← PEAKSHIFT / COMMUNITY POWER

A community power program for data centers

Give every home a battery.
Take back the peak.

Your project is stuck on two things money can't rush: a place in the interconnection queue and a town that wants you gone. Funding a small battery in every nearby home moves both. Neighbors pay less on the hottest days. You get roughly 100 megawatts of evening peak relief that never enters a study, installed in months.

100,000 homes ~100 MW peak relief 0 interconnection studies Months, not years

The bind

Two clocks are running, and neither one takes a check.

The site is ready. The power isn't. A new substation and the transmission to feed it run into the hundreds of millions and take years, if the queue clears at all. While you wait, the neighborhood shows up to the hearing to say no. Every month a 100 MW build sits idle is revenue you never earn back, and pouring money at the problem doesn't buy back the calendar.

3–5 yrTypical wait for a large interconnection and substation upgrade
$100M+Cost of the grid upgrade you're waiting on, before it powers a single rack
The townThe one input capital can't accelerate, and the one that stops projects cold

The reframe

The grid isn't short on power. It's short on room at 6 p.m.

The wall in front of you is the evening peak, the two hours when every air conditioner in the neighborhood runs at once and the local wires hit their limit. That peak lives in the homes around your site. Move it a few hours earlier and you free the exact headroom your load needs, on the exact feeders you're trying to connect to.

Neighborhood demand today Shifted by home batteries

The offer

Give every home a battery. It does the rest on its own.

A boombox-sized battery sits in the home and plugs into a wall outlet. The window AC plugs into the battery. No electrician, no panel work, no hole in the wall.

01 / NIGHT

Charge when it's cheap

The battery fills overnight, when power is cheapest and the grid is quiet.

02 / PEAK

Cover the 4–9 p.m. crunch

Through the evening peak the home's AC runs off the battery instead of the grid.

03 / ALWAYS

Never push back

No power flows to the grid, so there's nothing to permit and nothing to study.

Already built

The hardware is off-the-shelf. The software that schedules it is open source and running today. It's called PeakShift, it's brand-agnostic, and it already reads live electricity prices across the US, UK, Europe, and Australia.

What the neighborhood gets

The first data center that lowers the electric bill.

“We stopped fighting the data center when it started paying our July bill.”

The sentence you want said at the zoning hearing

Lower bills

Running the AC on cheap overnight power cuts the summer bill that hurts households most.

Backup power

When the grid goes down, the battery keeps the AC and the fridge running.

Nothing to buy

It's free, it's already installed, and it costs the household nothing to run.

What you get

About 100 megawatts of relief, and a community on your side.

~100 MWCoincident-peak reduction on the feeders you need, across 100,000 homes
0Interconnection studies, because nothing exports to the grid
MonthsTo deploy, at the speed of shipping and plugging in a box
Your 4CPShaving the system coincident peak lowers your own capacity and transmission-demand charge, the same intervals that set it
The honest version

This doesn't generate your power. It shaves the system peak during the exact intervals that set capacity charges (ERCOT 4CP, PJM PLC), which lowers your own bill in dollars, and it turns the neighborhood from opponent into beneficiary. For a project stuck on distribution capacity and a permit, that's the constraint that decides whether you build.

The math

100,000 homes. 200 MWh. Less than a year of delay costs you.

100,000 homes × 2 kWh200 MWh
Coincident peak relief~60–100 MW
Hardware, you fund it once$60–120M
At fleet / OEM volumeless
Capacity value, every year$6–10M/yr
Each household nets~$200/yr + a free battery
Net, against the alternativea bargain

The centralized version of this relief is a substation upgrade. It costs about the same or more, buys zero goodwill, and arrives years later behind its own queue. You fund the hardware once, and it pays back every year by shaving the exact intervals that set capacity charges, including your own. The household's offer is the free battery plus a lower bill plus backup power, not a token check, which is what keeps them enrolled. Measured against the revenue a finished campus earns, the program can cost less than the delay it removes.

It already works

Utilities and startups are already buying distributed over central.

ConEd, Brooklyn & Queens

Spent about $200M on distributed power instead of $1B on new wires to serve growing load, and kept the lights on.

Base Power, Texas

Raised $1B to put batteries in homes and trade the pooled capacity into the ERCOT market. The model is funded and scaling.

US Dept. of Energy

Its VPP Liftoff work puts pooled home batteries at roughly 40% the net cost of a utility-scale battery for the same service.

Every Electric, NYC

Gives free AC batteries to apartments and gets paid by Con Edison to shave the peak. Live today, homes earning $150 to $200 a season.

The pilot

Pick one site. Ten thousand homes. One summer.

A 10,000-home pilot around a single site runs roughly $6 to $12 million and proves the three things that matter: real peak reduction measured on your target feeders, a community that shifts from fighting you to thanking you, and dispatch economics from live grid programs. It scales the same way software does, by adding sites in a config file, not by pouring another foundation.

Mike German Steps Ventures mike@stepsventures.com

PeakShift is open source at github.com/steps-re/peakshift

Figures are planning estimates for discussion, drawn from public program data and typical hardware pricing as of 2026. Peak-relief and revenue numbers depend on local rates, climate, home participation, and grid-program terms, and would be sized against a specific site before commitment. The no-export design keeps the program behind the meter, so it needs no interconnection. Any future grid-export option is regulated per market and is not part of this offer.