American Power Fund IAmerican Power Fund I
American Power Fund I aggregates orphaned 10 to 15 MW power sites in ERCOT West into a 100+ MW portfolio and sells it to a hyperscaler. We acquire each site at $400 to $600K per MW, self-mine the capacity through the hold using fund-owned ASICs financed via equipment debt, and exit in years 3 to 5 at institutional pricing. Stranded Permian gas makes ERCOT West the cheapest large-scale power in North America. 10+ GW of AI data center capacity has already been permitted, contracted, or announced in the region, and fiber is being built today to service those projects. Every quarter that passes, our power infrastructure becomes more valuable to the next buyer. The fund pays an 8 percent annual preferred return to LPs and targets a 2.7 times net MOIC base case over a 5-year hold.
Fund Size
$50M
Hard cap $60M
Portfolio Target
100 MW
60 MW fund + 40 MW co-invest
Exit Comp
$1.5–3.0M/MW
Portfolio → hyperscaler
Target MOIC
2.7–4.5×
Base → upside case
The Pitch In One Paragraph
We acquire orphaned 10 to 15 MW power sites in ERCOT West at $400 to $600K per MW. This is a price range that public miners pass on (their deals need to be 50 MW or larger to move production guidance) and that institutional infrastructure capital cannot underwrite (their minimum check sizes start above $25M). Once the fund acquires a site, we deploy fund-owned ASICs financed through equipment debt secured against the hardware. At $48 per MWh realized power and base-case hashprice, the fund generates approximately $17M of annual operating cash flow on $50M of LP equity. That cash flow funds the 8 percent annual preferred distributions to LPs and produces meaningful current yield throughout the hold. The exit is where the LP multiple is created. We assemble four fund-acquired sites into a 60 MW anchor portfolio, expand to 100+ MW through co-invest and bolt-on acquisitions in years 3 and 4, and sell the assembled portfolio to a hyperscaler or institutional infrastructure buyer at $1.5 to $3.0M+ per MW. The buyer is paying a premium for what we already locked in: energized power, signed PPAs, and locations now sitting on the regional fiber buildout. Operating cash flow from self-mining funds the annual yield. Selling the assembled 100+ MW portfolio is what creates the multiple.
Fund Structure: Real Estate / Infrastructure Model
We are not a hosting company. We acquire ERCOT West power infrastructure (land, energized substation, bilateral PPA) at a 3 to 5 times discount to institutional comp. The fund deploys ASICs at every site to generate operating cash flow during the hold, then aggregates four fund-owned sites plus co-invest into a 100+ MW portfolio and sells it to a hyperscaler or institutional infrastructure buyer in years 3 to 5.
How it works, end to end
LPs commit capital to the fund
Total raise $50M with a $60M hard cap. Minimum commitment $250K friends-and-family or $1M institutional. The GP commits 5 percent of the fund. Capital is called as deals close, not at fund close.
Fund acquires four mid-cap ERCOT West sites
Acquisitions occur during the first 18 months at $400 to $600K per MW, totaling 60 MW across four fund-owned sites. Each site includes the land, the energized substation, the bilateral PPA, the building shell, and fiber where available. The fund owns each site outright.
Fund deploys ASICs and operates each site
ASIC capex is financed through equipment debt at 70 percent loan-to-value, secured against the hardware itself. The fund earns 100 percent of mining revenue at the realized power cost it already controls. At base-case hashprice and Peacock-equivalent economics, the operating fleet generates approximately $17M of annual cash flow on $50M of LP equity. Cash flow funds the 8 percent annual preferred return and produces meaningful current distributions during the hold.
Fund expands the platform toward 100 MW
During years 3 and 4, the fund extends from 60 MW to 100+ MW through three channels: bolt-on acquisitions funded by retained operating cash flow plus follow-on capital up to the hard cap; co-invest sleeves on individual deals offered to LPs and adjacent capital; and affiliate aggregation at exit, where fund sites are packaged with Gasbox-affiliated assets for institutional sale.
Fund exits the assembled portfolio
In years 4 to 5, the assembled 100+ MW portfolio is sold to a hyperscaler, an institutional infrastructure buyer (Brookfield, DigitalBridge, Stonepeak, Energy Capital Partners), or a public miner conducting roll-up M&A. Target net MOIC of 2.7 to 4.5 times on the LP's $50M of equity, depending on the exit comp realized.
Infrastructure PE (The Analogy)
Infrastructure funds acquire toll roads, pipelines, and cell towers — assets with stable cash flow and capital appreciation driven by long-term demand growth. Investors receive distributions during the hold and lump-sum proceeds when the asset is sold to a strategic buyer or recapitalized.
American Power Fund (This Fund)
The fund acquires ERCOT West power sites at $400 to $600K per MW. The fund deploys ASICs at every site for operating cash flow during the hold, then aggregates four fund-owned sites plus co-invest into a 100+ MW portfolio for sale to a hyperscaler or institutional infrastructure buyer. LPs receive current yield from operating cash flow and capital appreciation from the portfolio sale.
The fund's higher target MOIC versus traditional infrastructure PE is driven by two distinct value drivers stacked together. First, we acquire mid-cap sites at a 3 to 5 times discount because institutional capital cannot underwrite sub-$25M deals. Second, we aggregate the assembled portfolio into a 100+ MW package that institutional buyers will pay a premium for. Both spreads accrue to the LP.
The Capital Stack
$50M of LP equity controls approximately $123M of deployed capital through equipment debt and reinvested operating cash flow.
The fund is structured like a real-estate or infrastructure deal: LP equity sits at the bottom of the stack, equipment debt sits in the middle, and operating cash flow recycles into the next phase of deployment. The result is that a $50M raise funds a 56 MW operating fleet plus 4 site acquisitions plus working capital, rather than just one of those line items.
Total Capital Deployed: ~$123M
Source 1 · LP Equity
$50M
The fund's full raise. Pays for three things:
- Site acquisitions: $30M (4 sites × 15 MW × $500K per MW)
- Phase 1 ASIC equity: $13M (30% down on Year 1 hardware)
- Operating reserves: $7M (working capital and contingency)
Source 2 · Equipment Debt
~$60M
Two tranches secured against the ASIC fleet itself, deployed in sequence:
- Phase 1: ~$30M at first close (28 MW deployment)
- Phase 2: ~$30M in Year 2 (additional 28 MW)
- 60–70% LTV, 5-year amortization, market rates, debt service paid from mining cash flow
Lender candidates: Galaxy Digital Mining Finance, Foundry (DCG), Antalpha (Bitmain). Term sheet to be finalized concurrent with first acquisition.
Source 3 · Reinvested Cash Flow
~$13M
Year 1 operating cash flow funds the Phase 2 ASIC equity, eliminating the need for an additional LP capital call:
- Year 1 net cash flow on 28 MW: ~$8M
- Plus Q1–Q2 Year 2 cash flow: ~$5M
- Funds Phase 2 ASIC equity portion in full
How The 8% Preferred Return Gets Paid
The pref is paid out of operating cash flow from self-mining, not out of fresh LP capital and not out of exit proceeds. The math comfortably supports the obligation across the hold:
Preferred Obligation
$4M / year
8 percent on $50M of LP equity
Steady-State Cash Flow
~$17M / year
56 MW deployed, after equipment debt service
Pref Coverage
~4×
Excess distributes to LPs as additional current yield
In a deep-bear hashprice scenario where mining cash flow temporarily falls below the pref obligation, the fund holds $7M of operating reserves (sufficient to cover the pref for 1.75 years at zero operating revenue). This provides downside protection without diluting LP equity.
The Investment Thesis
Four reasons mid-cap ERCOT West is the highest-conviction infrastructure trade in North America right now.
1 · Mid-cap ERCOT sites are structurally orphaned
Public miners (Marathon, Riot, CleanSpark) need transactions of 50 MW or larger to move their production guidance. Institutional infrastructure capital (Brookfield, DigitalBridge, Stonepeak) will not underwrite sub-$25M deals. Hyperscaler capacity buyers want 100 MW commitments. Sites in the 10 to 15 MW range fall through every institutional buying mandate. They sit on the market longer, they clear at lower prices per MW, and they trade through relationships rather than auctions. The result is a persistent 3 to 5 times pricing gap between mid-cap sites and institutional packages of the same energized infrastructure.
2 · ERCOT West is becoming the cheapest hyperscaler power in North America
The Permian Basin produces more associated natural gas than its takeaway pipeline capacity can move. The stranded molecules clear into ERCOT West generators at sub-$30 per MWh marginal cost. This is the cheapest large-scale electricity in the continental US, and the cost advantage is structural rather than cyclical. Hyperscalers have figured this out. 10+ GW of AI data center capacity is currently permitted, in construction, or announced in the region. That capacity has to be built somewhere, and the somewhere is ERCOT West. We acquire energized infrastructure inside that buildout zone at mining-economics prices.
3 · The fiber buildout is the value catalyst
ERCOT West is currently fiber-thin. The installed fiber is adequate for Bitcoin mining but insufficient for the latency profile AI compute requires. That gap closes inside the fund's hold period. Hyperscalers deploying their signed PPAs in the region are forcing carriers to extend dark fiber routes to the substations connecting their loads. The fund acquires substations inside that buildout footprint, sometimes the exact substations the hyperscaler PPAs route through. Our power infrastructure was acquired at mining-grade prices. By year 3 or 4, the same infrastructure sits on AI-grade fiber. Every quarter that passes, our land, PPAs, and transformers become more valuable to the next buyer.
4 · Aggregation is the value-creation lever
A single 15 MW site sold to a public miner clears at $1.0 to $1.2M per MW because the buyer pool is narrow. A 100 MW portfolio sold to a hyperscaler or institutional infrastructure buyer clears at $2.0 to $3.0M per MW or higher, because the buyer is acquiring a ready-to-deploy compute campus rather than a single site. The fund's primary value-creation lever is aggregation. We acquire 60 MW directly with the $50M raise, expand to 100+ MW through bolt-on acquisitions and co-invest in years 3 and 4, and exit the assembled portfolio. The spread between mid-cap acquisition prices and institutional aggregation prices is the trade.
Why 100 MW Changes Everything
The exit-comp curve is non-linear. Aggregation is the single largest value-creation lever in the fund.
Mid-cap ERCOT sites trade at $400–600K/MW because no buyer with capital is structured to pay more for a 15 MW deal. The exit comp curve, however, is not linear. As portfolio scale increases, the buyer pool widens, the buyer's per-MW willingness to pay increases, and the unit becomes attractive to capital that could not touch the sub-scale form. The fund is structured to traverse this curve.
15 MW Single Site
$1.0–1.2M/MW
Public miners, regional aggregators, family offices. Narrow bidder pool.
60 MW Mini-Portfolio
$1.5–2.0M/MW
Mid-cap public miner roll-up, infrastructure capital co-invest, second-tier hosting operators.
100+ MW Portfolio · Base Case
$2.0–3.0M/MW
Brookfield, Stonepeak, ECP, hyperscaler subcontractor portfolios, and the major public miners as buyers.
100+ MW · AI-Grade Fiber
$3.0–5.0M/MW
Hyperscalers direct, neoclouds (CoreWeave, Lambda, Nebius, Crusoe). The buyer pays for ready-to-deploy AI compute capacity.
How We Get From 60 MW To 100 MW
The $50M raise funds the acquisition of 60 MW directly across four fund-owned sites in years 1 and 2. The fund expands toward 100+ MW during years 3 and 4 through three concrete channels:
- Bolt-on acquisitions. Operating cash flow retained in the fund plus follow-on capital up to the $60M hard cap funds the acquisition of one or two additional sites totaling 8 to 15 MW.
- Co-invest sleeves. Anchor LPs and adjacent capital are offered direct co-investment positions in deals beyond the fund's core, allowing the operating platform to scale without exhausting fund equity.
- Affiliate aggregation at exit. When liquidity arrives, the fund packages its sites alongside Gasbox-affiliated assets and partner-controlled sites into a single 100+ MW portfolio for institutional sale.
LPs in the fund participate proportionally in the aggregation premium regardless of which channel delivers the additional MW. The fund is structured so that the 100+ MW exit comp accrues to the fund's portion of the assembled portfolio.
Operating Model
Self-mine the capacity we own. Lease only when a tenant offers above our self-mine margin.
Primary Strategy
Self-Mine
Fund deploys ASICs (13 J/TH at $20K/PH) at the sites it owns. ASIC capex is financed via equipment debt (60–70% LTV) plus fund equity, repaid out of mining cash flow. The fund earns 100% of the hashrate revenue, not a fixed hosting rate.
Opportunistic Strategy
Lease to Tenants
The fund leases capacity to a tenant only when the offered hosting rate exceeds the fund's self-mine margin. In practice this is almost always an AI tenant on a fiber-grade site. Mining tenants offering $55–70/MWh do not clear the bar. At those rates the fund earns more by self-mining.
Why Self-Mine Is Primary
The fund owns the most expensive part of mining (energized power infrastructure) and finances the cheapest part (ASICs, via equipment debt). At $48/MWh realized power, the fund's self-mine margin is structurally higher than what any third-party miner can pay it as a hosting tenant — because that miner has to clear the same cost stack the fund already owns. Leasing to a mining tenant means giving away the spread we already captured by buying the site. The only time leasing wins is when a tenant brings demand the fund can't replicate itself — fiber-grade AI capacity at premium hosting rates. Until that tenant shows up, every MW the fund owns is more valuable in the fund's own ASICs.
Self-Mining Profit Mechanics
Per-MW economics under base-case Peacock conditions. Move the slider to see how hashprice changes the math.
Base Case · Per 1 MW · Annual
$48/MWh power · $38/PH/day · 13 J/TH · 95% uptime
Revenue & Cost Stack
Capital Math
Self-Mine vs. Lease · Same MW, Same Site
Self-Mine (Primary)
$572K
net cash flow per MW per year at base case ($38/PH, $48/MWh power)
Lease to Mining Tenant @ $65/MWh
$141K
net per MW · ($65 − $48 − $5) × 8,322 MWh · the fund gives away the spread it already owns
Lease to AI Tenant @ $85/MWh
$266K
net per MW · still below self-mine, but fiber-grade AI tenants typically clear at $90–110/MWh
The decision rule is simple: if a tenant offers more than the fund's self-mine net cash flow per MW (currently ~$572K/yr at base case), accept the lease. Otherwise, deploy ASICs. At today's hashprice, that bar is roughly equivalent to a hosting rate of $120/MWh — well above what mining tenants pay. AI tenants on fiber-grade sites can clear it; in those cases the fund leases. Otherwise it self-mines.
ERCOT West: The Fiber Inflection
Why the same site is worth fundamentally more in 24 months than it is today.
The Power Side
Stranded Permian Gas
The Permian Basin produces 6+ Bcf/d of associated gas, which is natural gas that comes up as a byproduct of oil production. Takeaway pipeline capacity to the Gulf is fully subscribed, leaving meaningful volumes stranded at the wellhead. ERCOT West generators monetize this gas at sub-$30/MWh marginal cost. The result is the cheapest large-scale electricity in the continental US, structurally below national grid costs by 40%+ for the foreseeable future.
The Demand Side
10+ GW Already In Motion
Hyperscalers have figured out the Permian power math. 10+ GW of AI data center capacity is permitted, in construction, or announced in ERCOT West. Pacifico Energy alone permitted 7.65 GW in February 2026. Those projects have to be built somewhere, and the somewhere is the same ~150 mile corridor that contains every site the fund is targeting. Hyperscaler capex is moving toward our power infrastructure, not away from it.
The Gap
Fiber Is Behind
ERCOT West is currently fiber-thin. Today's installed fiber is sufficient for Bitcoin mining; it is not sufficient for the latency profile AI compute demands. That gap is the inflection. Hyperscalers deploying signed PPAs are forcing carriers to extend dark fiber routes to the substations connecting their loads. By the time the fund exits, those routes pass within reach of the sites the fund acquired today.
Why This Re-Rates Our Sites
A mid-cap ERCOT West site without AI-grade fiber clears at $1.0–1.2M/MW today. The same site with AI-grade fiber clears at $3.0–5.0M/MW. The fund's underwriting does not require fiber to arrive at any specific site by any specific date. Base-case returns are calculated assuming fiber has not arrived. In that scenario, sites exit as mining-grade infrastructure to public miners and infrastructure capital. Fiber arrival is upside, and based on currently announced hyperscaler PPAs, it is upside the fund expects to capture on at least a portion of the portfolio. The asymmetry is the trade. Every quarter our sites sit unsold, the regional fiber buildout adds value we did not pay for.
What's Already Being Built In ERCOT West
We are not the only people who think this region matters. The receipts for the demand-side thesis.
The fund's underwriting does not depend on speculative future demand. 10+ GW of data center capacity is currently permitted, in construction, or announced in the Permian / ERCOT West corridor, the same corridor where the fund is acquiring sites. The projects below are the public, third-party-reported evidence. Each is operated by a different developer, financed differently, and at GW scale. Collectively they account for the regional power-demand and fiber-buildout dynamics the fund's thesis depends on.
GW Ranch
Pacifico Energy
Largest permitted data center campus in the U.S.
Pacifico Energy received TCEQ approval in February 2026 for what the developer describes as the largest permitted data center campus in the United States. Behind-the-meter gas-fired generation directly addressing Permian stranded gas economics. $5.5B in tax abatements approved alongside a 200-unit workforce housing development.
KCBD News · Feb 2026Pecos Campus
Core Scientific
Public Bitcoin miner pivoting 1.5 GW to AI compute
Core Scientific (NASDAQ: CORZ) is repurposing its existing Pecos Bitcoin mining capacity into AI-grade data center capacity, with CoreWeave/Poolside as the anchor tenant pipeline. Financed via a $3.3B senior secured note raise plus a $1B Morgan Stanley facility. This is the in-the-flesh proof case for the BTC-to-AI conversion path our fund preserves as upside optionality.
Data Center Frontier · Oct 2025Project Horizon
Poolside (with CoreWeave)
The cautionary tale, and why mid-cap structure matters
Originally announced October 2025 with CoreWeave as 15-yr anchor tenant on the first 250 MW. Project reportedly stalled in April 2026 after CoreWeave exited and Poolside's $2B Series C with NVIDIA stalled. Worth including: the demand thesis is real (the project was announced because the math works) but execution at single-developer GW scale carries concentration risk that mid-cap aggregation does not.
Poolside · Project Horizon AnnouncementWhat This Means For The Fund
GW-scale projects are being built in the same 150-mile corridor where the fund is buying 15 MW sites. We are not betting on demand; we are buying inside the demand zone. The Pacifico, Core Scientific, and Poolside projects together represent ~10 GW of announced capacity, which is orders of magnitude larger than the fund's 60 MW core. Two implications follow. First, the regional fiber, gas, and grid buildout that those projects require lifts the value of every adjacent piece of infrastructure, including ours. Second, Poolside's restructuring shows the execution risk of single-developer GW-scale deployments, a risk that mid-cap aggregation across multiple sites and multiple buyers structurally avoids.
Further Reading · Regional Context
Permian gas-powered AI buildout: Chevron, ExxonMobil, Diamondback entering
Texas Standard / Houston Public Media
Local pushback in Pecos: Texas towns and the data center expansion
Governing
Massive AI data center coming to Pecos County
Big Bend Sentinel · Feb 2026
Permian Basin gas-powered data center · emissions and permitting
Oil & Gas Watch · Feb 2026
Pacifico GW Ranch project page
Pacifico Energy
Core Scientific 8-K filing summary
StockTitan
Site-Level Economics
Toggle between Self-Mine (primary) and Lease (opportunistic) to model unit economics under each strategy.
This is the fund's primary operating mode. The fund deploys ASICs at every site it owns (13 J/TH efficiency, approximately $20K per PH of hashrate). ASIC capex is financed through equipment debt at 70 percent loan-to-value and repaid out of mining cash flow over five years. The fund earns 100 percent of mining revenue at the realized power cost it already controls. At Peacock-equivalent economics, this is the highest-margin use of the power the fund acquires.
Self-Mine P&L (Annual)
Yield on ASICs
37.2%
Payback
3.6 yr
ASIC capex: $15.38M · 769 PH at $20K/PH (13 J/TH gen). Financed via fund equity + equipment debt.
Self-Mine Notes
At 10 MW deployed, $38/PH/day hashprice, and $48/MWh realized power, the fund generates $5.73M of annual net cash flow. That is a substantial profit center on top of the fund's $4M annual preferred return obligation. ASIC capex of $15.38M is financed via equipment debt at 60–70% LTV; the equity portion is repaid out of mining cash flow. Yield on ASIC capex: 37.2%.
What The Fund Owns
Real assets with bilateral PPAs. Cash flow during hold. Lump-sum proceeds at exit.
The Real Assets
Land Parcels
ERCOT West & Central counties
Owned fee simple
Bilateral PPAs
Direct contracts with generators or REPs
3–5+ year terms
Substations
Transformers, switchgear, interconnect
Energized infrastructure
Building Shell + Container Pads
ASIC or GPU tenant ready
Move-in ready
Fiber Connectivity
Preserves AI tenant optionality
Where AI-grade
Self-Mine ASIC Fleet
Primary cash flow generator, financed via equipment debt + fund equity
~14 MW per site · ~1,077 PH
How LPs Get Paid
During Hold (Years 1–5)
Self-mine cash flow (~$8M/yr per 14 MW deployed) → 8% pref + excess to LP
Quarterly cash distributions
At Exit (Year 3–5)
Site sale: capital return + 80% of profits above pref
Lump sum proceeds
Fee Waterfall
Mgmt / LP preferred return / LP-GP carry split
2% / 8% pref / 80-20
GP Commitment
Aligns GP with LP outcomes
5% of fund
Min Commitment
Anchor LPs ($2M+) get LPAC seat + reduced mgmt fee
$250K F&F · $1M Inst.
Capital Timeline
From commitment to exit. What an LP actually experiences.
How This Fund Works
American Power Fund is a commitment-based vehicle. LPs do not wire their full check at signing. They commit a dollar amount, and capital is called as the fund acquires sites.
Commit
Month 0
Sign subscription docs. Wire $0. Commitment is binding but uncalled.
First Close
Months 0–6
Friends-and-family close at $5–10M. Institutional close takes fund to $50M target / $60M hard cap.
Capital Calls
Months 6–24
Pro-rata calls as each site closes. First call ~20–25% of commitment. 4–6 sites over the 24-month investment period.
Hold & Distributions
Years 1–5
8% preferred return paid quarterly from operating cash flow. Site-by-site liquidity events return capital as exits occur.
Exit Waterfall
Years 3–5
Capital returned + 80% of profits above pref. GP earns 20% carry. European waterfall with clawback.
Commit
Month 0
Sign subscription docs. Wire $0. Commitment is binding but uncalled.
First Close
Months 0–6
Friends-and-family close at $5–10M. Institutional close takes fund to $50M target / $60M hard cap.
Capital Calls
Months 6–24
Pro-rata calls as each site closes. First call ~20–25% of commitment. 4–6 sites over the 24-month investment period.
Hold & Distributions
Years 1–5
8% preferred return paid quarterly from operating cash flow. Site-by-site liquidity events return capital as exits occur.
Exit Waterfall
Years 3–5
Capital returned + 80% of profits above pref. GP earns 20% carry. European waterfall with clawback.
Inflows (LP Pays In)
Month 0 · Sub docs signed
$0
Month 6 · Site 1 capital call
~$220K
Month 12 · Site 2 capital call
~$200K
Month 18 · Site 3 capital call
~$200K
Month 24 · Site 4 capital call
~$200K
Total called
~$820K
Outflows (LP Receives)
Qtrs 3–20 · Pref @ 8%
~$330K cumulative
Years 3–5 · Capital return + carry waterfall
$2.0M–4.5M (base to upside case)
Total to LP
$2.3M–$4.8M
MOIC on called capital
2.8×–5.8×
Illustrative cadence. Actual timing depends on deal flow and site-level acquisition schedules. Mgmt fee waived on uncalled commitments.
Base case assumes single-site exits at $1.0–1.2M/MW. Upside case assumes portfolio exit at $1.5–2.0M/MW.
Investment Period
24 months
After this, GP cannot call new capital except for follow-ons or fund obligations.
Fund Life
5 yrs + 2× 1-yr extensions
Maximum 7 years from final close.
Recycling
First 18 months only
Proceeds from early exits may be redeployed; thereafter, all proceeds distribute to LPs.
Why This Structure
Commitment-based capital alignment means GP fees and LP capital are both tied to actual deployment. We do not earn management fees on uncalled commitments, and you do not pay for capital that has not gone to work. The fund's incentive is to deploy intelligently, not to gather assets.
Live Deal: Proof of Thesis
The fund's first acquisition. The arbitrage in real numbers.
Project Peacock · 15 MW · Reeves County, Texas
Full deal pageActive acquisition. Turnkey, energized ERCOT West site, 400 Mbps fiber on-site (adequate for ASIC ops; AI-grade upgrade tracked against regional fiber buildout). Acquired at $500K/MW ($7.5M). Realized power $48/MWh at 95% uptime with 4CP avoidance, $60/MWh at 99.9%. Self-mine plan: 14 MW deployed at $1.54M/MW ASIC capex, 70% equipment-financed. Single-site exit downside $16.5M (2.2× on basis) · 60 MW portfolio packaging $22–30M attributable · 100 MW hyperscaler exit $30–45M attributable.
The Acquisition
Capacity
Energized; transformer-engineered for 14 MW deployable
15 MW interconnect
Location
ERCOT West · inside the 10+ GW AI data center buildout corridor
Reeves County, TX
ESA Structure
Realized $48/MWh at 95% uptime w/ 4CP avoidance · $60/MWh at 99.9%
Constellation Index Solutions
Fiber
Mining-grade today · AI-grade upgrade tracked against regional buildout
400 Mbps on site
Acquisition Price
$7.5M · fund acquisition at standard mid-cap pricing
$500K / MW
Self-Mine Plan
ASIC capex $21.5M · 70% equipment debt · ~$8.0M/yr gross mining cash flow at $38/PH
14 MW · ~1,077 PH
The Exit Math · Three Paths
Single-site is the downside floor. Portfolio aggregation to 100 MW is the base case. AI-grade fiber is the upside.
A · Single-Site Exit (Downside Floor)
Exit comp
$1.0–1.2M / MW
On 15 MW
$15–18M gross
Gross MOIC on basis
2.7×–3.3×
IRR over 4 years
~22–28%
Buyer pool: regional aggregators, mid-cap public miners, family-office Bitcoin shops. This is the modal outcome. The fund works at this number.
B · 100 MW Portfolio Exit (Base Case)
Exit comp
$1.5–2.0M / MW
On 50–60 MW aggregated
$75–120M gross
Fund-level gross MOIC
4.0×–5.5×
IRR over 5 years
~32–40%
Buyer pool: Brookfield, Stonepeak, Energy Capital Partners, public miner roll-ups. Requires fund-level packaging at year 5.
C · AI Conversion (Tail Case)
Exit comp
$2.5–4.0M / MW
On 15 MW w/ AI lease-up
$37–60M gross
Gross MOIC on basis
5×–10×
IRR over 3–4 years
~45–60%
Buyer pool: CoreWeave, Lambda, Crusoe, Nebius, neoclouds, hyperscaler subcontractors. Requires fiber-grade site + tenant conversion. Possible on Peacock; not in base case.
How We Think About Exit
Exit comp ranges depend on transaction structure, not just site quality. A single 15 MW site sold to a public miner clears at $1.0–1.2M/MW because the buyer pool is narrow and the buyer applies a sub-scale discount. A 100 MW portfolio sold to a hyperscaler or institutional infrastructure buyer clears at $2.0–3.0M/MW because the buyer is acquiring a ready-to-deploy compute campus, not a single site. AI tenant conversion clears higher still. The fund underwrites to single-site economics as the downside floor and targets portfolio aggregation to 100 MW as the base-case exit.
Why This Deal Exists At This Price
Public miners are saturated in their existing ERCOT footprints and not adding 15 MW deals at current hashprice levels. Infrastructure capital won't underwrite a sub-$25M transaction. The result is a thin bidder pool for a fair-priced, energized asset. This is exactly the inefficiency the fund is built to capture.
Exit Value Modeling
Fund-level returns at exit. Adjust the exit comp and hold period to see how the spread translates into LP distributions.
How To Read This
The fund deploys $50M across ~60 MW at $500K/MW. Self-mining cash flow during the hold (approximately $8M/yr per 14 MW deployed at base case) funds the 8% pref and meaningful excess distributions, modeled separately and not double-counted here. The model below shows the exit-only return when sites are sold at the prevailing institutional comp. Operating distributions during the hold are additive to these numbers.
Exit Waterfall ($50M Fund)
Acquisition Basis
60 MW × $500K/MW
$30.0M
Gross Exit Value
60 MW × $2.00M/MW
$120.0M
↳ LP Capital Return
100% of LP commitments back
$50.0M
↳ Pref Distributions (Hold)
8% × 5 years from operating cash flow
$20.00M
↳ LP Share of Profit (80%)
Capital appreciation above acquisition
$56.00M
↳ GP Carry (20%)
Performance allocation to GP
$14.00M
LP Returns
Net MOIC
at $2.0M/MW exit, 5-yr hold
2.5×
Net IRR
Annualized return to LPs
20.3%
Total Cash to LPs
Capital + pref + 80% of profit
$126.0M
$1M LP example
Profit: $1.52M over 5 years
$2.52M total
GP Carry Earned
20% of profit above LP capital + pref
$14.00M
Market Context: Why ERCOT, Why Now
The macro tailwinds driving site comp appreciation.
ERCOT vs. National Power Cost
ERCOT West has averaged ~40% below national industrial power costs for five years running.
Green: ERCOT West · Gray dashed: National avg ($/MWh)
ERCOT Data Center Load Growth
Approved + queued load on track to 16× from 2022 to 2027.
ERCOT Mid-Cap Site Comp · Historical Trajectory
A 15 MW site that traded for $6.75M in 2020 trades for $31.5M today, a 4.7× appreciation in six years. Public miner M&A in ERCOT is currently clearing at $2.5–4M/MW for fully stabilized institutional assets. Mid-cap remains 3–5× below.
Fund Competitive Advantages
Why this team will close deals others can't.
Mid-Cap Deal Flow Access
Established relationships with ERCOT site developers and brokers. Mid-cap deals are not auctioned — they trade through relationships. We are on the call list when sub-15 MW inventory comes loose.
ERCOT Market Specialization
Concentrated entirely in the only US power market combining stranded gas, wind oversupply, bilateral contracting, and large flexible load policy. Regional expertise turns into deal underwriting speed.
Operator-Grade Diligence
GP team has built and run mining sites at 100+ PH scale. We know what to inspect on a substation, what to discount on an ESA, what to walk away from. Most institutional capital underwriting ERCOT lacks this depth.
Four Exit Channels, Not One
Public miners. Colocation operators (Compute North, Applied Digital, Iren). Infrastructure capital (Brookfield, DigitalBridge). AI/hyperscaler capacity buyers. We exit to whoever is bidding hardest at the time of liquidity.
AI Tenant Optionality
Sites with adequate fiber can lease to AI tenants at $70–90+/MWh, which is 25 to 50 percent above mining rates. Not every site qualifies, but selected sites carry that optionality at no incremental acquisition cost.
Hashprice Cycle Upside
Bitcoin hashprice currently sits approximately 90 percent below its 2021 all-time high. Self-mining gives the fund and its LPs direct exposure to the recovery. Costs are largely fixed (power, debt service, opex), so every dollar of hashprice improvement flows through to operating cash flow with leverage. Monthly distributions through the hold are paid in BTC or USD at LP election. We are structurally bullish hashprice into the next halving-driven cycle.
The Macro Thesis
Three forces that compound site comp appreciation through the fund's hold period.
AI Infrastructure Is the Defining Capital Allocation Story of 2025–2030
Microsoft, Google, Amazon, and Meta have collectively committed $400B+ in AI infrastructure capex through 2027. The bottleneck is not compute. It is power. Every hyperscaler is competing for permitted, grid-connected land near transmission. ERCOT is the only US market that can deliver 50+ MW capacity within 18 months on bilateral PPAs.
Bitcoin Miners Are the Accidental Infrastructure Developers
Over the past five years, Bitcoin miners built out 10+ GW of permitted, interconnected power infrastructure in ERCOT, faster than any data center developer in history. They did it cheaply (no building costs, modular ramp) and were willing to do the slow regulatory work that incumbents avoid. Now that capacity is being repriced upward as AI compute demand arrives.
The Stranded Energy Tailwind Is Permanent
West Texas wind and Permian Basin stranded gas create persistent reservoirs of sub-$30/MWh power that do not exist in other ISOs. ERCOT's energy-only market means the price reaches its marginal value, even when load is light. Mid-cap sites that can monetize this energy at hosting rates earn margin through the cycle.
Historical ERCOT Site Comp Context
Mid-cap sites have re-rated in every prior cycle.
Cycle 2020–2021
Avg Site Comp
$650K/MW
Mid-Cap Spread
$155K–$465K
Acquisition price for sub-15 MW sites
Pre-China ban → US migration · early ERCOT acquisitions
Cycle 2022–2024
Avg Site Comp
$1.1M/MW
Mid-Cap Spread
$275K–$825K
Acquisition price for sub-15 MW sites
Post-FTX consolidation · institutional comps stabilize
Cycle 2025–2026 (Current)
Avg Site Comp
$2.1M/MW
Mid-Cap Spread
$525K–$1.6M+
Acquisition price for sub-15 MW sites
AI compute overlap · public miner M&A clearing
Frequently Asked Questions
LP-grade diligence questions specific to the ERCOT mid-cap thesis.
For the full data room including ESA terms, exit comps, and capital stack details, contact jarret@gasbox.org
Current Operations
Gasbox currently operates mining infrastructure across 3 active sites, with live, revenue-generating deployments across the U.S. and internationally.
Active Sites
3
Total Hashrate
284 PH
Machines Online
1,216
Avg. Uptime
96%
Site-by-Site Breakdown
Iowa
Wind/Grid Blend
Brazil
Hydroelectric
Texas
Natural Gas
Iowa
Wind/Grid Blend
Hashrate
183 PH
MW
4.58
Machines
784
BTC/Day
0.194
Brazil
Hydroelectric
Hashrate
47 PH
MW
1.18
Machines
201
BTC/Day
0.050
Texas
Natural Gas
Hashrate
54 PH
MW
1.35
Machines
231
BTC/Day
0.057
Total · All Sites
Hashrate
284 PH
MW
7.10
Machines
1,216
BTC/Day
0.301
Daily BTC Output (All Sites)
0.301 BTC/day
Gross before operating costs & splits
Infrastructure Footprint
7.10 MW
Estimated power consumed across all sites
Fund Principals
American Power Fund I is run by operators, not financial engineers. Both principals have personally built MW-scale Bitcoin mining infrastructure, written the software that runs it, and managed LP capital and investor relations across prior Gasbox vehicles. Every part of the playbook this fund is executing, the principals have run before.
Jarret Porter
Managing Partner
Co-Founder of Gasbox. Has built and operated MW-scale Bitcoin mining infrastructure for nearly a decade, both on-grid in Iowa and off-grid on stranded flare gas in the Permian, drawing on a background in natural gas power generation. Currently operates live hashrate across three active sites in Iowa, Brazil, and West Texas. Built the Gasbox market intelligence terminal, a Bloomberg-style platform for Bitcoin network and mining economics. Operator-grade fluency across the full mining stack: substation engineering, ASIC procurement, energy markets, and fleet dispatch. Leads investor relations and LP capital for Gasbox. Before Bitcoin, built consumer lending businesses. Served as CMO of Loanry.com. Founded Niche Loans, an online lending fund of consumer-based assets that included NoCreditCampers.com and PayMyWedding.com, and sold the company in 2018, taking proceeds into Bitcoin mining. Graduated from New Mexico State University, where he played college baseball.
📞+1 (512) 710-8777
Justin Chiang
Co-Principal
Hardware engineer with deep specialization in AI compute infrastructure. His domain is GPU cluster topology, high-density power delivery, and the substation-to-rack engineering that hyperscalers actually pay for. Graduated from UC Berkeley with a double major in electrical engineering. Architected and maintains Gasbox's payment infrastructure software. The platform has shipped millions in distributions to LPs across prior Gasbox vehicles and converts ongoing mining revenue into USD bank deposits, stablecoins, or equity at each LP's election. Leads hardware procurement, datacenter electrical design, ASIC and GPU fleet management, and AI-readiness diligence on every site the fund acquires. Co-principal on the American Mining Fund. Active real estate operator running JM Estates. The cross-stack engineer fluent in both the power side and the compute side of the buildout, and the technical lead on the bridge between them.
📞+1 (858) 997-4690
Gasbox LLC · ERCOT Power Infrastructure · Texas
This is a confidential investment document. Only distribute to qualified investors.
GASBOX
Confidential · Invitation Only
Step 1 of 2
Enter your email to begin.
This document is distributed exclusively to invited parties. Your email is logged for access audit purposes.
Confidential investment document. Restricted to qualified, invited investors only. Unauthorized distribution is strictly prohibited.