The commercial model

29 October 2026 · proposed operating model · all Anyground prices and costs below are illustrative assumptions, not quotations

1. Sell a complete bounded result

The long-term business is a coherent catalog and accountable integration/service system for useful places. The launch business should not try to manufacture every subsystem or sell universal independence. It should deliver one defined capability whose value, operating obligations and support can be measured.

The recommended initial hypothesis is a professionally supported growing configuration on existing, known water and electricity, using a qualified enclosure and an experienced grower. The core product boundary is integration, repeatable handling/service interfaces, metering, local controls, a crop workflow and an explicit Care plan. Customers buy an installed scope and understandable work, not simply a box with nominal plant capacity.

A Care plan means duties are assigned; it need not mean a mandatory paid monitoring subscription. The customer or a trained partner may perform appropriate work. Essential local controls and export of owned records should remain available without a subscription. Managed agronomy, monitoring, scheduled maintenance and additional field response can be separately priced when they deliver real value.

No saleable Anyground hardware, supplier agreement, installation contract, paid pilot or customer acceptance is established by this document. The public website should invite understanding and research interest without implying orders are open. A working website is not evidence of a working farm.

2. Choose one buyer and one territory

The first service territory should be small enough that travel, training, spare stock and founder learning remain observable. “Anywhere” is the long-term platform direction, not a launch coverage map. Define who authorizes payment, who operates the equipment, who consumes output, who receives alerts and who can authorize a shutdown.

The first three are investigation candidates, not a commitment to launch in all three markets. Obtain at least two independent budget-holder commitments to a paid, explicitly experimental scope before interpreting demand as more than interest. Compare a narrower module with the whole-farm offer. If the module wins, do not inflate its scope to maintain a dramatic story.

Bottom-up market estimation should count eligible reachable sites, their relevant budget, likely conversion and available service capacity. Do not add global housing, agriculture, water and energy markets into a synthetic total. A huge theoretical market says little about whether the first ten installations can be supported profitably.

3. What belongs in an order

A future commercial proposal should separate:

• site assessment and engineering/configuration work;
• owned hardware and identified partner equipment;
• existing infrastructure accepted as a dependency;
• freight, unloading, lifting and access responsibility;
• site preparation, permits and specialist installation;
• connection, commissioning, training and acceptance;
• initial seeds/starts, nutrients, cleaning supplies and spare parts;
• ordinary customer work and optional managed work;
• support coverage, response scope, exclusions and escalation;
• warranty terms, replacement compatibility and decommissioning.

A paid assessment may be appropriate when configuration work is substantial. Credit it toward an order only if the economics support that promise. A pilot discount must be explicit and temporary; otherwise a subsidized installation becomes misleading evidence of repeatable pricing.

Distinguish the product's invoice from the customer's complete installed cost. The customer must also understand working capital, consumables, energy tariffs, demand peaks, labor, spoilage and maintenance. “No subscription” does not mean no operating cost. “Farm as a service” does not eliminate the equipment's capital requirement; it moves who finances it and who carries the residual-value risk.

4. A commercial reference, not an Anyground quote

Freight Farms' official FAQ currently lists a $215,000 starting purchase reference and a $3,000/month service route, and estimates 20–30 hours of labor per week for one container farm. The same account notes that shipping, training and configuration affect startup scope. These are vendor statements, checked 29 October 2026, not independent performance evidence or a quote for an Anyground configuration. Freight Farms FAQ

This reference is useful because it makes capital and ongoing human work visible. It does not establish what a sunlight greenhouse should cost, what crops can be sold in a particular town, or how a broader habitat should be priced. Never combine a vendor's maximum output and minimum consumption as if they were a matched guaranteed operating point.

5. Illustrative crop-cost stress test

This intentionally simple model retains the source report's benchmark so the public explanation can show the scale of the problem. It is not a proposed Anyground retail price, a bankable forecast, a supplier quotation, a food-safety recommendation or a design for a complete habitat.

Capital recovery uses:

annual capital = C × r × (1+r)^n / ((1+r)^n − 1)

With C = 250000, r = 0.08, n = 10, annual capital recovery is $37,257.37. Electricity is $10,950, labor $32,500, and other costs $12,400. Total modeled annual cost is $93,107.37, or $23.28 per saleable kg.

These are arithmetic sensitivities, not agronomically matched scenarios. The model excludes land rent, packaging, distribution, selling cost, taxes, corporate overhead, financing fees, profit and costs beyond the assumed maintenance reserve. It also excludes dwelling space, off-grid generation/storage, atmospheric-water equipment and original robotics development. Electricity tariffs can include demand charges; local support can include travel that this simple model misses.

Reducing human work from 25 to ten hours/week saves a modeled $19,500/year and lowers modeled cost to $18.40/kg before additional robot costs. The robot's capital recovery, software, servicing, consumables, supervision, safety arrangements and lost production must be added before calling this an economic saving. Time saved only has cash value if staffing cost actually falls or the released time produces valuable work; do not equate every saved minute with payroll reduction.

Do not double count food value. The same kilogram cannot generate both sale revenue and avoided procurement for one customer unless the accounting explicitly separates different transactions. Freshness, visitor experience and resilience can be real benefits, but price them separately rather than use them to imply cheap food.

6. Illustrative service-margin test

A platform that needs a concealed remote workforce may still be a service business, but it must charge for that work. Consider a hypothetical annual support contract at $6,000, with all inputs below invented solely for sensitivity:

If remote work doubles to 48 hours, the same model leaves $20, approximately 0.3%, before overhead. One additional $1,000 visit turns that into a loss. A nominal recurring revenue stream is not a moat if its obligations grow with every installed unit.

No $6,000 service price is recommended by this example. Real pricing needs fault frequency, travel routes, parts history, response commitments, agronomic duties and correlated supplier failures. A reserve estimated from independent failures can badly understate the risk of one defective component batch affecting many sites at once. Keep warranty exposure separate from customer-paid preventive service.

7. Platform and manufacturing economics

The proposed economics should improve through controlled repetition, not the mere existence of a broad catalog. Measure:

• shared BOM value rather than the percentage of identical-looking panels;
• assembly/test labor and first-pass yield;
• SKU count, variant stock and obsolete inventory;
• engineering hours and specialist visits per installation;
• warranty and recovery cost by configuration;
• refurbishment yield and actual second-life value;
• spare availability and repair lead time in the promised territory;
• installed and service contribution after channel margin.

Early production should use qualified suppliers and a small assembly/test operation. Heavy investment in a large factory before order, permitting and site-readiness evidence risks financing inventory that cannot become an accepted installation. Regional fabrication can reduce bulky freight, but requires controlled drawings, tolerances, training, inspection and supplier-change management. Digital files alone are not quality assurance.

Buy established climate, power, treatment and sensing equipment when it fits. Own the interface and acceptance work that otherwise falls between suppliers. Maintain approved alternatives for critical devices, but do not call an apparent drop-in substitute compatible until its electrical, chemical, firmware and service effects are tested. An inexpensive supplier change can create an expensive fleet problem.

An open interface is compatible with a defensible business. Customers may prefer Anyground because qualified combinations, repair records, understandable choices and accountable support save work. Restricting essential functions behind proprietary consumables or a cloud subscription would weaken the ownership promise and could make remote applications less dependable.

8. Cash, leases and scale

Initial customer ownership or an established financing partner is preferable to self-financing a leased fleet before asset life and service cost are known. A lease requires cash for hardware before the stream of receipts arrives, plus provisions for default, retrieval, repair, storage and remarketing. Attractive monthly pricing can conceal a large balance-sheet obligation.

A funding plan should separate reusable product development, customer-specific engineering, working capital and frontier research. Build it from actual team, supplier, workshop, test, qualification, insurance and installation costs in the chosen geography. This document intentionally does not invent a fundraising target or guaranteed return. Research sponsorship can fund a difficult mechanism without proving the core commercial offer has demand.

The prospective platform moat is accumulated, reviewed operational knowledge: which configuration works, which fault matters, how it is recovered and what expansion costs. Combine that with controlled manufacturing and service coverage. A large image library, an AI label or a list of component vendors is not equivalent evidence.

9. Precommitted gates

These are proposed decision rules. Customer and engineering teams must freeze numerical acceptance thresholds before trials; three cycles or sites do not establish lifetime reliability. Positive customer acceptance is not a substitute for technical qualification, and technical qualification is not proof of willingness to pay.

10. First ninety days

Weeks 1–3: observe two candidate buyer types, collect matched records, identify who pays and operates, compare conventional alternatives, obtain water/weather/utility information and document the work that customers actually want removed.

Weeks 4–6: select one job, crop/workflow and territory; assemble a preliminary quote-backed scope with partner responsibilities; time a baseline; prototype the smallest useful Service Spine and a failure-prone replacement task. Agree what evidence would disprove the product hypothesis.

Weeks 7–12: demonstrate local metering, one bounded sequence, branch isolation and independent manual recovery; cost the full installation and support; freeze an experimental pilot agreement and acceptance plan. Do not promise a retail launch date merely because the bench demonstration works.

The commercial north star is not the cheapest-looking cabin or the largest nominal crop capacity. It is a valuable installed capability whose complete cost, human work and recovery behavior remain understandable as it repeats. Only that evidence earns the catalog's expansion toward complete living environments.