Energy attribution platform

Know what each customer actually costs you — measured, not estimated.

GreenShift reconciles real datacenter power — host, process, and PDU/BMC signals — with live Kubernetes workload telemetry, and attributes it down to the VM, node, namespace, pod, workload, and tenant that used it.

  • Spin-off from the IPCEI-CIS research project EMULATE (FH Dortmund)
  • In pilot deployment with an IT service provider since Q3 2026; letter of intent signed with a regional carrier
  • TOP 10 finalist, KUER.NRW Business Plan Competition 2026
Why GreenShift

The trust boundary

Every operator knows total site consumption down to the watt. Which tenant caused it? Nobody knows — measurement stops at the rack and PDU, and visibility into tenants' guest systems is off-limits for good reason. That's exactly the boundary that EU energy labelling requirements (EnEfG) and your customers' CSRD sustainability reports now require you to cross. Estimates don't hold up to audit; flat-rate allocation cross-subsidizes inefficient tenants.

In September 2026 the EU adopted a mandatory energy and water efficiency labelling scheme for data centres of 500 kW and above, with the first labels due in August 2027.

Two measurement layers, one reconciliation

GreenShift combines a physical reference measurement at the PDU with relative load shares from a software agent — and continuously reconciles the two. The result: attribution per tenant and workload based on actual measurements, not modelling assumptions.

What this means for you: audit-ready, customer-level energy disclosure from live measurement data instead of an annual one-off analysis; the end of cross-subsidy, since consumption is billed to whoever causes it, and can be passed through to tenants to refinance the service itself; and a tenant portal and reports that can be white-labelled in your own branding — GreenShift stays the technology supplier in the background.

Built for real infrastructure

Design principles, not afterthoughts.

Fail open, outbound-only, explicit freshness, server-side authorization, Prometheus-native metrics and data minimization — the constraints the architecture is built on.

Read the design principles →