BITF — knowledge base

Overview

Bitfarms (BITF) is shifting from internationally diversified Bitcoin mining toward North American power and data-center infrastructure for AI and high-performance computing (HPC), while retaining Bitcoin mining as an operating and monetization option. The strategy centers on converting Pennsylvania energy assets acquired through Stronghold into HPC campuses, prioritizing the United States and reducing emphasis on Latin America.

Management intends to own durable power and data-center infrastructure rather than GPUs, which have shorter economic lives and require recurring replacement capital. Initial development is focused on Sharon, Washington and Panther Creek, with Scrubgrass representing a larger but earlier-stage opportunity. Execution depends on interconnection approvals, customer commitments, specialized partners, regulatory conversions and substantial site-specific financing.

CEO Ben Gagnon summarized the capital-allocation philosophy: “Yeah, I think I'm clearly a Bitcoin maxi at heart, but the other thing, I'm also a profit maxi.” [[s:76@00:13:06]]

Key facts & figures

Management says the Stronghold acquisition was underwritten primarily around converting Pennsylvania assets to AI/HPC use, not simply operating power plants; this internal rationale is unverifiable without supporting transaction records. [[s:76@00:17:03]]
Bitfarms claims more than 1 GW of prospective Pennsylvania AI/HPC capacity, but the figure is unverifiable and includes conceptual capacity without assured interconnection or construction. [[s:76@00:22:32]]
Management cited purported firm commitments for 80 MW at Sharon in late 2026, 50 MW at Panther Creek in late 2026 and an additional 300 MW at Panther Creek around January 2028; contract terms and delivery dates remain unverified. [[s:76@00:32:57]]
Scrubgrass remains at the conceptual-study stage. Management believes it could exceed 1 GW within three to four years, but conceptual load studies do not guarantee interconnection, financing, construction or energization. [[s:76@00:34:56]]
A $300 million Macquarie facility is intended to finance Panther Creek’s first 50 MW building and associated first-phase infrastructure; whether it fully covers the project is unverified and subject to scope, budget and cost overruns. [[s:76@01:03:25]]
Later Panther Creek phases and developments at other sites are expected to require separate project-level financing.
Planned HPC power contracts would use firm utility service at market prices, with energy-price exposure passed through to customers rather than retained by Bitfarms.
Smaller sites are expected to target enterprise customers, potentially at higher margins; large campuses are intended for hyperscalers and neocloud providers.
Bitfarms expanded its executive, board and advisory capabilities and is using partners including T5 for specialist data-center development and operations.
Quebec power allocations may be contractually or tariff-restricted to cryptocurrency mining and could require conversion before supporting AI/HPC; the specific requirement is unverified without utility contracts and applicable tariff rules. [[s:76@01:00:51]]
Waste-coal generation can remove Pennsylvania refuse piles associated with spontaneous fires and acidic, metal-contaminated runoff, although combustion also creates emissions and ash requiring controls; the underlying reclamation claim is accurate. [[s:76@00:39:33]]
Management’s description of Pennsylvania generation as overwhelmingly natural-gas-driven was misleading: gas is the largest source and coal has declined, but nuclear remains material and renewable development is significant. [[s:76@00:37:19]]
The claim that coal-refuse runoff reaches an “Appalachian River,” then the Missouri River and Gulf was inaccurate; western Pennsylvania generally drains through the Ohio and Mississippi, while eastern areas drain toward the Susquehanna, Delaware, Chesapeake Bay or Atlantic. [[s:76@00:41:23]]
A claim that Pennsylvania has about 12 coal-refuse power plants and doubled and extended related tax credits through 2036 is unverified pending current plant and legislative records. [[s:76@00:41:51]]
Riot reportedly held approximately 26.5 million Bitfarms shares, just below 5%, following an August 18 filing; the figure and resulting disclosure implications are unverified without current securities filings. [[s:76@01:06:55]]

Thesis & bull case

Scarce power pipeline: Existing energy assets and PJM interconnection positions could allow Bitfarms to reach energized AI/HPC capacity faster than greenfield competitors if claimed delivery schedules hold.
Infrastructure rather than hardware exposure: Owning power and data-center facilities avoids direct ownership of rapidly depreciating GPUs and the recurring capital required to refresh them.
Multiple monetization paths: Sites can continue supporting Bitcoin mining while being developed for potentially longer-duration and more predictable HPC infrastructure revenue.
Customer segmentation: Smaller facilities could address enterprise demand at higher margins, while large Pennsylvania campuses could attract hyperscalers and neoclouds requiring hundreds of megawatts.
Risk transfer: Passing market electricity costs through to HPC customers could protect project margins from power-price volatility, subject to actual contract terms.
Experienced partners: T5 and an expanded management, board and advisory bench may reduce Bitfarms’ data-center capability gap.
Project-level capital: Site-specific financing can limit reliance on unrestricted corporate capital and better align debt with contracted project cash flows.
Pennsylvania upside: Scrubgrass could become a transformational campus if conceptual capacity progresses to approved interconnection, customer contracts and financing.
Environmental and policy angle: Coal-refuse generation can support land reclamation and cleanup, potentially improving local and political support despite associated emissions.
Capital allocation: Management defended share repurchases as opportunistic when BITF traded materially below its assessment of intrinsic value, potentially increasing per-share exposure to the power pipeline.
Gagnon’s framing of the Stronghold rationale was explicit: “And at BitFarms, we were underwriting the transaction as the ability to convert them off into HPC and AI.” [[s:76@00:17:11]]

Risks & bear case

Execution gap: Bitfarms is transitioning from Bitcoin mining into specialized data-center development, construction and operations, where it lacks the established record of incumbent operators.
Pipeline quality: More than 1 GW of claimed Pennsylvania potential should not be treated as deliverable capacity; much of the Scrubgrass opportunity remains conceptual.
Interconnection risk: PJM studies, firm-service arrangements, network upgrades and energization schedules may be delayed, resized or rejected.
Customer risk: No disclosed hyperscaler, neocloud or enterprise customer commitments were identified. Capacity without bankable leases may be difficult to finance.
Capital intensity: The $300 million Macquarie facility addresses only the intended first 50 MW Panther Creek phase; later phases and other sites need additional capital.
Dilution and leverage: Financing several campuses may require equity issuance, expensive debt, joint ventures or asset-level concessions, reducing shareholder economics.
Construction risk: Data centers require complex cooling, networking, redundancy and power-quality systems. Cost inflation, design changes and delays could undermine returns.
Partner dependence: Reliance on T5 and other external specialists reduces capability gaps but introduces counterparty, coordination and economics-sharing risks.
Timing risk: Initial HPC revenue is not expected until late 2026 or early 2027, leaving a lengthy period of spending before cash generation.
Bitcoin exposure: Mining economics, Bitcoin prices, network difficulty and machine efficiency remain relevant while sites await conversion.
Regulatory risk: Quebec sites may need tariff or contract changes; Pennsylvania projects face utility, environmental, permitting and policy uncertainty.
Environmental liabilities: Burning coal refuse can aid reclamation but still produces air emissions and ash, potentially creating compliance costs and opposition.
Geographic retrenchment: Reduced focus on Latin America may cause impairments, exit costs or weaker returns on previously developed assets.
Buyback trade-off: Repurchasing shares during a capital-intensive pivot may preserve per-share value if the stock is undervalued, but it also consumes liquidity needed for development.
Management credibility: Several site, financing and capacity claims remain internally sourced or unverifiable; delivery against milestones is necessary before assigning full value.
Interview framing included factual errors unrelated to the core strategy, including the inaccurate claim that Bitcoin moved from roughly $0.73 to $1,000 within 12 months; the move occurred over roughly three years. [[s:76@00:07:52]]
The statement that Sherbrooke is Quebec’s largest city outside Montreal was inaccurate; Quebec City and several other municipalities are larger. [[s:76@00:59:11]]

Timeline of developments

2025-09-15: CEO Ben Gagnon presented Bitfarms’ strategic pivot toward North American AI/HPC energy infrastructure, identified Sharon, Washington and Panther Creek as the fastest prospective routes to HPC revenue, described Scrubgrass as a potentially greater-than-1-GW but conceptual opportunity, and outlined T5 involvement, pass-through power pricing, a $300 million Macquarie facility and infrastructure-focused ownership rather than direct GPU investment. [[s:76]]

Open questions

Which Pennsylvania megawatts have executed utility agreements, completed interconnection studies and binding energization dates rather than conceptual or prospective status?
What are the exact milestones, covenants, interest costs, collateral and permitted uses under the $300 million Macquarie facility?
Does the Panther Creek facility cover the complete first 50 MW phase, including network upgrades, substations, buildings, cooling and contingency?
When will Bitfarms announce binding AI/HPC customer contracts, and will they be take-or-pay agreements sufficient for project financing?
What lease rates, contract durations, escalation clauses, utilization assumptions and return thresholds does management target?
How much equity or corporate support will be required for Panther Creek’s later 300 MW phase and for Sharon, Washington and Scrubgrass?
What portion of the claimed Pennsylvania pipeline is firm, studied, queued or merely conceptual?
Can late-2026 energization be maintained amid PJM queue congestion, equipment lead times and construction constraints?
What economics will T5 receive, and which development, construction and operating responsibilities remain with Bitfarms?
How will Bitcoin mining operate during conversion, and what revenue or asset impairments could result from taking mining capacity offline?
Which Latin American assets will be retained, sold, scaled down or impaired?
Can Quebec mining allocations be converted to HPC tariffs, on what timeline, and at what power price?
How will coal-refuse generation’s reclamation benefits compare with emissions, ash-management and regulatory costs?
What customer profile produces the best risk-adjusted returns: enterprise, hyperscaler or neocloud?
How much capital was used for buybacks, at what average price, and how does that compare with near-term project funding needs?
Has Riot fully exited or merely fallen below the 5% reporting threshold?

Notable predictions to track

Late 2026: Sharon is expected to receive 80 MW of firm capacity; verify energization, customer contracting and revenue commencement. [[s:76@00:32:57]]
Late 2026: Panther Creek’s initial 50 MW is expected to become available; verify whether the Macquarie facility fully funds completion and whether capacity is leased.
Late 2026 or early 2027: Management expects the fastest Pennsylvania sites—Sharon, Washington and Panther Creek—to begin generating initial HPC revenue.
Around January 2028: Panther Creek is expected to receive an additional 300 MW; verify utility delivery, project financing and tenant commitments.
Within three to four years: Scrubgrass could exceed 1 GW, although it remains at conceptual-study stage and should be treated as a contingent projection rather than committed capacity. [[s:76@00:34:56]]
Future HPC contracts are expected to pass market electricity costs through to customers, limiting Bitfarms’ energy-price risk.
Smaller enterprise-oriented sites are expected to produce higher margins than large hyperscaler or neocloud campuses.
Bitfarms expects infrastructure ownership to produce more durable economics than owning GPUs because GPU fleets depreciate quickly and require frequent replacement.