At the intersection of metal lattices, hydrogen, and materials engineering, LENR points toward a new energy architecture.
This page organizes LENR around three axes: reaction mechanism, validation status, and deployment scenarios.
LENR refers to a research area exploring excess heat and possible nuclear-scale effects that may emerge from interactions between hydrogen and engineered metal lattices. The deck frames it under condensed-matter nuclear science (CMNS) rather than conventional high-temperature fusion.
This page does not treat the claims as settled fact. Instead, it restructures the attached material into a readable web narrative around mechanism, evidence, and possible deployment paths.
Special processing creates vacancies and cracks, treating the reaction field itself as an engineering variable.
High-density hydrogen in defect sites is presented as the central motif linking the whole deck.
The narrative aims at reproducibility, materials design, and application development—not just theory.
The material proposes the following four-stage hypothesis. No mechanism producing practical nuclear reaction rates under low-energy conditions has been established.
The deck says engineered vacancies and cracks are created inside the metal lattice to define the reaction field.
The material assumes hydrogen concentration at defect sites. The “one million” factor has not been independently verified by AIfrontier.
High hydrogen density, electron screening and quantum tunneling are being investigated as possible influences on reaction rates. No practical low-energy mechanism is established.
Electrochemical loading, gas pressure, and electrical stimulation are presented as ways to initiate the reaction.
Clean Planet calls its nickel-based nano-composite and hydrogen heat technology “QHe” and reports testing and product development. QHe is the company’s technology name; independent, standardized verification of COP or commercial performance has not been established here.
A 2025 Nature paper reported about a 15% increase in neutron-producing D-D fusion driven by external ion acceleration after electrochemical deuterium loading. It did not report direct helium-3 detection, excess heat, or net energy. Limited re-evaluation does not constitute scientific acceptance of reproducible LENR heat.
Clean Planet and Miura announced a joint-development agreement for an industrial boiler using QHe as a heat source. AIfrontier is not identified as a party to that agreement. Distributed power remains a future-use hypothesis.
A natural fit where decarbonization and efficient heat supply matter at the same time.
Supplying power near computation loads aligns with the deck’s infrastructure-scale framing.
The conclusion explicitly links the technology story to climate response and the post-fossil energy transition.
Beyond one product category, the deeper value may lie in defect control, catalysts, and reaction-system engineering.
This page does not provide test conditions, measurement boundaries, or independent reports supporting completed claims of 750-hour operation, COP 4.0+, a 10 kW demonstration, finished independent validation, a production date, or ¥1/kWh. They are not presented as achieved results.
Assessment requires whole-system heat balance, uncertainty analysis, consistency with nuclear products, independent replication, durability, safety, and techno-economic analysis including capital cost. Timing, output and pricing remain undecided.
Home battery VPP operations and distributed AI data centers, run by one control system.
Where LENR asks about the power source itself, LNER asks how to bundle the power and compute that already exist. Batteries and heat go into homes; compute and batteries go into sites; and one control layer treats both as a single supply-demand system.
Two markets, sold from the same assets
There are two places to sell. The power market (ΔkW in the balancing market, the capacity market, wholesale and retail) and the compute market (inference sold by the hour, API access). When one side's prices fall, the other keeps the same assets earning — a structural difference from a battery-only aggregator, or a compute-only operator.
A four-layer model — from physics to markets
LAYER 0 · HOMES
Supplying flexibility and customer relationships
Home batteries in the 13 kWh class, smart distribution boards, heat-pump water heating and HVAC, EV and V2H. Homes are not where compute is placed — they are the layer that secures large amounts of low-cost flexibility, able to offer the grid both upward DR (charging) and downward DR (discharge and curtailment).
LAYER 1 · SITES (NODES)
Where compute and heat live
On the roofs and spare space of apartment buildings, commercial facilities and warehouses, under low-voltage commercial contracts (20–50 kW): liquid-cooled inference nodes, industrial batteries, rooftop solar, and heat-pump waste-heat recovery. When grid connections take years, an existing connection is itself an asset.
LAYER 2 · CONTROL — LNER Orchestrator
Solving power and compute against one objective
Per-device metering, aggregation of low-voltage resources, demand and price forecasting, market bidding and settlement, inference workload scheduling, load shifting between sites, imbalance management, and audit. LNER's core asset is not the hardware but this layer and its operating data.
LAYER 3 · MARKETS & CUSTOMERS
Stacking revenues so policy shifts don't decide the outcome
The balancing market, capacity market and wholesale/retail; inference API customers; heat offtakers; municipalities and finance. Rather than depending on a market whose unit price is set by policy, the plan grows self-consumption optimisation, inference and heat — revenues that move independently of subsidies.
Dual flexibility — coordinating storage and time-shiftable compute in one optimiser
Night
Low prices
Battery charge overnight
Inference batch jobs
Morning
Ramp-up
Battery standby, reserves
Inference normal load
Midday
Solar surplus, cheap
Battery charge from surplus, sell upward DR
Inference run at maximum
Evening
Peak, high prices
Battery discharge, sell downward DR
Inference curtail, defer, shift to other sites
Late night
Prepare for tomorrow
Battery ΔkW standby
Inference batch jobs
The core of the design
A battery stores electricity; a GPU turns electricity into computation. Run them separately, and the first is exposed to falling unit prices while the second is exposed to power costs and connection limits. LNER makes battery charge/discharge and the time or location shift of deadline-flexible compute jointly controllable, strengthening both at once. A compute queue does not store energy; it shifts demand in time. Only customer-approved batch workloads are considered flexible.
Don't throw the heat away
Heat recovered from liquid cooling is raised in temperature by a heat pump and sent to facility hot water and heating, or to neighbouring consumers. Unlike electricity prices, the avoided cost of heat does not follow market prices, which makes it a stabiliser for revenue. PUE 1.3 or lower is a voluntary design target. PUE does not measure compute efficiency or heat-use value, and legal applicability depends on facility conditions.
Policy tailwinds
Japan’s balancing market began in FY2021. Use of low-voltage small resources and per-device metering was designated to begin in FY2026 subject to required system work; actual eligibility depends on product, region and implementation date. Support for waste-heat utilisation and PUE regulation push in the same direction. The ground for aggregation as a business is already laid.
Evaluation gates — not a committed deployment plan
Years, household counts and site counts are undecided. Any pilot depends on regulatory eligibility, financing, customer consent, grid connection, equipment procurement and sufficient time-shiftable workloads. Interactive inference and workloads with strict response-time obligations are excluded by default.
Where it meets LENR
LENR deals with the power source itself; LNER deals with how the power and compute that already exist are bundled. If LENR advances to practical deployment, it could connect to LNER's site nodes as a distributed source. The two proceed independently, however — LNER does not depend on LENR becoming practical. What they share is a stance of never treating energy and computing separately.
Current status
A private organization presenting LENR-related research. This page does not establish the scope of any contract, joint research, or independent validation.
Visit WebsiteA company pursuing LENR commercialization. A person with the same name as AIfrontier’s representative is publicly listed as its director; it is not presented as an independent validator. The cooperation scope requires confirmation.
Visit Website