Vertical Farm Post-Mortems: What Killed Them

In three years, nearly every well-funded vertical farm in the United States went through bankruptcy court or shut its doors. Companies that raised hundreds of millions of dollars ran out of money. The most interesting part of the story is what saved the biggest one: AeroFarms came out of Chapter 11 by narrowing down to microgreens. Here is the breakdown company by company, using their own filings, and what actually transfers to a shelf holding twenty trays.

A note on sources first. The numbers for AppHarvest, Kalera, and Local Bounti come from filings with the Securities and Exchange Commission: forms 8-K, 10-K, and 10-Q. Executives sign those personally. Where a company is private and files nothing, I use its own press release. Where even that is missing, I name the trade publication in the text.

Who went down and when

One clarification so nobody gets misled. AppHarvest grew tomatoes in high-tech greenhouses, not on racks under lights. I keep it in this list because its economics broke along the same logic, and because its filings are the most detailed of the group. It was never a vertical farm. How a closed room under lights differs from a greenhouse in the logic of climate control is explained in the wiki: grow room versus greenhouse.

A dollar of tomatoes cost $3.91 to grow

This is the central number of the whole article. Here is AppHarvest over three years, in millions of dollars.

Look at the two middle rows for 2022. Sales of $14.59 million against cost of goods sold of $56.99 million. Divide one by the other and you get 3.91. Every dollar the company collected for tomatoes cost it almost four dollars to produce, before office payroll, interest, or marketing.

There is no exotic finance here. It is the same arithmetic you run on a tray: what it cost to grow against what somebody paid. When the first number is bigger, more volume digs the hole deeper. AppHarvest was building new greenhouses at exactly the point where every pound sold already lost money.

For the same line calculated at your scale, see the full cost breakdown of one 1020 tray, from substrate through labor.

The ending was mundane. On July 19, 2023, the company borrowed $2.69 million at 10 percent, due July 24. A five-day loan. On July 23, one day before it came due, AppHarvest filed for Chapter 11 in the Southern District of Texas. The next day Nasdaq notified it of delisting.

One customer for all of the revenue

The 2022 annual report contains a sentence worth reading twice. Verbatim: “Substantially all of our net sales in 2022 and 2021 were generated from the sale of tomatoes… under an agreement with one customer, Mastronardi.” Every dollar of revenue moved through a single buyer.

The risk factors in that same report carry the other half of the deal. AppHarvest had agreed not to compete with Mastronardi outside Kentucky and West Virginia. Selling to anyone else across the rest of the country was off the table by contract.

When there is one buyer, that buyer names the price. A five percent cut is not negotiated, it is announced. At your scale this is the grower whose entire harvest goes to one restaurant or one store. Spreading that risk across accounts is the subject of a separate piece on selling to restaurants.

A $59 million robot, written off in nine months

On April 7, 2021, AppHarvest bought Root AI, a robotic harvesting startup. The total consideration in the footnotes is $59.108 million, of which $49 million was paid in stock. It bought at the top, three months after going public.

By December 31 of the same year the entire purchase was written down to zero. The filing calls it “a non-cash charge of $59.7 million… to fully impair the carrying value of goodwill and definite lived intangible assets included in our acquisition of Root AI.” Nine months from purchase to full write-off.

A year later, in 2022, the company wrote down another $50.1 million from the book value of the greenhouses and the equipment inside them. It conceded that what it had built was worth less than what it had spent.

The lesson is plain and it stings. Equipment does not fix agronomy or economics. A robot that picks tomatoes changes nothing when a pound of tomatoes costs more than it sells for. At small scale this shows up as a $2,000 rack bought by someone who has never sold a tray.

What investors were promised: the AeroFarms forecast in an SEC filing

The most revealing document in this whole story is public. In 2021 AeroFarms planned to go public through a SPAC, and the proxy statement included management’s own financial forecast, prepared in late 2020. The company appears in that document under the code name Dream Holdings.

The document states the growth rate outright: a revenue compound annual growth rate of 168 percent over five years. Now set that against what happened. For 2023 the forecast promised seven farms and $54 million in revenue. In June 2023 AeroFarms filed for Chapter 11 in Delaware.

Notice the shape of the curve. Three years of losses, then a sharp inflection into profit. That is the shape of nearly every business plan handed out at a microgreens workshop. The inflection always sits a year or two past the point where the money runs out.

Revenue doubles, and it fixes nothing

Local Bounti is interesting because it is still alive. The company still files with the SEC, which makes its numbers more instructive than any bankruptcy. Four years, in millions of dollars.

Revenue grew two and a half times in four years. That is good work by a sales team. Now look at the gross profit line: it holds near 12 percent of sales and refuses to move. And look at the debt line, up fourfold over the same stretch.

The first half of 2026 makes it sharpest. Gross profit for six months was $2.58 million. Interest expense alone for those same six months was $8.52 million. Every head of lettuce sold in half a year does not cover the interest on the debt.

Accumulated deficit as of mid-2026 stands at $550 million, and stockholders’ equity is negative $197 million. In August 2026 the company raised another $12.5 million through a convertible note. The terms of that note and the November 4 vote on 58 percent dilution are covered separately: the $12.5 million loan from the controlling owner.

If one line from this article is worth keeping, it is this one. Gross margin does not heal with volume. When a pound of greens returns 12 percent, doubling sales doubles the labor, the working capital, and the risk. The hole stays exactly where it was.

The disease that closed a $2.3 billion farm

Bowery Farming was the second-largest vertical farm in the country. It raised over $700 million, was valued at $2.3 billion, and counted Justin Timberlake and Natalie Portman among its investors. In early 2022 it took on another $150 million in debt from KKR to fund expansion.

On November 3, 2024, staff received an email announcing the closure of every site. WARN notices covered 187 workers: 104 in Bethlehem, Pennsylvania, and 83 in Nottingham, Maryland. The New Jersey farm and the New York headquarters closed as well.

Beyond weak demand for premium greens, trade coverage points to a plant disease that wiped out yields across several farms at once. The sources I checked never name the pathogen, so I am not going to invent one.

The point holds either way. A closed system, sold in every pitch deck as the core advantage, works in both directions. Once a pathogen is inside, it travels the racks at the speed of the irrigation water. A company with a billion-dollar valuation and a staff of agronomists could not get ahead of it. What that looks like with the best-known waterborne pathogen, whose zoospores swim from root to root through a shared reservoir, is covered in the wiki on Pythium in hydroponic systems.

Electricity: why Infarm left Europe

Infarm was the European star of the sector: a Berlin startup putting growing modules directly inside supermarkets, with over $500 million raised. In November 2022 it laid off roughly 500 people, more than half its workforce, and began winding down European operations. The Berlin headquarters closed.

The stated reason was energy prices. That is the closest parallel to a small grower in this whole article. A vertical farm sells converted electricity, and when the rate jumps, the product gets more expensive overnight while the shelf price sits still.

The same trap has a smaller version. A grower running trays in a spare room pays the residential rate. The day the operation becomes a business, commercial rates and demand charges are a different number entirely. That switch, and what it does to a tray, is worked through in the piece on startup budgets.

What survived: microgreens and strawberries

Here is the part worth the most attention. Two companies on this list went through court and are still operating. Both did the same thing: they narrowed down to a single crop with a high price per pound.

AeroFarms filed on June 8, 2023, secured $10 million in financing from existing investors, and came out that September. Its own press release named what it was concentrating on: the farm in Danville, Virginia, and microgreens. The exact wording is “AeroFarms microgreens have become the dominant market leader at retail.”

Plenty filed on March 23, 2025, and emerged on May 29. Out of everything it had been doing, strawberries at the Richmond farm are what remained. The press release puts it as “we are laser focused on strawberries.”

The logic is identical in both cases. Field lettuce sells for pennies, and a vertical farm will never beat it on price. How far above pennies microgreens actually sell, and to whom, is covered in the per-ounce price benchmarks. Microgreens and strawberries sell for many times more per pound, and at the same cost for light and labor they can carry the cost structure. Why a multi-tier rack pays off mainly with low-light crops, and where the lower tiers lose their light, is explained in the wiki on vertical growing systems.

For a reader of this site the conclusion lands somewhere unexpected. An industry that burned billions, climbing out of the hole, arrived at the exact crop you start with on a shelf in a spare room.

What does not transfer to your shelf

Before drawing conclusions, name the differences. Otherwise the takeaway becomes “growing greens indoors cannot work,” and that is wrong.

These companies were killed by capital structure, not by growing. Local Bounti carries $483 million in debt and interest payments that swallow the entire gross margin. Bowery took $150 million from KKR for expansion and had to show that expansion. AeroFarms promised investors sixteen farms in six years, which meant building on a schedule rather than on demand.

You have none of that. A shelf of twenty trays has no debt, no quarterly report, and no promise of 168 percent annual growth. You can stop at ten trays and run them for years, and nobody gets to call that a failure.

The second difference is logistics. Vertical farms were built to supply grocery chains, where product travels hundreds of miles and lives in cold storage. You hand over a tray cut that morning, or living in the tray. That removes refrigeration, freight, and shrink from the math, which are precisely the lines the big operations burned on.

What does transfer is the arithmetic. It works the same for a $200 million farm and for a rack on a porch.

Seven lessons for a grower with twenty trays

1. Gross margin gets calculated first. What it cost to grow against what somebody paid. While that number is negative, every new rack enlarges the loss.
2. Volume does not cure margin. Local Bounti doubled revenue and kept the same 12 percent.
3. One buyer sets the price. AppHarvest’s contract with Mastronardi barred it from selling to anyone else across most of the country.
4. Equipment does not fix economics. A $59 million robot was written off in nine months, because the problem was the price per pound.
5. A closed loop moves disease as fast as it moves water. Bowery closed with a billion-dollar valuation and a staff of agronomists.
6. Power rates jump while the shelf price sits still. That is what pushed Infarm out of Europe.
7. The crop has to carry the cost structure. Both survivors dropped lettuce and moved to microgreens and strawberries.

Five things to check in your own numbers

Closing their company after ten years, the founders of Upward Farms wrote: “We found that vertical farming is almost infinitely complex - as we tackled challenges, new ones emerged.” They walked away from a 250,000 square foot facility while it was still under construction.

Your advantage over them is that a mistake on twenty trays costs twenty dollars, not two hundred million. Use it. Run these five numbers before buying another rack. The loudest name on this list reached the end of its story: AeroFarms was sold to a family office on undisclosed terms after a $1.2 billion valuation.

• Gross result on one tray last month: revenue minus seed, substrate, clamshell, label, water, and power. If it is negative, volume is the wrong lever.
• The share of revenue coming from your largest account. Past half, that is a dependency rather than a channel.
• The rate you actually pay per kilowatt-hour, and what that comes to per tray.
• What happens to cash flow if one batch goes to compost because of mold. That was the last straw at Bowery.
• Whether the crop you grow most of carries the highest margin, or only the highest yield. Those are different questions.

The first item, worked out line by line at U.S. prices, is in the tray cost breakdown and in the review of what a small operation actually earns. The scale is different from anything above. The arithmetic that broke those companies is the same one. Measured across 176 surveyed farms, that arithmetic becomes nine reasons microgreens businesses fail. What remains of the category after those failures is counted in the first federal microgreens census.