"*" indicates required fields
Energy Saver from Zinkworks is an AI-powered predictive energy management solution that automates telecom network energy efficiency. Most RAN power is consumed when there’s little or no traffic. Energy Saver forecasts demand at cell level using a patented graph and sequence model, then reduces unused capacity automatically in real time. The platform is designed around closed-loop network automation and aligns with TM Forum AN L4.
The problem
Mobile networks were architected for the peak hour. But peak demand only lasts for a small part of the day. For most hours, base stations still draw near-peak power while network load drops by as much as 10×. Operators end up paying full price to run half-empty cells, and the bill keeps climbing.
* Illustrative model
Operators are paying peak-hour energy costs around the clock. In many cases, €20–30M of every €100M energy bill goes to keeping the RAN always on, even during low-traffic periods. Estimate your savings
Modelled across 100,000+ cells in a Central European Tier-1 5G estate, using Zinkworks’ field-validated savings and €120/MWh wholesale power rates.
Projected energy reduction across capacity cells, while maintaining coverage and QoS.
Drag the dials to match your footprint.
Energy Saver makes the RAN demand-aware. It forecasts cell-level traffic and applies the right mix of power-saving actions through standard RAN interfaces. By learning from live network outcomes, it reduces energy use while protecting coverage and QoS.
Every prediction is compared with real network outcomes, so Energy Saver learns what works and improves with every deployment.
A patented graph and sequence forecasting engine built for cross-cell prediction. Instead of analysing cells in isolation, it learns traffic behaviour across neighbouring cells and forecasts demand from minutes to hours ahead, informing every downstream action.
Each lever stands alone — operators start with one, then layer in the rest. KPI guardrails preserve quality of service throughout.
Switch carriers off as the forecast falls. Bring them back ahead of the next traffic ramp. Coverage on the primary carrier is preserved throughout.
Dim antenna Tx power to match the served footprint. Fewer users, smaller cell, less radiated energy. Cell-edge users protected by KPI guardrails.
When load stays below threshold for a sustained window, power down non-essential components and enter deep sleep. Wake automatically on demand. The deepest lever per base station.
Scale virtualised network functions with forecast load. Release compute when demand drops, reserve it ahead of the next peak. Cloud-native; scaling lands as Kubernetes / VNF lifecycle events.
All figures modelled. Per-network results vary with traffic mix, carrier configuration and starting baseline.
Energy Saver is a closed-loop RAN energy optimisation product. It coordinates four levers — DCAD (Dynamic Carrier Adaptation), ATPC (Adaptive Transmit Power Control), LBSM (Load-Based Sleep Management) and VNF-DS (Virtualised Network Function Dynamic Scaling) — to cut energy use across your base-station estate without touching coverage or KPIs.
With all four levers enabled, validated network modelling shows roughly 30 kWh/day/base station of saved energy. At a typical €0.28/kWh tariff that’s about €3,066 per base station per year — so a 5,000-base-station estate saves around €15.3M/year, and a 10,000-base-station estate roughly €30.7M/year.
No. Every action is policy-bound, KPI-guarded, and reversible. Energy Saver watches load, throughput, latency and call-drop metrics in real time and automatically rolls back any action that breaches a guardrail. You can also run it in shadow mode first to confirm savings before activating live writes.
Energy Saver is vendor-agnostic. It integrates via standard O-RAN SMO interfaces and runs against the major RAN vendors. We can tailor the demo to your specific SMO target, vendor mix, and operational lifecycle.
Most operators move from demo to a shadow-mode pilot within weeks, not
months. The demo is structured around three concrete steps:
Which of DCAD, ATPC, LBSM, VNF-DS are already in flight in your network? Where would you want to start — and what would you want to leave alone? A 90-minute working session.
100–200 cells with good data and clear KPI guardrails Baseline kWh and € per cell, per day. Agree what “success” looks like in writing.
Deploy NTP + PMC on the cluster. Compare against baseline. Decide what to scale, what to refine, what to drop together.
Request a tailored walkthrough of the closed-loop optimisation engine, the four energy levers, and a savings projection grounded in your real base-station count and tariff.
