Decentralized Infrastructure · AI Economy
Decentralized energy infrastructure for the AI economy.
PowerLink develops closed-loop power systems for locations the grid cannot economically reach — converting stranded gas, waste streams, critical minerals, and hydro into on-site power.
We believe that whoever controls energy and compute will control the next industrial era.
The decentralized utility layer for AI.
Infrastructure for sovereign compute.
The bridge between energy and intelligence.
See how the platform works.
The next AI bottleneck is electricity.
The grid was built for a distributed, low-density world. AI is a concentrated, hyperscale load. The infrastructure gap is structural, not cyclical.
Median time in queue to commercial operation for U.S. projects reaching COD in 2024.[1]
RTO clearing price, 2025/2026 delivery year ($269.92/MW-day) vs. 2024/2025 ($28.92/MW-day).[3]
Sources: [1] LBNL, Queued Up (2025 ed.). [2] Wood Mackenzie transformer survey, Q2 2025. [3] PJM 2024/2025 BRA results. Figures as of dates shown.
The market conditions cited above are third-party industry data describing grid-wide constraints. They do not guarantee demand for PowerLink's solutions, that PowerLink will secure customers, or that the Company will successfully execute its business plan.
Centralized Power Is Breaking.
The grid was engineered for a different era. AI demands infrastructure the grid was never designed to deliver.
Power Demand vs Grid Capacity
Illustrative index of AI-driven power demand vs. incremental grid capacity, 2024–2035. Demand path derived from third-party projections [LBNL, Goldman Sachs, IEA, 2024]; capacity path and index construction reflect PowerLink modeling and are illustrative, not a forecast of actual MW.
Methodology and key assumptions: both series are indexed to a common 2024 base value rather than plotted in MW. The demand path applies the growth rates implied by published third-party projections of U.S. data-center and AI electricity demand [LBNL, Goldman Sachs, IEA, 2024] through the sources' forecast horizons, then extends them to 2035 at progressively slower growth rates assumed by PowerLink. The capacity path assumes that incremental grid capacity actually reaching commercial operation grows only about 1–2% per year, reflecting the interconnection-queue and equipment lead-time constraints cited in the preceding section. Both paths are PowerLink management assumptions for illustration only; actual demand and capacity may differ materially.
Legacy Architecture
Residential / Commercial- 01Designed for distributed municipal demand
- 02Average grid infrastructure age 40+ years[13]
- 03New transmission lines typically take 10–15 years[13]
- 042,600 GW in U.S. interconnection queues at year-end 2023; approximately 2,290 GW at year-end 2024[4]
Decentralized AI Infrastructure
PowerLink- 01Built for distributed AI loads under 25 MW
- 02Modular generation deployed at the source
- 03Reduces dependence on interconnection queues
- 04Targeted to deploy in months rather than multi-year grid timelines, subject to permitting, site readiness, and commissioning
Deployment-timeline basis: the months-rather-than-years target is a management estimate based on the modular, containerized design of PowerLink's systems. It is not based on a completed PowerLink deployment — no PowerLink site has yet been deployed at commercial scale — and actual timelines will depend on permitting, site readiness, equipment availability, financing, and commissioning.
In our view, the question is not whether decentralized power wins — it is who builds it.
Stranded resources, converted into reliable power at the source.
The model: one feedstock, multiple intended revenue streams — tipping fees on the way in, energy and (where deployed) compute on the way out, recovered materials in between. Capabilities shown reflect the platform as designed; deployment varies by site and stage.
01 / Inputs
02 / PowerLink Platform
03 / Outputs
How the model is designed to scale.
Each deployment is intended to generate multiple revenue streams that can be reinvested into additional nodes. Actual economics will depend on execution and are not guaranteed.
- 01
Acquire stranded resources
Long-duration feedstock access targeted.
- 02
Deploy modular systems
Containerized or warehouse-scale systems.
- 03
Generate reliable power
Closed-loop generation at the source.
- 04
Monetize data center compute
Intended recurring revenue per MW where deployed.
- 05
Recover materials
Steel, carbon black, diesel, propane, as designed.
- 06
Reinvest to fund the next site
Capital intended to be recycled across the portfolio.
Forward-looking. Describes the intended business model, not results or projected returns. Subject to feedstock availability and pricing, downtime, commodity and power-price volatility, financing, and regulatory conditions. No assurance of profitability or that cash flow will be reinvested as described.
Who builds it.
Operators, engineers, and capital builders with infrastructure pedigree.
Brad Hoagland
Chairman and CEO
CFA charterholder. Former CFO of Ecoark Holdings. Built PGIM's London office from inception to $2B AUM in six years. Bucknell B.S. Economics.
Dan Koehler
COO
Digital infrastructure and executive operations. Leads operations at Kratos Digital Mining. International best-selling author on ethical leadership.
Gene Taylor
Chief Global Development & Integration Officer
30+ years in U.S. and geopolitical business strategy. Capital formation in complex and underserved markets. Cross-cultural stakeholder integration.
Michael Pollack
Chief Global Financial Officer
35 years in public accounting. Advised 100+ publicly traded companies and 250+ private firms. Reverse mergers and SEC initial registration.
Britt Swann
VP of Finance
Financial modeling and underwriting for new ventures. Led finance and corporate development for Vantage entities. $500M equity raise and MLP IPO experience.
Jeff Fehlan
Chief of Hardware Engineering
40+ years in design, engineering, and manufacturing. Civil, aerospace, and semiconductor pedigree. Michigan State B.S. Agricultural Engineering Technology.