After AppHarvest: What Vertical Farming Really Needs to Work

This article is a follow-up to Vertical Farming in 2026: How AI Made It Profitable, which covered the $7.5B market, the companies that survived the 2024–25 shakeout, and how AI climate control changed the economics. Here, we dig into the persistent drawbacks, the specific AI applications doing real work in 2026, and what the path to $25–40B actually requires.

Why does vertical farming still cost 10–20x more energy than field farming?

Vertical farms consume 10 to 20 times the energy of conventional fields, driven by continuous LED lighting and HVAC systems that never stop running.

In an open field, the sun provides all the light and the atmosphere manages most of the climate work. In a vertical farm, every photon hitting a leaf comes from an artificial source, and every degree of temperature is actively managed by HVAC equipment running around the clock. LED efficiency has improved dramatically over the past decade—but powering 50,000 square feet of grow lights for 18–22 hours per day, every day of the year, still consumes significant electricity.

A 2023 analysis published in Nature Food found that vertical farms emit 4 to 10 times more greenhouse gases per kilogram of produce than field-grown equivalents, depending on the energy grid they draw from. Where farms operate on renewable energy—as several Nordic operations now do—that number drops substantially. Where they draw on coal-heavy grids, the environmental math inverts.

AI optimization systems from Plenty and Bowery have reduced energy cost per pound by 30–40% compared to 2022 designs. That's real progress. But it narrows a 10–20x gap to an 8–15x gap, not to parity. The arithmetic still doesn't work for low-margin produce. The companies that survived the shakeout did so by choosing crops where that gap could be absorbed by premium pricing—not by eliminating it.

Why can't vertical farms grow wheat, corn, or soybeans profitably?

Vertical farming works for crops where premium pricing or freshness justifies higher costs. Commodity staples like wheat and corn generate too little revenue per square foot.

The economics of vertical farming depend on revenue per square foot of grow space. Leafy greens, herbs, and premium berries cycle quickly—lettuce turns in 35–45 days—and command prices high enough to cover energy and capital costs. A kilogram of premium basil might sell wholesale for $20–$30. A kilogram of wheat sells for under $0.50.

Staple crops also require far more physical space proportional to their caloric value. Wheat and corn are tall plants with long growth cycles and modest yield value per square meter. In a vertical environment, that means more grow levels, more lighting, more energy, more capital—and revenue that doesn't come close to covering it. The World Resources Institute has noted that vertical farming's food security contribution is limited to specialty and high-value crops; for caloric staples, field agriculture and improved logistics remain the more resource-efficient option.

This is a structural constraint, not a technology gap. No AI optimization system can change the caloric density of wheat or the price ceiling on soybeans. It's why every profitable vertical farm in 2026 grows some combination of greens, herbs, microgreens, or berries—and why expansion into staples remains a long-term aspiration rather than a near-term operating reality.

What went wrong with AppHarvest and the other companies that failed?

AppHarvest raised $475M and filed for Chapter 7 bankruptcy in 2023. Kalera and Local Bounti restructured. All three scaled infrastructure before solving unit economics at the crop level.

AppHarvest is the clearest case study. The company went public via SPAC in 2021, raised hundreds of millions, and built facilities targeting tomato production at scale. Tomatoes are a borderline crop for vertical farming—higher value than staples but not high enough to absorb the energy budgets that AppHarvest's facilities required at launch. When input costs rose and retail pricing pressure increased, the margin cushion that never fully existed disappeared.

Kalera, a Dubai-founded lettuce grower with US operations, faced similar dynamics: real product quality, insufficient unit economics, and a capital structure built on optimistic growth projections. Local Bounti's greenhouse-vertical hybrid model showed more resilience but still required restructuring when targets slipped.

The common thread isn't the product or the technology. It's sequencing. These companies scaled infrastructure before achieving profitable operations on a per-unit basis. Facilities came online before AI optimization systems were capable enough to justify the energy cost at the volumes being targeted. Survivors like Bowery and Plenty took a different path: prove the unit economics work on small crop runs first, deploy AI to reduce per-pound cost, then expand. That sequence matters in any capital-intensive business.

How does vertical farm produce compare to field-grown nutritionally?

Studies show vertical farm produce matches field-grown on most nutrients. The practical advantage is freshness and calendar consistency, not nutritional superiority.

Research published in Frontiers in Plant Science found that hydroponic and aeroponic systems can produce leafy greens and herbs with comparable or superior vitamin C, beta-carotene, and antioxidant profiles compared to field-grown equivalents—largely because controlled light spectrum and nutrient delivery can be tuned to maximize specific compounds. LED spectrum in particular can be dialed to conditions that stimulate the plant's natural stress responses, boosting protective compounds like anthocyanins.

The practical advantage at the consumer end isn't nutrition per se—it's freshness at the point of consumption. Field-grown produce may travel 1,000–2,000 miles and spend 7–14 days in cold chain before reaching store shelves. Vertical farm produce grown within 25 miles of a store can arrive within 24–48 hours of harvest. Nutrient degradation is time-dependent. A fresh vertical farm head of lettuce and a nutritionally optimized but 10-day-old field-grown head are not the same product—and increasingly, consumers can taste the difference.

How is AI cutting vertical farm costs and improving yields in 2026?

AI-managed grow environments now predict 20% higher yields, detect pest and disease pressure early with computer vision, and reduce labor requirements by roughly 50% through automation.

Three AI applications are producing the most measurable impact: predictive yield analytics, computer vision for disease detection, and warehouse-style automation for seeding, transplanting, and harvest. Each targets a different cost line in the operating budget.

Predictive analytics—trained on sensor data from millions of grow cycles—now produce yield forecasts accurate enough that farms can optimize nutrient delivery, light schedules, and harvest timing to achieve roughly 20% higher yield per square meter compared to manual scheduling. Better yield from the same infrastructure directly improves unit economics without adding capital or energy cost.

Computer vision systems scan crop canopies for early signs of disease and pest activity, detecting problems 48–72 hours before they become visible to the human eye. Catching issues early reduces crop loss from an industry average of 12–15% toward 3–5%. On a high-turnover crop like basil or lettuce, a 10% reduction in waste changes the margin profile at scale.

Automation is the most significant labor cost lever available. Seeding, transplanting, and harvesting in vertical farms are physically demanding, repetitive tasks. Robotic systems now handle 60–80% of those operations at facilities in Scandinavia and select US operations. The result is roughly 50% fewer labor hours per pound of produce, with more consistent product quality than manual handling allows.

Which companies are leading AI deployment in vertical farms globally?

Plenty and Bowery anchor North American AI leadership. Nordic Harvest is the European robotics benchmark. YesHealth and Planet Farms are demonstrating regional models that others will follow.

Nordic Harvest's Copenhagen facility—one of Europe's largest fully automated vertical farms—adapted warehouse logistics robotics to the grow environment rather than building purpose-built ag systems from scratch. That approach brought proven supply chain technology into agriculture at lower development cost and faster deployment. It's become a reference model for operators looking to scale without proportionally scaling headcount.

Planet Farms in Italy has paired AI climate optimization with on-site photovoltaic generation and time-of-day scheduling—running energy-intensive lighting during peak solar hours and shifting to lower-intensity periods when grid draw would peak. By integrating renewable generation with AI-managed grow cycles, they've achieved carbon-neutral operations and qualified for EU farm-to-fork subsidy support. This renewable-plus-AI integration template is the architecture that makes the environmental case for vertical farming credible at scale.

Where is the vertical farming market headed by 2030 and 2032?

Market projections place vertical farming between $25B and $40B by 2032, with Asia and Europe as the fastest-growing regions, driven by urban density and food security policy respectively.

Grand View Research projects the global vertical farming market at $25.4B by 2030, with continued growth through 2032. The upper range of $40B assumes AI and renewable integration cuts operating costs enough to expand the viable crop range beyond current greens and herbs—opening peppers, tomatoes, and small fruits to commercial vertical production at scale.

Hydroponics currently accounts for approximately 53% of market share within vertical farming, driven by its relative simplicity versus aeroponic systems and its proven performance with leafy crops. Container farms—modular, portable units that can be paired with on-site renewable generation—are emerging as the fastest-growing format within the sector, with cost structures lower than purpose-built facilities.

Asia is the fastest-growing region by volume. YesHealth in Taiwan has demonstrated what a scaled, profitable vertical farming network looks like in a high-density urban market where freshness and food safety concerns drive significant willingness to pay. South Korean and Japanese operators have expanded rapidly with government backing. Europe's growth is policy-driven—sustainability mandates and food security investment after the 2022 supply disruptions created funding and procurement commitments that didn't exist five years ago.

What needs to change for the 2032 projections to hold?

Cheaper energy access, a broader profitable crop range beyond greens and berries, and policy subsidies in major markets are the three variables the industry can't optimize its way around.

Energy is the most important variable. Every scenario that reaches $30B+ assumes meaningful further reductions in energy cost per pound—either through AI optimization pushing current efficiency further, through renewables integration bringing effective energy price down, or both. Container farms paired with solar or wind generation are the most credible near-term model: several pilot projects in the US Midwest and Northern Europe have demonstrated energy cost structures competitive with grid-powered traditional greenhouses. At scale, the modular container format could rapidly expand vertical farming's geographic footprint beyond urban cores where land cost justifies premium real estate.

Policy subsidies are the second lever. The Association for Vertical Farming has identified public procurement—particularly school meal programs and institutional purchasing mandates—as the most direct mechanism for creating stable demand at prices vertical farms can survive on. Political interest in domestic food security has created openings that didn't exist in 2019.

The boldest prediction worth naming directly: AI agent orchestration could manage 50% of vertical farm operations by 2030. Agentic systems that route nutrient delivery, schedule harvests, coordinate logistics, and adjust grow parameters without human intervention are closer to deployable than they were two years ago. The same autonomous operations logic emerging in software engineering—AI systems that manage pipelines end-to-end—is being trialed in controlled agricultural environments where feedback loops are tight and sensor data is clean. Whether that reaches 50% of all vertical farm ops by 2030 is an ambitious projection. The direction is not in doubt.