Sidewalk Infrastructure Partners builds the things the grid forgot.

It is 2026, and the grid is the bottleneck

Every conversation about artificial intelligence eventually runs into the same wall: a power outlet. Training models and running data centers takes electricity - a great deal of it - and the wires carrying that electricity were designed for a calmer century. This is the wall Sidewalk Infrastructure Partners spends its days pushing against.

SIP is an infrastructure developer and holding company. It does not behave like a venture fund chasing the next app, and it does not behave like a utility waiting for permission. Instead it finds the parts of the physical world that are creaking - the power grid, the data center, the recycling stream - and builds operating businesses to fix them. Two of those businesses, Renew Home and Verrus, are now running in the real world rather than on a slide.

We kept building more, when we should have built smarter

For a hundred years, infrastructure had one answer to rising demand: build another one. Another power plant. Another lane. Another data center humming somewhere off a highway. It worked, until it didn't. The cost of "another" climbed, the permits piled up, and the climate bill came due.

The founders saw a different lever. The grid already has enormous slack hiding inside it - a water heater that could run an hour earlier, a thermostat that could nudge itself a degree, a data center that could lean on its batteries for ninety seconds while the grid catches its breath. Coordinate millions of those small moves and you get the equivalent of a power plant, without pouring a single foundation. The problem was never a shortage of electrons. It was a shortage of flexibility.

Two Sidewalk Labs veterans, one unfashionable idea

In 2017, inside Alphabet's Sidewalk Labs, Jonathan Winer and Brian Barlow started working on the unglamorous layer beneath smart cities - the pipes, wires, and roads. In 2019 they spun the effort out as an independent company. A year later, in May 2020, they raised $400 million from Alphabet and the Ontario Teachers' Pension Plan. Patient money for a patient problem.

Their bet was that the gap holding infrastructure back was not technical but social: technologists and infrastructure operators rarely sit in the same room. So SIP appointed itself the convener - developing a thesis, proving a fix, then dragging engineers, incumbents, and policymakers to the same table until something got built.

Not one product - a method that keeps producing them

SIP runs a loop: spot the bottleneck, develop a thesis, build and validate a solution, then scale it into a company that can stand on its own. Two of those companies are now doing real work.

The numbers behind the flexibility argument

Flexibility sounds nice. The case for it lives in the gap between how fast a Verrus facility can shed load and how rarely it actually needs to. Below, a few of SIP's headline figures, drawn to scale.

The backers tell their own story. Alphabet provides the technical vantage point and the founding DNA. The Ontario Teachers' Pension Plan and the StepStone Group bring the kind of capital that can wait years without flinching - exactly the patience infrastructure demands. As Winer put it, this is "multi-year not multi-decade" work: not a quick flip, not a science fair.

Unlock infrastructure's potential

SIP's stated aim is to unlock infrastructure's potential by building businesses that supply the energy and compute an AI-driven economy needs. Stripped of the polish: make the grid and the data center flexible enough to keep up with the thing everyone keeps plugging into them.

It is a mission that resists the usual startup vocabulary. There is no viral moment, no overnight scale. There is a thermostat nudging itself at 4 p.m. so a neighbor's lights stay on, and a data center quietly leaning on its batteries while the grid exhales. The work is invisible by design - which is, of course, the surest sign infrastructure is doing its job.

Demand isn't slowing down. The grid has to bend.

The arithmetic is unforgiving. Compute demand climbs, electrification spreads, and the easy answer - build another plant, string another line - gets slower and pricier every year. The clever answer is to make what already exists more responsive. If SIP's bet holds, the next decade of infrastructure is less about pouring concrete and more about coordination: millions of small, flexible moves adding up to capacity that used to require a smokestack.

Skeptics will note that Verrus's first sites are still coming online, and that "flexible infrastructure" has been promised before. Fair. The proof will be in the kilowatts. But the direction is set, and the capital is patient.