The public brief

The product is permission, not produce.

Across the country, data centers are stalling — not on engineering, but on the people who live next door. Intelligent Harvest sells the one thing the build-out can’t buy: a community’s yes. This is the research in the open, no NDA — the short version of the case. The full brief lives behind the gate.

Scroll
01 The bottleneck isn’t the grid

Data centers don’t die in the server hall. They die at the hearing.

A trillion-dollar build-out is colliding with a permission crisis. Public reporting ties roughly $156 billion in U.S. data-center projects to local opposition.1 The honest version: opposition touched those projects — permitting, power availability, and economics all play a role too — but the direction is unambiguous, and consent has become the deciding variable.

What the public wants in return isn’t a tax-revenue slide. It’s a benefit that is local (it stays with the people bearing the cost), visible (you perceive it without taking anyone’s word), and durable (it lasts the life of the facility). Words and money leave nothing behind. A pipe nobody can see changes no one’s mind.

The one thing every operator is short on

Permission is the scarce input in the entire industry. The town wants something it can see from the road and buy at the market — not a banner, not a press release. That is the gap Intelligent Harvest fills, and it is the whole business.

What your town gets
The benefit you can see from the road.
Not a tax slide — fresh food on local shelves and year-round jobs that stay in town. Drag to size it to a project near you.
300,000
lbs of fresh local food a year — strawberries, oranges, mushrooms, turmeric & ginger
20
year-round jobsRooted in place — the recurring local employment a server hall can’t add.
4
crop zonesStrawberries, oranges, gourmet mushrooms, and turmeric & ginger, on one heat loop.
365
days a yearThe waste heat runs around the clock, so the greenhouse grows through winter.

Design target Food and jobs scale a ~2.5-acre, four-zone flagship and are internal design targets, not measured results.

For the non-engineering reader

Two simple ideas do the work here. First, a greenhouse wants gentle warmth — which happens to be the weak, low-grade heat a data center gives off, the kind almost nothing else can use. Second, the industry is already switching to liquid-cooled servers for its own reasons, and that switch makes the heat easier to capture while cutting water use at the same time. We’re building on a change that’s already happening, not asking anyone to change course.

02 How it works

Every data center is already a heat plant.

A data center turns electricity into heat and throws nearly all of it away. Reuse has lagged in the U.S. not because the heat is absent, but because air-cooled systems made it uneconomical to catch — air is a poor heat carrier. Water isn’t: by volume it carries thousands of times more heat. As the largest operators move to direct-to-chip and immersion cooling for their own efficiency, server heat now arrives in a closed liquid loop — hotter, concentrated, and easy to pipe.

That single transition is the one that makes a neighboring greenhouse feasible. We’re not asking the industry to change course; we’re building on the course it’s already on. And the same shift independently cuts the footprint communities object to most — a peer-reviewed life-cycle assessment led by Microsoft researchers found liquid cooling reduces greenhouse-gas emissions by 15–21%, energy demand by 15–20%, and blue-water use by 31–52% versus air cooling.2 Solving the water fight and enabling reuse are, physically, one decision.

The grade of the heat is already enough

The standard objection is that data-center heat is “low-grade” — warm, but cooler than many uses need. That’s exactly why district heating struggles: building networks want 60–70 °C. A greenhouse is the rare customer that wants precisely what a data center most easily gives. Comfortable growing air sits around 18–27 °C, with mild root-zone warmth through low-temperature loops. Water-cooled reject heat (roughly 50–60 °C) clears the greenhouse’s comfort line before any lift at all.3 The low grade that is a deal-breaker for almost everyone else is the whole point here.

Why a greenhouse

Low-grade heat doesn’t travel. The hard part of reuse was never capturing heat — it’s finding an anchor offtaker with continuous, year-round demand, close enough to use the warmth before it’s lost. A greenhouse dissolves the distance problem by sitting on the fence line: zero transmission distance, and an appetite for heat that runs 24 hours a day, every day of the year. The data center rejects far more heat than the greenhouse can use — up to 0.86 MWh of reusable warmth per MWh of IT load4 — so the recovery gear stays small and the project can start now. And it converts an invisible by-product into something a resident can see from the road and taste at the market. It solves the physics and the politics in one move.

03 It already works

This isn’t a hypothesis. It’s running on four continents.

A clever mechanism that’s never been built is a hypothesis. This one is demonstrated. Data-center heat warms greenhouses, homes, pools, and fish farms today — and Europe is roughly a decade ahead, in part because several countries now require reuse.

Boden, Sweden. A 32‑MW data center heats an approximately 90,000‑square‑foot greenhouse, enabling vegetable production near the Arctic Circle.5
Across Europe. Hyperscale and colocation operators feed municipal heat networks in Odense, Mäntsälä, and Kista, with other recovered-heat uses running from pellet-drying to commercial trout and lobster farming.6
The U.S. pipeline is converging on this exact model. A liquid-cooled facility in Marietta, Ohio already heats a local greenhouse; proposed campuses in West Virginia and Ohio would co-locate data centers with controlled-environment agriculture on the same waste heat.7

When multiple independent teams arrive at the same design, that isn’t coincidence — it’s convergence on what the constraints allow. The open question is no longer whether the pattern works. It’s who executes it cleanly, at community scale, with the framing that actually earns permission.

04 The graveyard

A wave of brilliantly funded farms collapsed. We studied every one.

Between 2022 and 2026, controlled-environment farms with enormous capital behind them went bankrupt. We’re not blind to it — we built the model to do the opposite of what killed them.

Plenty
~$940Mraised
SoftBank, Walmart, Bezos. Chapter 11, March 2025.
Bowery Farming
~$700Mraised
Once valued at $2.3B. Shut down late 2024.
AppHarvest
~$700Mraised
SPAC-funded greenhouse tomatoes. Liquidated 2023.
AeroFarms
~$300Mraised
The poster child. Chapter 11 in 2023; survives on microgreens.
Fifth Season · Kalera · a long tail
of shutdowns from 2022 onward.

Figures from public bankruptcy filings and industry reporting, 2022–2026.

Why they died
1
Energy was the killer. Operating costs were dominated by electricity for artificial lighting and climate control. They replaced free sunlight and warmth with a power bill — against produce that couldn’t command a high enough premium to cover it.
2
Produce was the whole business. Thin margins on lettuce and herbs had to carry enormous capital and debt. When demand softened, nothing else held the model up.
3
They scaled before the economics worked. Raise hundreds of millions, build showcase facilities, figure out the farming later — tech company first, farm second.
4
No secured demand. Few locked off-take before building. Capacity arrived ahead of buyers.
05 Why we’re different — by design, not by hope

We inverted every line on that list.

We took the survivors’ discipline — one anchor crop, secured offtake, a right-sized build — and paired it with the one advantage none of them had: heat that costs nothing. Point for point, the model is the opposite of what failed.

What killed them
What we built instead
Energy was their #1 cost.
That cost is our free input — the data center’s waste heat is a liability someone else pays us to take away.
Produce was the whole business.
Produce is our proof, not our product. Revenue is the operator’s permission fee — we’re structurally indifferent to whether the strawberries turn a profit.
They scaled a VC-funded fleet before the economics worked.
One operator-funded flagship. No speculative scaling, no burning investor capital racing ahead of the math.
They were tech companies reinventing pipes and robots.
We buy the engineering and sell the permission — proven greenhouse and heat-recovery gear, procured as a service.
They built before securing demand.
Our buyer is committed by LOI before a pane of glass goes up.
06 What the free heat is worth

The moat the bankrupt farms never had.

A waste-heat greenhouse never buys heating fuel — so the heat’s value isn’t speculative. It equals the fuel bill the greenhouse doesn’t pay. The farms that died were paying for the one input a data center throws away for free. Quantified at flagship scale, that advantage is worth real money — and it lands largest exactly where these projects are built: rural towns that heat with the costliest fuels.

~$75K–$650K
per year — the free-energy advantage at flagship scale, depending on climate and the fuel the host town would otherwise burn.
~$16M
one-time cost of a flagship greenhouse — about 0.01% of the blocked U.S. pipeline it’s designed to help unlock.
~700 t CO₂
per year — carbon avoided versus a gas-heated greenhouse, on the order of 150 cars off the road. The farm’s heating footprint, zeroed out.

Design target And these aren’t “revenue.” The greenhouse is permission insurance, not a farm that must turn a profit — which is exactly what lets it absorb a bad harvest and still cost a rounding error against the facility it protects. Figures are internal Intelligent Harvest projections, to be confirmed by built performance.

07 The risk we take most seriously

If the greenhouse becomes theater, it deserves to fail.

The sharpest objection to a project like this isn’t technical. It’s that a greenhouse becomes a green fig leaf — a photogenic distraction that buys approval while a facility’s real impacts continue. We think that critique is correct about the danger. It’s precisely why the work has to be done one way.

The difference between a benefit and a billboard is measurement

Intelligent Harvest commits to third-party-metered, publicly reported delivered heat and crop output — the same logic the best policy proposals use, tying any incentive to verified energy actually delivered rather than promised. A claim you can audit is not greenwashing. A claim you cannot is. The entire thesis collapses the moment the benefit becomes unverifiable — so we put it on the record, by design.

08 The water architecture

The same loop has to answer for the water, too.

Heat makes the case; water decides whether a town believes it. So we treat water the way we treat heat — with a meter, not an adjective. One water system, four climate configurations, and a single published net-withdrawal number.

The greenhouse doesn’t change how much water the data center itself uses — that’s the operator’s footprint, and the move to liquid cooling is already cutting it 31–52% against air cooling (a Microsoft-led life-cycle assessment). What we own is narrower and checkable: a greenhouse built to recycle most of what it draws, capture rain and condensate, and report the net on a third-party gauge. Not “water-positive” as a pledge — a number you can read.

70–90%
less water than open-field farming, when a sealed greenhouse recaptures its transpired water (Dr. Greenhouse).
4
climate configs on one shared spine — net-zero in Virginia, net-positive in the desert, a stormwater sink on the coast.
1
third-party net-withdrawal meter, published — the water twin of the heat meter.
Why the coast is the opposite problem

On the Virginia coast the water question inverts — surplus, not scarcity. Hampton Roads has among the highest relative sea-level rise on the East Coast, and the land is sinking about twice as fast as the sea is rising. There the greenhouse’s job is to absorb stormwater and never pump the stressed aquifer — aligned with the region’s number-one water priority, and honest about what it can’t do: it doesn’t stop the sea.

Read the full water architecture →

The close

The product is permission. The produce is proof.

A roughly $16 million greenhouse, sized to be trivial against the billions it protects, changes the answer a community gives. It’s a farm. This is America — and America has always bet on the farmer.