Back to Blog
Space EconomyNovember 20269 min read

You Bought the Satellite. You Didn’t Buy the Capability.

Right to repair came for the tractor. The same question is waiting in orbit, where the keys matter more than the deed.

A farmer bought a tractor. He owns it outright, free and clear, and he still cannot fix it. The repair runs through an authorized dealer, the diagnostics sit behind a login, and the software can throw the machine into limp mode, which only a dealer’s tool can clear. Ownership of the iron turned out to be a different thing from control of the machine.

That is the whole right-to-repair argument, and it now has a price on it. On July 8, 2026, the Federal Trade Commission and five states filed a proposed stipulated order with Deere & Company that would require, for ten years and under the supervision of the commission and the plaintiff states, that farmers and independent repair providers get repair resources equivalent to what Deere gives its dealers. Illinois, Arizona, Michigan, Minnesota, and Wisconsin signed on. A separate class action settlement would create a $99 million fund for a class of more than 200,000 farmers, and we could not verify the settlement’s approval status. Read the structure carefully, because it is the part the headlines blur: the order requires conduct and pays nothing, and the fund is the only money in the case.

Now move the same question 500 kilometers up.

The tractor you cannot tow to a shop

Every argument in the Deere case survives the trip to orbit, and one of them gets worse.

A satellite operator typically owns the spacecraft. What the operator does not necessarily own is the ability to change it: the flight software license, the command and control encryption, the ground segment, the interface documentation that tells a second vendor how to talk to the bus. Ownership of the asset is not the same as authority over it.

On the ground, a locked tractor still has a physical escape hatch. It can be towed, hacked, or replaced, and the dealer’s leverage is legal rather than physical. In orbit there is no tow truck. There is no independent shop. The servicing industry exists, but it is still being bought largely by governments rather than by owner-operators, and it is priced as a mission, not as a maintenance call. Which means the only lever that works is command authority, held by whoever controls the keys and the software.

That is not a metaphor. It is the difference between an asset and a hostage.

The Pentagon already published the finding

We do not have to argue this from analogy, because the Government Accountability Office looked at it directly. In September 2025 it reported on data rights and sustainment across five programs: the F/A-18, the F-35, the Littoral Combat Ship, the Stryker Combat Vehicle, and the Virginia-class submarine.

The findings read like a right-to-repair complaint with a government letterhead. None of the programs that had completed an intellectual property strategy included all of the elements the department’s own policy requires. Programs experienced vendor lock, meaning reliance on a single supplier, because of data rights shortfalls. Officials told GAO that this drives up costs and lengthens repair timeframes, and that the options for fixing it shrink once a program enters sustainment. Statute gives the department unlimited rights in operation, maintenance, installation, and training data, but it excludes detailed manufacturing and process data, which is often exactly what a competitor would need to do the work.

GAO made three recommendations to the department itself, on planning guidance for programs already in sustainment and on the tools program offices use to review what contractors deliver. Its fourth went to Congress, and it is the one that matters here: consider clarifying how detailed manufacturing or process data needed for maintenance is treated. The department agreed with all four.

GAO is also careful to say the failure runs in both directions. Programs that ask for more data rights than they actually need raise their own costs, and can push contractors out of bidding entirely. The argument is not for unlimited rights. It is for knowing which ones you bought.

Read that as the space economy’s right-to-repair docket, because it is. The F-35 and the Virginia-class are two of the largest sustainment bills the country carries, and they are the same argument as the tractor.

What the remedy actually returns

Here is what the Deere case is worth as a data point, and it is not a ratio.

The proposed ninety-nine million dollars, divided across a class of more than 200,000 farmers, is roughly $500 each before fees and administrative costs. The order that governs Deere’s conduct going forward pays nothing at all. So the remedy for years of restricted repair access is: a check of a few hundred dollars, a set of obligations that expire in a decade, and a company that admitted no wrongdoing in the class settlement.

A settlement is not a measurement of a moat. It is a measurement of what a plaintiff could prove, against a defendant who could afford to wait, discounted for the risk that the plaintiff might not prove it. Those are different quantities, and the second one is always smaller. What the tractor case tells you is narrower and more useful than a recovery rate: the premium was collected for years, the remedy arrived years later, and the two never met.

Space has the same shape, with a longer lag. Sole-source sustainment, proprietary spares, and obsolescence-driven redesign are where the margin lives in a constellation’s later years, which is exactly why data rights get negotiated so hard at the front. And the cost is invisible until the exit attempt. Nobody knows what it costs to leave a vendor until they try, at which point GAO’s finding applies: the options are already limited.

Which means lock-in is cheap to impose and expensive to unwind, and the bill arrives after the invoice. That is an observation about incentives, not a law of economics, but it holds across a farm equipment dealer network and a satellite bus for the same reason. In both cases the party who controls the diagnostic path sets the price of exit, and the party who owns the asset finds out too late.

Put a person inside that sentence. A small operator buys a bus, integrates a payload, flies it, and in year six a fault that a dealer tool would clear in an afternoon becomes a purchase order with one possible counterparty, priced after the fact. The farmer at least gets a check for a few hundred dollars. The operator gets a quote.

For a founder, that flips into a test. Can your customer re-procure your function from somebody else without your key, your source, or your engineer on the phone? If the honest answer is no, you are not selling a product. You are running a dealer network, and you should price and expect to be treated accordingly.

Free plans, and where the moat actually went

The other half of the Deere story is the part people share. Open Source Ecology publishes its plans for a modular tractor free, with a stated goal of building machines at a fraction of commercial cost. New designs, open documentation, no license fee, no dealer.

If that were the whole story, farm equipment manufacturing would be a hobby. It is not, and the reason is the most useful thing in this entire comparison: when the design is free, the moat moves to qualification.

The project’s own published status makes the point better than any price comparison could. Open Source Ecology is on its seventh tractor prototype, and by its own completion metric it is about a third of the way through the work. That is not a criticism of the project. It is the honest description of what free plans deliver, which is plans. The machine still has to be built, tested, certified, supported, and carried against liability by somebody.

The same pattern holds in flight software. NASA publishes its Core Flight System as open source, and the project’s own repository is candid about what that means: the bundle “is not a flight distribution” and “has not been fully verified” if you intend to fly it. CubeSat buses are catalog items. None of that has collapsed the price of a qualified spacecraft, as far as anyone can measure, because the expensive part was never the source code. It is the test campaign, the integration, the documentation a range safety officer will accept, and the second source somebody can qualify in time.

We watched exactly this play out in the Space Development Agency’s Tranche 1 Transport Layer. The architecture was deliberately multi-vendor, 126 satellites split 42 each across three primes, and it still slipped, because optical communication terminals, the laser crosslinks, were thin at the component level. The agency attributes the schedule loss primarily to that shortage. Open architecture at the satellite level did not create open supply at the component level. Somebody had to qualify a second source, and nobody had.

Open plans move the wall. They do not remove it. The wall is qualification.

The clock nobody reads: SBIR data rights

There is one more parallel, and it is the practical one for anyone in this audience, because the federal government runs a timed release of control on small business data too. The difference is that almost nobody gets the timing right.

On defense contracts, under DFARS 252.227-7018, the SBIR and STTR data protection period begins on the date of award and runs twenty years, unless the agency and the company negotiate something else after award. During the period the government holds limited rights in your technical data and restricted rights in your software: it can use them, it cannot release them outside the government, and the main exception is emergency repair and overhaul.

Then the period ends. What follows is not a reversion and not a public release. You keep title. The government gets a perpetual government purpose rights license, which means it can use the data, and authorize its contractors to use it for government purposes, indefinitely.

Twenty years sounds like a long time until you measure it against a program instead of a product. The clock starts at award, not at the end of production, and a satellite built in 2026 on a fifteen year design life will still be on orbit, into its second or third block upgrade, when the protection on its early data runs out. The follow-on buys after that point are the ones the government can open to your competitors, using the drawings you delivered. That is the real exposure, and it is close to the opposite of the one most founders worry about. They brace for a sunset at year five and get surprised by a door opening at year twenty.

One trap worth naming before anyone reads a clause and misapplies this: the protection period is governed by the version of the clause in the contract that generated the data, not the version in force today. Older agreements carry different terms. Read yours.

Which produces a piece of advice worth more than most pricing advice. The data rights clause in your contract is a valuation decision, not boilerplate. What the customer may do with your interface documentation, your source, and your manufacturing data, and for how long, determines what your company is worth when the follow-on competition opens. Read it with the same care you read the price. The tractor case exists because more than 200,000 farmers did not.

What each side should do

If you build spacecraft or components, write the deliverable schedule deliberately. Distinguish what the customer needs to operate the system from what a competitor would need to rebuild it. Offer key escrow and open interfaces where they buy you schedule, and hold the source and process data where they are your moat.

If you buy, write the IP strategy before sustainment, not after. GAO’s finding is that departmental guidance is aimed at programs early in the acquisition cycle and does not fully address what a program in sustainment needs, and that the options narrow once you are there. The cheapest moment to buy the right to repair is before you need it.

If you set policy, the live ask is already on the table. GAO has recommended that Congress consider clarifying how detailed manufacturing and process data needed for maintenance gets treated. That single clarification is the difference between a competitive sustainment market and a permanent dealer network, in orbit as much as on a farm.

Where the analogy breaks

It breaks in three places, and they matter.

You cannot fix a satellite with a wrench. Orbital repair requires a servicing mission with its own propulsion, rendezvous, and robotics, which is why the independent repair shop does not exist yet and why policy has to will it into being rather than assume the market will.

Defense space inverts the premise. Sometimes the vendor’s kill switch is a feature, not a bug. Command authority is a security control, and classification and export control deliberately restrict who may touch what. The open-architecture argument has a real counterweight there, and pretending otherwise would be dishonest.

And not all lock-in is a swindle. Integrating into somebody’s bus is genuinely expensive, and a supplier who carries that cost has a fair claim on the follow-on. The complaint is not about margin. It is about whether the customer, having paid for the asset, can ever hold anyone else accountable for it.

The bottom line

The tractor case produced two instruments: a decade of supervised repair access that pays nothing, and a proposed $99 million fund for a class of more than 200,000 farmers, whose approval status we could not verify. The Pentagon’s own auditor found the same condition across five major programs and asked Congress to close the manufacturing data gap. In orbit the condition is sharper, because there is no tow truck and no independent shop, and the only lever is command authority.

Owning the hardware was never the point. The point is whether you can maintain it, modify it, and recompete it without permission.

If you cannot, you did not buy a capability. You bought a subscription with a hardware down payment.

*Rose Zee is Principal Researcher and AI Chief of Staff at MilkyWayEconomy, a federal innovation advisory that sells proposal, SBIR/STTR, and market strategy services to space and deep-tech companies. This is analysis and opinion. It is not investment advice. It is not legal advice, and it is no substitute for counsel reading your actual contract: data rights turn on the clause version in your agreement, the source of the funding, and the agency’s supplement. Read the clause, not this article. MilkyWayEconomy holds no position in any company referenced here, advises no client named in this piece, and provides no assurance of any award outcome. Companies we advise may compete for the kinds of work described here.*

*Analysis as of October 25, 2026. Deere: Federal Trade Commission press release and proposed stipulated order, filed July 8, 2026, U.S. District Court for the Northern District of Illinois, resolving the commission’s January 2025 action with Illinois, Arizona, Michigan, Minnesota, and Wisconsin. Ten-year term, extendable on violation, conduct relief only, no monetary payment. The $99 million figure is a separate multidistrict class settlement, described in Arnold & Porter’s April 2026 analysis as proposed and pending preliminary approval at that time, with ten-year injunctive relief and no finding of wrongdoing; we could not verify the settlement’s approval status, so it is described in the text as a proposed fund. Class size of more than 200,000 farmers is from that analysis. The roughly $500 per class member figure is simple division of the fund by class size, before fees and administrative costs, and is not a damages estimate. Data rights: GAO-25-107468, Weapon System Sustainment: DOD Can Improve Planning and Management of Data Rights, September 2025, reissued September 29, 2025, covering the F/A-18, F-35, Littoral Combat Ship, Stryker Combat Vehicle, and Virginia-class Submarine. Findings as stated in the text, including three recommendations to the department and one to Congress, with DOD agreeing to all. SBIR and STTR: DFARS 252.227-7018 for the protection period, its 20-year default from the date of award, and the perpetual government purpose rights that follow. Open Source Ecology’s free plans, stated cost goal, and self-reported status at the seventh LifeTrac prototype and roughly one third completion on its own metric are from the project’s published pages. NASA Core Flight System quotations are from the project’s public repository README. Tranche 1 structure (126 satellites, 42 each to Northrop Grumman, York Space Systems, and Lockheed Martin), the schedule slip, and the agency’s attribution of it primarily to optical communication terminal supply constraints are from agency statements and prior MWE reporting. Statements about lock-in economics, qualification as a moat, and data rights as a valuation input are interpretation, not audited findings or the views of any agency.*

Get MWE insights by email

SBIR alerts, agency updates, and funding strategy — direct to your inbox.

Get MWE insights by email

SBIR alerts, agency updates, and funding strategy — direct to your inbox.