedium
  8. 8Supplier switch and tariff exposureLow

Bull arguments show how well they survive the devil’s advocate (strong / medium / weak); bear arguments show the damage they could do after the counter-argument (high / medium / low).

Bull case — all positive points (most important first)

🐂 Bull #1The speed-to-power shortage is real, documented by competitors, and lasts until at least 2028

  • The point: buyers pay for whoever can deliver firm power soonest. On competitors' own words, ERock's 12–18 months is two to three years ahead of the alternatives.
  • Evidence:
    • Caterpillar (2026-08-04 call): "towards the back half of 2028 and into 2029 on turbines", large gas prime engines "late 2028 for sure". It is taking orders "into 2029 and 2030" and says "no one is slowing down at the moment" [17].
    • Generac (Q2 call): 40–45-week lead times, rivals "70, 80 weeks"; prices rarely go "in reverse" [18].
    • GE Vernova: 116 GW of gas backlog plus slot reservations [23].
    • ERCOT's large-load queue is ~475 GW, about 90% data centres, against a 91 GW all-time peak [50].
    • ERock's capacity is sold out "through 2027 and into 2028" [7].
  • Why it matters: sold-out capacity plus deposits means 2027 revenue depends on ERock's factory output, not on winning orders.
  • 😈 Devil's advocate:
    • Shortages invite supply. Cat's 10 MW line (~1.5 GW by early 2028), Generac's tripled packaging, Cummins' new prime-power platform and INNIO/VoltaGrid's 3.8 GW all land in 2027–28 [17,18,20,25].
    • ERock's better pricing is only expected in "2028 and beyond" [7], just as that supply arrives.
    • Abbott's freeze can reduce the number of Texas sites ready to receive power [12].
  • Verdict: Strong and stronger than three days ago. Competitor transcripts independently confirm the gap through 2028. Beyond that it is unproven.

🐂 Bull #2Marquee customers, backed by cash

  • The point: the biggest AI buyers have chosen ERock and pre-paid.
  • Evidence:
    • Anthropic 470 MW, the largest award in company history [3,39].
    • Meta's subsidiary chose ERock for McCloud [8].
    • Microsoft 63 MW [1].
    • Contract liabilities were $528.4M at 30 June [2].
    • Backlog of ~$1.7B (up ~10x year on year) counts only firm contracts with deposits [3,7].
    • Analysts: 8–9 brokers, all Buy. The only target explicitly built up is Wolfe's: 14x 2028 EBITDA of $400M [26,30].
  • Why it matters: deposits fund inventory, and reference customers sell the next order.
  • 😈 Devil's advocate:
    • Anthropic's order is equipment only [7]. Its site is unknown, so we don't know whether that site is affected by the Texas freeze.
    • Anthropic's own compute commitments are enormous (e.g. a reported $45B, 460 MW deal with Nscale [48]), so a priority reshuffle is possible.
    • Customers can terminate for a fee [1].
  • Verdict: Strong The cash is in hand and few small caps can show these names. Scope is narrow (equipment only) and the site is undisclosed.

🐂 Bull #3Engine supply secured for 2027–28 (new)

  • The point: the scarce part inside ERock's product is now locked in for the backlog period.
  • Evidence:
    • HD Construction Equipment: 390 billion won ($290M), 2027–28, 22-litre gas long blocks for "large-scale data centers operated by major US technology companies" [6].
    • CEO quote: "secures supply through 2028" [5].
  • Why it matters:
    • It removes the single biggest supply risk flagged in dive #1.
    • Interpretation: if long blocks are a modest share of a genset's ~$357K revenue, $290M of blocks supports far more than $1B a year of equipment revenue. That implies management plans well above today's run rate. Block unit prices are not disclosed, so this is directional only.
  • 😈 Devil's advocate:
    • HD has no North American track record in continuous-duty gas blocks [6].
    • A new engine in a certified emissions package may need re-certification and testing. Not found whether CARB-DG covers the HD engine.
    • Korean-made blocks face US tariff exposure (Section 232 and others). ERock says it cannot always pass costs through [1].
  • Verdict: Medium Strong. Supply is contracted, but a new supplier carries ramp and qualification risk that nobody has disclosed.

🐂 Bull #4One asset, three revenue lives, plus a service tail

  • The point: a RockBlock can serve as bridge, then backup, then grid capacity over 20–25 years, and each sale seeds an O&M contract.
  • Evidence:
    • McCloud is designed to run as Meta's bridge for ~5 years, then become El Paso Electric's flexible capacity [8,9].
    • Service attach is 95%+ and the renewal rate is 100% (company claims) [4,7].
    • Installed base is 1,104 MW [3].
  • Why it matters: recurring revenue could grow from a 2+ GW installed base.
  • 😈 Devil's advocate:
    • The McCloud "second life" is exactly what the Texas judges questioned. If El Paso Electric cannot charge ratepayers after the bridge period, the plant's flexible-capacity phase may never happen as designed [8].
    • Annual recurring revenue (ARR) of $23.6M is still tiny, and Q2 services revenue fell [3].
  • Verdict: Medium weaker than in dive #1. The flagship example of the model has just been challenged by a regulator.

🐂 Bull #5An asset-light model with a strong cash position

  • The point: customers fund the plants, deposits fund inventory, and ERock has no debt.
  • Evidence:
    • $626.6M unrestricted cash, no debt, undrawn $250M credit line [2,3].
    • A new 1.2 GW line costs $15–20M [7].
  • Why it matters: growth needs no new equity, unlike rental-fleet rivals such as Solaris [24].
  • 😈 Devil's advocate:
    • Excluding deposits, cash is only ~$98M (secondary calculation).
    • H2 must turn adjusted EBITDA positive.
    • The TRA (Tax Receivable Agreement) diverts 85% of future tax savings to pre-IPO owners [1].
  • Verdict: Medium

🐂 Bull #6Low emissions and a small footprint ease siting

  • The point: the cleanest gas engine available (company claim) makes permits and neighbour acceptance easier.
  • Evidence:
    • CARB-DG certification [1].
    • No SCR needed [1].
    • ~50% smaller footprint and no water use [7].
  • Why it matters: where permits are the bottleneck, being cleanest helps.
  • 😈 Devil's advocate:
    • Texas' freeze applies to all data-centre permits regardless of how clean the engine is [12].
    • The fast-track Texas gas-generator standard permit caps a site at six engines and 400 hours a year per engine (draft text). So ERock's large sites need full New Source Review permits like everyone else's [44].
    • Critics now attack "minor-source" permitting of clustered generators as a loophole [45].
  • Verdict: Weak Medium in Texas, Medium in California.

🐂 Bull #7Texas rules require backup generation disclosure and emergency curtailment

  • The point: SB 6 makes large loads disclose on-site backup and lets ERCOT order them to switch to it or curtail. That builds a structural role for dispatchable on-site gas [51,16].
  • Evidence:
    • SB 6 text [51].
    • Large-load rule 25.194 reportedly took effect 2026-10-08 with $50,000/MW security (secondary, unverified).
    • Batch Zero, ERCOT's first batch study for big new loads, is paused on 17 loads (6.6 GW). Final results are still officially 2027-04-09 [15].
  • Why it matters: if the grid gets slower and interruptible, on-site power becomes standard.
  • 😈 Devil's advocate:
    • The same politics produced a permit freeze and an audit. Abbott plans to ask the January 2027 legislature to "eliminate any financial incentives for data centers" [13,16].
    • Hostility toward data centres reduces the number of ERock's customers in Texas.
  • Verdict: Weak (was Medium). The direction of Texas policy has turned against data centres.

Bear case — all negative points (most important first)

🐻 Bear #1Everything rides on an H2 2026 delivery cliff

  • The point: guidance needs revenue to jump ~5x half on half, from a new plant.
  • Evidence:
    • H1 revenue was $71.6M. Full-year guidance is $435–465M, so H2 needs ~$363–393M [2,3].
    • H2 adjusted EBITDA must reach roughly +$30–36M, against −$26.9M in H1 [3].
    • Management's own 2026 backlog conversion is $360–390M, which only just covers guidance [7].
    • Hyperion started assembly in Q2 [3].
    • Three material weaknesses remain unremediated: segregation of duties, accounting staff and IT controls [2].
    • The one public Q3 consensus figure (~$20M revenue, TipRanks) looks stale. If it were right, Q4 would need $345M+ [30].
  • Why it matters: missing the first-ever guidance would damage the new management team's credibility.
  • 🛡️ Counter-argument:
    • Revenue is booked at factory acceptance, and bill-and-hold allows booking even if a customer's site is late [2].
    • Inventory more than doubled to $106M [2].
    • Q2 beat consensus [30].
  • Severity: High Q3 results (~10–17 Nov) are the next test.

🐻 Bear #2Texas politics now hit the core projects (new)

  • The point: ERock's two biggest risks in the state are no longer hypothetical.
  • Evidence:
    • McCloud (Meta/El Paso Electric):
      • The PUCT judges' proposal for decision (~2026-09-23) found that El Paso Electric did not consider alternatives, ran no request for proposals and did no cost-benefit study. It recommends approval only if El Paso Electric "does not pass capital and operating costs on to its ratepayers", and denial otherwise [8].
      • The City of El Paso opposes the application [10].
      • A PUCT vote is reportedly set for December [11, summary].
      • Construction has already begun [3].
      • El Paso Electric already holds its TCEQ air permit [8].
    • Statewide freeze:
      • Abbott's 2026-09-21 order tells TCEQ to "pause the issuance of all permits related to data center projects until ERCOT completes its review", expected mid-December. It covers islanded, behind-the-meter sites too, according to law firms. Public Citizen disputes that reading [12,13,14].
      • TCEQ reports on compliance by 2026-10-19.
      • Whether fast-track standard permits and permits-by-rule are caught is unclear [12].
    • Disclosure: ERock's 10-Q (2026-08-14) reported "no material changes" in risk factors [2]. The company has not commented on either event (nothing found).
  • Why it matters:
    • McCloud is ~$290M of backlog and likely ERock's top customer.
    • Most of ERock's other data-centre backlog is in Texas, and new Texas sites can't get air permits until the freeze lifts.
  • 🛡️ Counter-argument:
    • Meta already pays all McCloud costs for the bridge period. The dispute is mainly about who pays afterwards [8,11], so approval with a no-ratepayer condition is a plausible outcome that keeps the generators ordered.
    • Already-issued permits appear outside the freeze [12].
    • The freeze is tied to a December audit, not open-ended.
    • Delays to grid hook-ups can increase demand for bridge power.
  • Severity: Medium High. A denial, or a condition that makes El Paso Electric walk away, would be a visible hit to ERock's largest customer. A long freeze would slow 2027–28 orders.

🐻 Bear #3Customer and geographic concentration

  • The point: a handful of customers and one state decide the outcome.
  • Evidence:
    • Top customer 37% (Q1) and 28% (Q2) of revenue [2].
    • One customer holds 62% of receivables [2].
    • Texas is 80%+ of revenue [1].
    • Data centres provided "substantially all" backlog growth [2].
    • Both plants are in Houston [1].
  • Why it matters: Bear #2 shows how a single regulatory decision on one customer can move the whole company.
  • 🛡️ Counter-argument:
    • Deposits and termination payments protect ERock's cash.
    • Customers are creditworthy.
    • The Anthropic order dilutes concentration from 2027.
  • Severity: High

🐻 Bear #4Bigger competitors are adding capacity into 2027–28

  • The point: the gap ERock exploits is the one competitors are now filling.
  • Evidence:
    • Caterpillar's 10 MW gas line ramps to ~1.5 GW over 18 months, and Cat has a $72B backlog [17].
    • Generac: a ~$1.6B data-centre backlog and two hyperscaler deals of ~$700M+ each, tripling packaging capacity [18,19].
    • Cummins: a "several GW" hyperscaler deal and a gas prime-power platform in development [25].
    • INNIO/VoltaGrid: 1.5 GW (Feb 2026) on top of 2.3 GW, delivery by 2028 [20].
    • Wärtsilä: 1.2 GW of data-centre orders [21].
    • Bloom: guiding $3.9–4.2B revenue [22].
  • Why it matters: ERock's ~19–22% gross margin leaves no room for a price war. Its pricing upside is pencilled in for 2028+, when this supply arrives.
  • 🛡️ Counter-argument:
    • Everyone is sold out. Generac says prices rarely go "in reverse" [18].
    • Cat's 10 MW unit targets big prime-power sites, not ERock's small modular blocks [17].
  • Severity: Medium High.

🐻 Bear #5Lock-up, short build-up and an insider-tilted structure

  • The point: a small free float meets a large unlock.
  • Evidence:
    • Lock-up ends 2026-12-06 (a Sunday, so effectively 12-07). Morgan Stanley can release holders early [1].
    • Class A shares outstanding: 48.2M. Class B units: 171.2M. Reserved for Class M units: 52.4M [2].
    • EIP's 13D reports 102.8M shares on an as-exchanged basis [35].
    • The founder holds ~22.6M [36].
    • No S-3 or Form 144 filed as of 2026-10-11 [34].
    • No open-market insider purchases since the IPO. Insider sales were IPO-related, at $19.85 [34,36].
    • Short interest rose from 6.25M (08-14) to 12.0M (09-30) with 4.3 days to cover, ~25% of Class A [32,33].
    • A Hagens Berman "investigation" solicitation (not a lawsuit) is online [46].
  • Why it matters: supply around December plus heavy shorting can depress the price whatever the business does. The TRA permanently diverts value to pre-IPO owners.
  • 🛡️ Counter-argument:
    • At ~56% of the IPO price, sponsors have little reason to dump stock.
    • A high short interest can also squeeze on good Q3 news.
    • The lock-up date is public.
  • Severity: Medium

🐻 Bear #6A thin-margin business priced for growth

  • The point: the stock already assumes a tripling of revenue.
  • Evidence:
    • Gross margin 18.6% (adjusted 22.2%) in Q2 [3].
    • Consensus revenue: ~$450M (2026), ~$1.42B (2027) and ~$2.2B (2028, unverified) [30,31].
    • Enterprise value (EV) ~$2.7B on ~273M units at $11.95 [31,37].
    • Wolfe's bull model needs mid-20s% gross margin by 2028 [26].
  • Why it matters: both the volume ramp and margin expansion must happen.
  • 🛡️ Counter-argument:
    • Fixed costs are low, so operating leverage at 3x volume can be large.
    • Newer orders are reportedly priced higher [7].
  • Severity: Medium

🐻 Bear #7Leadership track record and control weaknesses

  • The point: a new CEO with a poor last public-company outcome, an unproven team, and controls not yet fixed.
  • Evidence:
    • Three CEOs in ~12 months before the IPO [1].
    • Stem under Carrington: FY2024 net loss of $854M and a 1-for-20 reverse split in 2025 [40,41].
    • Material weaknesses still open [2].
    • The founder's Form 4 was filed ~3 months late [36].
  • Why it matters: a 5x ramp is when weak controls produce errors.
  • 🛡️ Counter-argument:
    • Amthor and Blakely have run the business for a decade.
    • Stem's 2024 impairments were partly sector-wide.
  • Severity: Medium

🐻 Bear #8Supplier switch and tariff exposure

  • The point: 2027–28 production depends on a new Korean engine supplier.
  • Evidence:
    • HD's deal is its first large continuous-duty contract in North America [6].
    • The prospectus flags sole-source parts, past supplier delays and tariffs it may not pass on [1].
    • The earlier Generac alliance (2021) would have expired around September 2026 [43].
  • Why it matters: an engine-qualification hiccup in early 2027 would hit the year the backlog is meant to convert.
  • 🛡️ Counter-argument:
    • HD Hyundai is a large, established engine maker.
    • A contracted supply is better than the uncontracted position seen in dive #1.
  • Severity: Low Medium.

Tailwinds NOT in the forecasts

  • A further hyperscale order of 300 MW+, or Anthropic adding installation and O&M — 33% chance within 36 months · Impact if it happens: High

    • What it is: a new large order, or Anthropic's equipment-only purchase order widened to full scope.
    • Why it's not priced in: BofA's model reportedly counts Anthropic product sales only [38]. Consensus 2027 assumes conversion of existing backlog.
    • What would trigger it / the sign to watch: backlog above ~$2.2B at Q3 or Q4, or a third assembly line (the 65%-contracted trigger) [7].
    • Why this probability: the pipeline reportedly tripled and HD's ~$290M block order implies planned volume above backlog. But the Texas freeze may delay site-specific orders, so slightly lower than dive #1's 35%.

  • Utilities copy the "bridge-then-flex" model with a clean ratepayer structure — 21% chance · Impact: Medium

    • What it is: a utility orders RockBlocks with the data-centre customer paying throughout, avoiding McCloud's ratepayer fight.
    • Why it's not priced in: utility backlog shrank in H1 [2].
    • Sign to watch: a utility filing naming ERock, or the PUCT's December McCloud order setting a usable template.
    • Why this probability: the judges' ruling makes utilities more cautious, but a clear approval-with-conditions could create a template.

  • The permit freeze creates a pre-permitted-sites premium for ERock's existing backlog — 18% chance · Impact: Low–Medium

    • What it is: customers with permits in hand accelerate, and ERock's bill-and-hold deliveries become more valuable.
    • Why it's not priced in: consensus treats the freeze as neutral or negative.
    • Sign to watch: Q3 call commentary on pull-forward requests.
    • Why this probability: speculative, with no evidence yet.

  • Margin surprise before 2028 — 17% chance · Impact: Medium

    • Why it's not priced in: the CFO pointed to 2028+ [7].
    • Sign to watch: adjusted gross margin above ~25% in any 2027 quarter.
    • Why this probability: older contracts dominate 2027 deliveries, and new HD engines may carry tariff costs.

Headwinds NOT in the forecasts

  • The Texas permit freeze or January 2027 legislation delays new Texas data-centre sites in ERock's 2027–28 order book — 27% chance of a visible delay within 18 months · Impact if it happens: High

    • What it is: TCEQ's pause extends beyond the December ERCOT audit, or the legislature curbs data-centre incentives and permits. Customers' sites then slip, and deliveries either shift to bill-and-hold or are deferred.
    • Why it's not priced in: guidance (2026-08-11) predates the freeze. No analyst action on the freeze was found.
    • Sign to watch: TCEQ's 2026-10-19 report, ERCOT's December audit, the Q3 call, and bills filed for the session opening January 2027.
    • Why this probability: the freeze is real and broad, but it is tied to a dated audit, already-issued permits appear safe, and ERock's 2026 deliveries are mostly for sites already underway.

  • The PUCT denies McCloud, or El Paso Electric rejects the conditions, and the project is delayed or restructured — 14% chance · Impact: High

    • What it is: commissioners follow the judges' "deny" branch, or El Paso Electric won't accept a no-pass-through condition, and Meta has to find another owner or structure.
    • Why it's not priced in: ERock's filings and guidance treat McCloud as under construction [3].
    • Sign to watch: the PUCT open-meeting agenda (December), El Paso Electric or Meta exceptions to the proposal for decision, and any ERock 8-K.
    • Why this probability: Meta already pays during the bridge period, so a conditioned approval is the likeliest outcome. Outright denial of a plant already under construction is rare, but the judges' language is unusually sharp.

  • The H2 2026 ramp misses guidance — 34% chance · Impact: Medium

    • Why it's not priced in: consensus ~$450M sits at the guidance midpoint [31].
    • Sign to watch: Q3 revenue. Roughly $140M+ would keep the year plausible (Interpretation: assumes about a 40/60 split between Q3 and Q4).
    • Why this probability: a 5x step-up from a new plant. The HD deal doesn't help 2026.

  • A large customer delays, downsizes or cancels — 19% chance within 36 months · Impact: High

    • Sign to watch: backlog falling without matching revenue, or termination-fee income.
    • Why this probability: deposits deter cancellation, but the McCloud review and Texas freeze raise site risk.

  • Heavy selling around the lock-up, amplified by a crowded short — 42% chance of a noticeable price impact around December 2026 · Impact: Low–Medium

    • Sign to watch: an S-3, Form 144s, or a marketed secondary.
    • Why this probability: short interest doubled into the event, and 20M+ Class A shares plus exchangeable units unlock.

Investment Risk Score: 62/100

Risk area Weight Score (0–100) Why
Business & competition 25% 60 Competitor calls confirm a sold-out market to 2028, slightly better than feared (62 before). But Cat, Generac, Cummins and INNIO capacity lands in 2027–28, and ~20% gross margin leaves no buffer.
Financial health & dilution 20% 45 Unchanged. $627M cash and no debt, but ~$98M excluding deposits. Loss-making until H2. TRA.
Execution (delivering the plan) 20% 72 Unchanged. A 5x H2 ramp from a new plant with open control weaknesses, plus a 2027 supplier switch to HD.
Dependency & concentration (customers, suppliers, commodities, key people) 15% 75 Engine supply is now contracted (better), but the top customer is very likely McCloud, now under regulatory review (worse). Net unchanged.
Regulation, policy & geopolitics 10% 70 Up from 55. Texas permit freeze (BTM included), the McCloud proposal for decision, Batch Zero paused, an anti-incentive legislative push, and tariffs on Korean engines.
Sentiment & expectations (how much good news the market already expects) 10% 58 Up from 55. All-Buy coverage and 3x 2027 consensus, against short interest that doubled to ~25% of Class A into a 12-07 lock-up.

Overall = the weighted average = 15.0 + 9.0 + 14.4 + 11.25 + 7.0 + 5.8 = 62.45, which rounds to 62 (high). Scale: 0–25 low · 26–50 moderate · 51–75 high · 76–100 very high.

The score rises by one point. Demand and engine-supply evidence improved, but Texas regulation became a concrete risk to ERock's largest project and its home market. Execution and concentration still drive the score, and the Q3 report is the next point where it can move meaningfully either way.

Conclusion

  • What kind of company this is: a newly listed, still loss-making assembler-integrator selling "time to power" during a shortage. Owning it is a bet that ERock converts ~$1.7B of deposit-backed orders into delivered gensets in 2026–28, before larger engine makers close the gap, and that Texas politics does not strand its customers' sites.
  • The 2–3 things that will decide the outcome:
    1. Delivery execution: Q3/Q4 2026 against guidance, then 2027 on HD engines.
    2. Texas regulation: the PUCT's McCloud decision and how long the TCEQ freeze lasts.
    3. New orders beyond Anthropic and Meta at better prices, before 2028 competition arrives.
  • Signposts:
    • Bull confirmed if:
      • Q3 revenue is ~$140M+ and guidance is reaffirmed.
      • The PUCT approves McCloud (even with a no-ratepayer condition) and El Paso Electric and Meta accept.
      • Backlog rises above ~$2B, or a third production line is announced.
      • TCEQ lifts the pause after the December audit.
    • Bull broken if:
      • Guidance is cut, or revenue is pushed into 2027.
      • McCloud is denied, or El Paso Electric withdraws.
      • The freeze extends into 2027 or the legislature restricts on-site gas for data centres.
      • Backlog or contract liabilities fall without matching revenue, or a new material weakness or restatement appears.
  • Balanced verdict:
    • The strongest bull case is now better documented than three days ago. Caterpillar's and Generac's own calls confirm that nobody else can deliver firm gas power before 2028. ERock is sold out into 2028, holds $528M of customer cash, and has locked in its key input through 2028.
    • The strongest bear case also got sharper. ERock's biggest known project is under a hostile regulatory proposal, and its home state has frozen data-centre permits, while the company must quintuple output in one half-year under unremediated controls, ahead of a large unlock and a doubling of short interest.
    • The evidence leans positive on demand and supply, and negative on execution and Texas policy. Until Q3 shows delivered revenue and the PUCT and TCEQ outcomes are known, risk stays high.

Open questions & data gaps

  • Exposure of ERock's backlog to the TCEQ freeze: which customer sites (especially Anthropic's 470 MW) already hold air permits. Not found. High importance.
  • McCloud contract protection: whether ERock's generator sale to El Paso Electric is protected (paid deposits, termination payments) if the PUCT denies. Not found. The PUCT docket (possibly 59076, unconfirmed) and the exact vote date were not read directly. High importance.
  • Top customer identity: still undisclosed. El Paso Electric is inferred. High importance.
  • HD engine qualification: whether the 22-litre HD block is covered by ERock's CARB-DG certification and tariff treatment. Not found. Medium importance.
  • Q2 call transcript: still only AI summaries (Investing.com, Quartr). Verbatim Q&A is not verified. Medium importance.
  • Analyst consensus for 2027–28 EBITDA and EPS, and Q3 consensus: aggregator data conflict (2027 revenue $0.95B vs $1.42B; Q3 $20M looks stale). Medium importance.
  • Reasons for JPMorgan's (to $21, aggregator-only) and Guggenheim's ($23→$20) target cuts: not found. Low importance.
  • Short-interest float base: MarketBeat's 25.4% uses ~47M shares. True tradeable float may be ~28M. Low–Medium importance.
Sources
  1. ERock, Inc. Form 424B4 prospectus — SEC EDGAR — 2026-06-10 — https://www.sec.gov/Archives/edgar/data/2110029/000119312526265898/d12401d424b4.htm
  2. ERock, Inc. Form 10-Q, quarter ended 2026-06-30 — SEC EDGAR — 2026-08-14 — https://www.sec.gov/Archives/edgar/data/2110029/000119312526346922/eroc-20260630.htm
  3. ERock Reports Second Quarter 2026 Results (8-K Ex. 99.1) — SEC EDGAR — 2026-08-11 — https://www.sec.gov/Archives/edgar/data/2110029/000119312526344714/eroc-ex99_1.htm
  4. Q2 2026 Earnings Presentation — ERock IR — 2026-08-11 — https://d1io3yog0oux5.cloudfront.net/_87038f422296da5577b2b93c40dd3b1a/erock/db/3576/34525/pdf/Aug+11+-+EROC+Earnings+Presentation_Design_V2.pdf
  5. ERock, HD Construction Equipment Sign Multi-Year Engine Long Block Supply Agreement — ERock newsroom — 2026-10-07 — https://erock.com/newsroom/erock-hd-construction-equipment-sign-multi-year-engine-long-block-supply-agreement-to-accelerate-manufacturing-capacity/
  6. HD Construction Equipment signs supply deal with ERock (~390 billion won) — Korea Herald (Herald Business) — 2026-10-07 — https://biz.heraldcorp.com/article/10895760
  7. Earnings call transcript: ERock jumps 23% after Q2 2026 revenue surge (AI-assisted summary, not verbatim) — Investing.com — 2026-08-12 — https://www.investing.com/news/transcripts/earnings-call-transcript-erock-jumps-23-after-q2-2026-revenue-surge-93CH-4855247
  8. Texas Electric Utility Only Considered Gas to Power $10 Billion Meta Data Center — Inside Climate News — 2026-09-24 — https://insideclimatenews.org/news/24092026/el-paso-electric-only-considered-gas-to-power-meta-data-center/
  9. El Paso Electric plans $473M gas plant — Transformer Magazine (citing KVIA) — 2026-01-20 — https://transformer-magazine.com/news/el-paso-electric-plans-473-m-gas-plant/
  10. City Opposes Proposed McCloud Generation CCN — City of El Paso — 2026-06-18 — https://www.elpasotexas.gov/assets/Press-Releases/2026.06.18-NEWS-RELEASE_City-Opposes-Proposed-McCloud-Generation-CCN-Cites-Deficient-Analysis-and-Risks-to-Ratepayers.pdf
  11. El Paso ratepayer protection / McCloud PUC vote scheduled for December (search summary only) — KTSM 9 News — 2026 (date not shown) — https://www.ktsm.com/news/el-paso-ratepayer-protection/amp/
  12. Governor Abbott Orders TCEQ to Pause All Permit Issuance for Data Center Projects — Bracewell — 2026-09-23 — https://www.bracewell.com/resources/governor-abbott-orders-tceq-to-pause-all-permit-issuance-for-data-center-projects/
  13. Governor Abbott Directs TCEQ to Halt All Data Center Permits Pending ERCOT/TWDB Audits — Troutman Pepper Locke — 2026-09-22 — https://www.troutman.com/insights/governor-abbott-directs-tceq-to-halt-all-data-center-permits-pending-ercot-twdb-audits/
  14. Another Abbott data center letter, but no real action (search summary only) — Public Citizen — 2026-09 — https://www.citizen.org/news/another-abbott-data-center-letter-but-no-real-action/
  15. Batch Zero Verification and Audit, PUCT Open Meeting presentation — ERCOT — 2026-08-20 — https://www.ercot.com/files/docs/2026/08/20/Batch-Zero-Verification-and-Audit-PUCT-Presentation.pdf
  16. Abbott orders full audit of Texas data center interconnection queue — POWER Magazine — 2026-08-04 — https://powermag.com/abbott-orders-full-audit-of-texas-data-center-interconnection-queue-threatens-to-deny-grid-access/
  17. Caterpillar 2Q 2026 Earnings Call Transcript — Caterpillar IR — 2026-08-04 — https://s25.q4cdn.com/358376879/files/doc_financials/2026/q2/2Q-2026-Earnings-Call-Transcript.pdf
  18. Generac (GNRC) Q2 2026 Earnings Call Transcript — Motley Fool — 2026-08-07 — https://fool.com/earnings/call-transcripts/2026/08/07/generac-gnrc-q2-2026-earnings-call-transcript
  19. Generac Q2 2026 results (8-K Ex. 99.1, summary only) — SEC EDGAR — 2026-07-29 — https://www.sec.gov/Archives/edgar/data/0001474735/000143774926024731/ex_994826.htm
  20. INNIO Secures Additional Major Order from VoltaGrid: 1.5 GW — INNIO — 2026-02 — https://www.innio.com/en/news-media/press-releases/innio-secures-additional-major-order-from-voltagrid-1-5-gw-for-behind-the-meter-power-generation/
  21. Wärtsilä Half-year Financial Report January–June 2026 (summary only) — Wärtsilä — 2026-07 — https://www.wartsila.com/fwd/result-presentations/interim-report-q2-2026
  22. Bloom Energy Q2 2026 results (8-K Ex. 99.1) — SEC EDGAR — 2026-07-28 — https://www.sec.gov/Archives/edgar/data/0001664703/000162828026050150/ex991_q226financialresults.htm
  23. GE Vernova reports Q2 2026 results (summary only) — GE Vernova — 2026-07-22 — https://gevernova.com/news/press-releases/ge-vernova-reports-second-quarter-2026-financial-results-raises-2026-financial
  24. Solaris Energy Infrastructure Q2 2026 results (8-K Ex. 99.1) — SEC EDGAR — 2026-08-05 — https://www.sec.gov/Archives/edgar/data/0001697500/000162828026053383/sei-ex99x6302026.htm
  25. Cummins Q2 2026 results (8-K Ex. 99, summary only) — SEC EDGAR — 2026-08 — https://www.sec.gov/Archives/edgar/data/0000026172/000002617226000026/cmi2026q28-kex99.htm
  26. Wolfe Research initiates ERock with Outperform — Investing.com — 2026-07-06 — https://www.investing.com/news/analyst-ratings/wolfe-research-initiates-erock-stock-with-outperform-rating-93CH-4775924
  27. BofA initiates ERock coverage with Neutral at $16 — Investing.com — 2026-07-06 — https://www.investing.com/news/analyst-ratings/bofa-initiates-erock-stock-coverage-with-neutral-rating-at-16-93CH-4776108
  28. Barclays initiates ERock with Overweight on power scarcity — Investing.com — 2026-07-06 — https://www.investing.com/news/analyst-ratings/barclays-initiates-erock-stock-with-overweight-on-power-scarcity-93CH-4776181
  29. Morgan Stanley initiates ERock with Overweight — Investing.com — 2026-07-06 — https://www.investing.com/news/analyst-ratings/morgan-stanley-initiates-erock-stock-with-overweight-rating-93CH-4775738
  30. ERock consensus estimates and rating actions — Investing.com / TipRanks — retrieved 2026-10-11 — https://www.investing.com/equities/erock-consensus-estimates ; https://www.tipranks.com/stocks/eroc/forecast
  31. ERock forecast and statistics — StockAnalysis — retrieved 2026-10-10/11 — https://stockanalysis.com/stocks/eroc/forecast/ ; https://stockanalysis.com/stocks/eroc/statistics/
  32. EROC short interest history — MarketBeat — retrieved 2026-10-11 — https://www.marketbeat.com/stocks/NYSE/EROC/short-interest/
  33. EROC short interest and short volume — Equibles — retrieved 2026-10-11 — https://equibles.com/stocks/eroc
  34. EDGAR filing index and submissions for CIK 2110029 — SEC — retrieved 2026-10-11 — https://www.sec.gov/cgi-bin/browse-edgar?action=getcompany&CIK=0002110029 ; https://data.sec.gov/submissions/CIK0002110029.json
  35. Schedule 13D, Energy Impact Partners group — SEC EDGAR — 2026-06-18 — https://www.sec.gov/Archives/edgar/data/0002110029/000119312526275971/xslSCHEDULE_13D_X02/primary_doc.xml
  36. Form 4 / Form 3 and Schedule 13G/A, W. T. McAndrew Jr. — SEC EDGAR — 2026-09-04 — https://www.sec.gov/Archives/edgar/data/0002110029/000149315226041632/ownership.xml
  37. EROC daily price history — Yahoo Finance — 2026-06-10 to 2026-10-09 — https://query1.finance.yahoo.com/v8/finance/chart/EROC
  38. This unique power company is a play on the Anthropic IPO, BofA says (read via summaries; page blocked) — CNBC — 2026-08-13 — https://www.cnbc.com/2026/08/13/this-unique-power-company-is-a-play-on-the-anthropic-ipo-bank-of-america-says-it-has-room-to-run.html
  39. ERock scores Anthropic deal, sees order backlog soar to $1.7B — Energy Capital HTX — 2026-08-19 — https://energycapitalhtx.com/erock-anthropic-deal
  40. Stem, Inc. Form 10-K for FY2025 — SEC EDGAR — 2026 — https://www.sec.gov/Archives/edgar/data/1758766/000175876626000016/stem-20251231.htm
  41. Stem Announces CEO Transition (8-K exhibit) — SEC EDGAR — 2024-09-16 — https://www.sec.gov/Archives/edgar/data/1758766/000175876624000166/exhibit99-pr11.htm
  42. Power Solutions International celebrates shipment of 1,000th unit to Enchanted Rock (search snippet only) — PSI — undated — https://investors.psiengines.com/node/11536
  43. Generac Power Systems and Enchanted Rock announce alliance — PR Newswire — 2021-09-14 — https://prnewswire.com/news-releases/generac-power-systems-and-enchanted-rock-announce-alliance-301376038.html
  44. Air Quality Standard Permit for Natural Gas Electric Generating Units (draft text) — TCEQ — 2024/2025 — https://www.tceq.texas.gov/downloads/permitting/air/announcements/2024-005-oth-nr-ngegusp24-std-pmt-final-prop5.pdf
  45. This permit common for dry cleaners is now being used to build AI power plants — Floodlight via Stacker/KEYT — 2026-07-17 — https://keyt.com/stacker-ai/2026/07/17/this-permit-common-for-dry-cleaners-is-now-being-used-to-build-ai-power-plants/
  46. ERock (EROC) investigation (law-firm solicitation; search summary only) — Hagens Berman — 2026 — https://www.hbsslaw.com/cases/erock-inc-eroc-investigation
  47. Anthropic considers Texas data center expansion — Axios Austin — 2026-08-26 — https://www.axios.com/local/austin/2026/08/26/anthropic-texas-data-centers-bastrop-rockdale
  48. Anthropic signs $45B compute deal with Nscale for 460 MW in West Virginia — VKTR — 2026-08 — https://www.vktr.com/ai-market/anthropic-signs-45-billion-compute-deal-with-nscale-for-460-megawatts-in-west-virginia/
  49. Enchanted Rock delivers 148 MW of gas-fired power for California water districts — Microgrid Knowledge — 2024-06 — https://www.microgridknowledge.com/microgrids/critical-services/article/55090548/enchanted-rock-delivers-148-mw-of-gas-fired-power-for-california-water-districts
  50. ERCOT Update to Senate Business & Commerce — ERCOT — 2026-07-29 — https://www.ercot.com/files/docs/2026/07/29/ERCOT-Senate-July-29-Panel-1-Assessing-The-Grid.pdf
  51. Texas SB 6 (enrolled text, 89th Legislature) — Texas Legislature Online — 2025 — https://capitol.texas.gov/tlodocs/89R/billtext/html/SB00006F.htm
All EROC Bulls vs Bears reports
Disclaimer. AI-generated research and podcast (Claude by Anthropic, voices by Google Gemini) for educational purposes — not financial advice and not a recommendation to buy or sell. Willem (MR.INVEST) owns this stock. Check the company’s own filings before making any decision.
← All Bulls vs Bears reports