Plexus · Distribution Grid Intelligence

The distribution grid has capacity. We find it.

Plexus runs always-on power flow on your distribution grid, using data you already collect. It shows what can connect, and what to do when nothing can.

See Plexus in action How it works
Plexus hosting capacity view: the published ICA map for a Greensboro, NC feeder, with the agent roster, live grid status and thermal overloads
The problem

The grid's most dynamic layer doesn't know what it can handle.

New load and DER are arriving faster than distribution was ever planned for. The next decade brings more.

20–30%
DER penetrationOn grids built to carry power one way. It climbs every year as electrification, data centers and new DER stack up.
$19B+
Infrastructure liabilityRacked up by US utilities since 2018. This is what it costs to learn an asset's real limit too late.
~50%
Capacity sitting unusedAverage US utilization at peak. The headroom is real. Nothing in the planning stack can prove where.

All three come back to one question: what can still connect, and what to do when nothing can.

What Plexus does

What can connect. And what to do when nothing can.

Know what can connect
A hosting capacity map for the whole feeder, always current. Built on real power flow, not a nameplate estimate.
Know what to do when it can't
Every constraint comes back named: which element, which hour, what it costs to fix. Priced before you commit capital.
Know it on today's grid
The model re-runs when your network changes. Answers reflect the feeder as it stands today, not last year's study.
The Platform

One model. Any short-term planning question.

Every agent runs against one live model of your network, built from data you already collect and solved with full AC power flow. Adding a new agent takes days.

Built from data you have
No new hardware
Meter reads, network models, load shapes. The files your planners already send to regulators.
Full AC power flow
Regulator-grade studies
A real solve on the real network, every hour of the year. No surrogate models.
Condition derating
Capacity that reflects age
Where you have condition data, capacity reflects it. Everything else runs at nameplate.
Running today Hosting Capacity Reliability Risk NWA Dispatch Transformer Protection Line Loss DER Interconnection Load Interconnection + build your own
Why Plexus

Most tools give you an approximation.

Nameplate models, AI surrogates, one-off studies. None of them is the grid you are running today.

8,760
hours of AC power flow behind every answer
<30 min
to re-run the full year when the feeder changes
<$5
what that refresh costs in compute
12 weeks
from utility data to first pilot result
Watchfuly
Real physics, always current
AC power flow, re-run against the live network with real asset condition. Every answer reflects today's grid.
Approximations & snapshots
A model of the model
A learned approximation, or a nameplate snapshot. Neither one is your grid.
Legacy study tools
Queued by hand
An engineer sets up each study by hand. Milsoft, the standard at most co-ops, has no native bidirectional power flow.
Data & workflow layers
Easier access, same answers
They make existing data and workflows easier to use. The physics underneath does not change.
Condition derating
Asset condition →
Real capacity
Nameplate
→ Power flow
Other tools stop at a health index. Plexus turns asset condition into the circuit parameters the solver actually uses: kVA, ampacity, fault rate.
How it works

Plexus runs on the data you already have.

No new hardware. Point Plexus at your feeder model and load data and it runs, on-premises or in your own cloud VPC. Nothing leaves your environment.

Your existing data
Your network model, load shapes, and SCADA or AMI.
Works with your data today
Plexus
Solves full AC power flow across the year, derated by asset condition.
Agents
Flag risks, price fixes, and answer interconnection requests as conditions change.
Day 1
Feeder model + load
Topology, nameplate ratings, line impedances, and load profiles at the transformer level. That is enough to go live.
No customer PII or billing data. Feeder topology is the same file you already send to FERC.
Enrichment
Asset condition → derating
A condition score per transformer, from inspection records, oil samples, or just age and loading history.
Not required to go live. Add it over time. Every record improves the result.
Higher fidelity
Interval smart meter data
Better per-load diversity if you have it. Class profiles work if you do not.
Plexus walks around the AMI wall. If you have it, we use it. If not, we work without it.

Ready to see your grid clearly?

Drop your name and email and we'll set up a walkthrough. No commitments.

EPRI OPAI Consortium Member  ·  Esri Startup Partner  ·  Research Triangle Park, NC