Hyperchargers on one device
One customer runs 30 high-power chargers at up to 1,000 kW each through a single Edge device – including load management in power classes where every wrong decision costs real money.
NeLeSo Edge L · AI appliance
The AI appliance for large sites and high-power charging: a local operator portal, AI fault diagnostics and automatic optimization of charging profiles – with no cloud required if you choose. Field-proven with 30 hyperchargers at 1,000 kW each on a single device.
From the field
One customer runs 30 high-power chargers at up to 1,000 kW each through a single Edge device – including load management in power classes where every wrong decision costs real money.
Edge L manages 100 charge points and more on one device – OCPP server, load management, operator portal and AI all running together on one fanless box in the control cabinet.
Fully operable offline: no cloud, no remote access, no data leaving the site. The remote support tunnel remains strictly opt-in – and can be disconnected at any time.
Local AI · 100+ TOPS
Edge L brings the analysis to where the data is created: an NPU with more than 100 TOPS runs compact, quantized models directly on the device. No cloud LLM, no API key – your OCPP logs, sessions and meter values never leave the site.
To be honest: these are compact local models, not a replacement for a large cloud LLM. For diagnostics, forecasting and tuning at the site, that is exactly the right size of tool – fast, offline and completely under your control.
Illustrative run of the on-device AI diagnostics: a charge point reports a fault at night, Edge L retrieves similar cases from its own OCPP logs via RAG, delivers a finding and proposes adjusted charging profiles – all on the device, no cloud involved.
Feature set
OCPP 1.6J and 2.0.1 locally, dynamic load management via Modbus RTU/TCP, SunSpec and EEBUS, PV surplus charging, offline-first with optional cloud sync – the complete Edge foundation, just with far more headroom.
CPMS Core directly on the device: charge points, sessions and faults in the browser at the site – no cloud login, even in air-gapped operation.
Anomaly detection on sessions and MeterValues, load forecasting for the site and local fault diagnostics via RAG over your own OCPP logs.
The AI derives optimized charging profiles and limits from your site history – in suggestion mode for approval or fully automatic, your call.
Billing compliant with German calibration law (Eichrecht), retrofittable directly on the device – from €2,900 one-off plus €49/month.
100+ charge points per device and high-power chargers up to 1 MW – field-proven with 30 hyperchargers at 1,000 kW each in production.
Datasheet
| Platform | Industrial NPU box with 100+ TOPS and 16 GB RAM – standard industrial hardware with the NeLeSo stack pre-installed |
|---|---|
| Interfaces | 2× ETH · RS485 · CAN |
| Form factor & environment | Fanless · DIN rail/wall mount · −20…+60 °C |
| Capacity | 100+ charge points · high-power chargers up to 1 MW |
| Operating modes | Offline-first · fully air-gap capable · optional cloud sync · opt-in remote tunnel |
| Updates | Signed A/B updates with automatic rollback · maintenance windows under your control |
| Delivery | ≤ 10 working days · 24-hour burn-in in our test lab · first-boot wizard |
| Warranty | Extended hardware warranty to 5 years from €790 |
Hardware: standard industrial device from an established manufacturer with the NeLeSo stack pre-installed (distributor model) · All prices net, plus VAT.
FAQ
Yes. All models run on the device’s NPU, and fault diagnostics uses RAG over the OCPP logs stored locally. Edge L is fully air-gap capable – no internet connection is required for operation, and the remote support tunnel remains strictly opt-in.
The AI derives optimized charging profiles and limits from your site’s history – sessions, meter values, load curves. In suggestion mode you review and approve every change before it is applied; automatic mode can be enabled once you trust the suggestions.
Yes. Through the optional connection to a central platform – yours or ours – several Edge sites can be monitored together. Each site stays autonomous: if the connection drops, operation simply continues locally.
Data sovereignty, plain and simple. OCPP logs and operational data never leave your site. Compact local models on the NPU are all it takes for fault diagnostics, anomaly detection and forecasting – Edge L is deliberately not a general-purpose cloud LLM replacement.
Next step