2025年世界CEAセンサスが示す屋内農業の未来像

The Census at a Glance

The Agritecture/CEAg World Global CEA Census is the closest thing the indoor agriculture industry has to a comprehensive self-assessment. The 2025 edition collected responses from 316 operations across 54 countries, covering everything from economic viability and technology adoption to workforce development and business model innovation. It is not a press release or a pitch deck. It is farm operators reporting on their actual conditions, and that makes it one of the most valuable data sources available for understanding where this industry genuinely stands.

The headline narrative that emerges from the data is one of maturation. The indoor farming industry is transitioning from its first phase—characterized by ambitious technology demonstrations, venture-scale fundraising, and rapid expansion—into a second phase defined by operational discipline, unit economics scrutiny, and careful scaling. The era of “move fast and break things” in indoor agriculture is over. What replaces it is the harder, less glamorous work of proving that controlled-environment production can be a sustainable business, not just an impressive technology.

Profitability: The Gap Is Narrowing

The census findings on economic viability confirm what anyone paying close attention already suspected: most indoor farming operations are still not profitable, but the distance between current cost structures and break-even is shrinking.

The operators closing that gap share a few common characteristics. They have moved beyond single-channel revenue dependency—selling produce at retail—toward diversified income streams that leverage the assets and expertise they have already built. They are disciplined about matching production capacity to contracted demand rather than growing speculatively. And they are making pricing decisions based on actual cost of production rather than on competitor pricing or aspirational margin targets.

The most important shift in the profitability conversation is that the industry is now talking about unit economics at all. Three years ago, the dominant metric was scale—how many square feet, how many tons of production capacity, how many millions raised. The census data reflects a community that has internalized the painful lesson that scale without margins is just a faster way to run out of money. The ERP Gap in Indoor Farming: Why Most Farms Are Still Running on Spreadsheets

AI Adoption: Early Stages, High Expectations

The census data on AI adoption reveals a gap between aspiration and reality that is worth examining carefully, because it has practical implications for how operators should allocate technology budgets.

Most farms are in the early stages of AI integration—using sensors and cameras for monitoring and data collection, but not yet deploying predictive models that optimize climate control, irrigation, or harvest scheduling in real time. The reasons are practical rather than philosophical. Integration complexity is the most frequently cited barrier: most indoor farms run a patchwork of systems from different vendors, and making those systems talk to each other—let alone feeding their data into AI models—requires integration work that many operations lack the engineering resources to undertake.

Cost is the second barrier. AI solutions marketed to indoor farming operations often require significant upfront investment and ongoing subscription fees that are difficult to justify when the farm itself is not yet profitable. Operators are understandably skeptical of technology vendors promising AI-driven yield improvements when the more fundamental question of whether the business model works has not been answered.

The census suggests that the farms making the most effective use of AI are those that approach it incrementally—starting with data collection and monitoring, building historical datasets, and then layering in predictive capabilities as the data foundation matures. The farms that skip directly to “AI-optimized everything” without first establishing reliable data pipelines tend to generate expensive noise rather than actionable insight.

Business Model Innovation: Beyond Selling Produce

One of the most interesting findings in the census is the emergence of revenue models that extend beyond the traditional produce-sales paradigm. The operators demonstrating the most creative thinking about sustainability are the ones asking a different question: what else can this facility and this expertise generate?

Data services are an early but growing revenue category. Farms that have built robust environmental monitoring and crop performance datasets are finding that their data has value to seed companies, agricultural researchers, and other growers. Contract growing arrangements—where a farm produces specific crops under agreement for restaurant groups, institutional food services, or specialty retailers—provide revenue predictability that speculative retail sales cannot. Technology licensing allows operators who have developed proprietary systems or processes to monetize their intellectual property without the capital requirements of additional facilities.

Consulting services represent another emerging revenue stream. Operators who have navigated the learning curve of building and running an indoor farm possess hard-won expertise that newer entrants will pay for. The irony is not lost on the industry: some of the most valuable knowledge in indoor farming was generated by the mistakes and near-failures that almost destroyed the companies now in a position to sell that knowledge.

This diversification trend signals an industry that is becoming more pragmatic about how value is created and captured. The farms that survive long term will likely look more like technology and services companies that happen to grow food than like traditional agricultural operations that happen to use technology.

The Workforce Challenge

The census confirms what hiring managers in indoor farming have been saying anecdotally for years: the workforce gap is real, it is structural, and it is not going away on its own.

The industry needs people who can bridge agricultural science and technology—individuals comfortable with both plant physiology and software systems, with both nutrient chemistry and data analysis. That combination of skills is rare because the educational pathways that produce it barely exist. Traditional agriculture programs do not teach software programming. Computer science programs do not teach crop science. The result is a persistent talent gap that forces operators to either hire from one domain and train in the other, or compete intensely for the small pool of candidates who have somehow assembled the full skill set independently.

The census data suggests that software programming is now considered a core workforce need alongside plant science and operations management—a shift that would have been unthinkable in agriculture a decade ago. As farms increasingly depend on integrated software platforms for everything from climate control to inventory management to financial reporting, the ability to configure, troubleshoot, and extend those systems becomes as important as the ability to diagnose a nutrient deficiency or optimize a lighting recipe.

The Software Gap

Threading through several of the census findings is an implicit observation about software infrastructure: most indoor farms are running on fragmented, inadequate technology stacks that create operational friction and limit the industry’s ability to scale.

The AI adoption barriers—integration complexity, data silos, lack of standardized interfaces—are fundamentally software problems. The workforce challenge—needing programmers alongside agronomists—is a symptom of software systems that require too much custom engineering to operate. The business model limitations—difficulty tracking unit economics, inability to benchmark performance—reflect the absence of integrated farm management platforms that connect environmental data, production metrics, and financial outcomes in a single system.

The industry’s shift toward integrated digital cultivation platforms—systems that unify climate control, crop monitoring, resource tracking, and business analytics—is one of the census’s most actionable findings. The farms that close the profitability gap first will disproportionately be the ones that solve the software problem first, because software is what connects the operational decisions (how much light, how much water, when to harvest) to the financial outcomes (cost per kilogram, margin per crop, return on facility investment) that determine whether the operation survives.

What Comes Next

The 2025 Census paints a picture of an industry at an inflection point. The first wave of indoor farming was about proving that the technology worked. It did. The second wave was about scaling that technology to commercial production. That happened, but the economics did not always follow. The third wave—the one the census data suggests is now underway—is about proving the business model. CEA Industry Mid-Year Report: Consolidation, AI, and What’s Next

The operators who will define this third wave are not necessarily the ones with the most advanced technology or the largest facilities. They are the ones with the clearest understanding of their unit economics, the most disciplined approach to matching production with demand, and the most integrated software systems for connecting operational data to financial outcomes. They are the ones who have internalized the census’s central lesson: that indoor farming’s future depends less on growing better plants and more on building better businesses.

Three hundred sixteen respondents across 54 countries is not the entire industry, but it is a large enough sample to be instructive. The data tells a story of hard-won maturity—of an industry that tried to grow too fast, got burned, and is now rebuilding with clearer eyes and sharper discipline. That is not a story of failure. It is a story of an industry that is finally doing the difficult work that sustainable growth requires.