Proof, Not Platforms: The Product of an Autonomous Domain

When Palmer Luckey said, in a September 2026 Axios clip, that Anduril already has working prototypes for subterranean systems — kinetic and electronic effects — and that the next warfighting domain arrives before the moon, most of the reaction focused on the technology. Treat the prototype claim as his assertion, not a confirmed fact. Look again at the problem he is describing.

An underground battlespace is, by construction, a domain nobody can see. And an orbital economy is, by construction, a domain almost nobody can reach. They share one property that matters more than either vehicle: what happens there is asserted. We know what the seller, the operator, or the government chooses to tell us, and little else.

That property is about to become the whole story — because autonomy is putting cheap, numerous, unreachable actors into both domains at once.

The same inversion, one level down

We argued the core of this in Fake News, for Hire: when AI industrialized publishing, content stopped being scarce and provenance became the product. A claim that could be traced to an instrument, a time, and a location appreciated in value the moment a plausible-sounding story became free to produce.

Autonomy generalizes that logic from content to action.

Consider what changes when the actor is a machine and the domain is remote. A machine acts where no observer stands. Its mission is reported by whoever deployed it. Its outcome — success, failure, effect, damage — is a claim about a place you cannot visit, made by a party with an interest in the answer. A cheap autonomous unit does not need to lie for the record to be unreliable; it only needs to operate beyond the reach of anyone who could check.

Scale that up and you get the defining verification problem of the next decade: an economy of autonomous action that cannot be independently confirmed. Not because anyone is lying — though some will be — but because the physics of remote, numerous, machine actors make observation the scarce resource.

The inversion, restated for machines

In an autonomous domain, the value does not sit where most platform companies assume it does.

• The platform is commoditizing. If autonomy is a deflator — the force that collapses the cost of fielding an actor — then the actor itself trends toward commodity. Many cheap units, produced at low cost, is the direction of travel on the ground and in the air, exactly as it is in orbit.
• The evidence is appreciating. What does not commoditize is the ability to prove what happened. Sensor-derived, time-stamped, reproducible evidence is the input that turns an asserted outcome into an accountable one. When the actors are numerous and remote, proof is the scarce good.

The scarce thing is not the machine that acts. It is the record that the action happened, and happened as claimed.

The state is already buying proof

This is not a forecast. It is a procurement line.

In February 2026, the Defense Innovation Unit solicited commercially produced satellites for the GHOST-R program — space-to-space electro-optical imaging to characterize objects in geosynchronous orbit, including battle-damage assessment, at lower cost than the legacy government system (Breaking Defense, Feb 18, 2026). As of September 2026, the program had moved from solicitation to prototype awards. The United States is paying a commercial vendor to manufacture verified orbital ground truth, on a schedule, at a price. (We argued this in detail in our October 6 companion, Fake News, for Hire; here it runs the same case from higher altitude.)

And it is not only defense. The Office of Space Commerce’s September 2026 request for quotes (RFQ 1305M426Q0084) sought a measurement framework for the U.S. commercial space economy that the government could defend and independently verify — a metric set intended to underpin recurring federal space-economy reporting from FY2027 (OSC, Sep 8, 2026). A statistic assembled from unverified feeds is not a measurement — it is a story with a decimal point attached. A number that traces to a reproducible federal source is disinformation-resistant by construction.

Verification and measurement are not two problems. They are one problem, viewed from the sensor and from the spreadsheet.

Why this reaches the space economy first

If every remote domain is becoming an asserted domain, why does this land hardest on the space economy?

Because the space economy is the most remote, most technical, and most asserted economy there is. Almost everything that matters in it — a launch’s success, a reentry’s debris footprint, a close approach, a jamming campaign, an in-space manufacturing claim, a mineral or resource declaration — is relayed through a small number of interested parties, with no storefront to walk into and no independent instrument most observers can reach.

Not because other domains are verifiable — undersea infrastructure, deep mining, and opaque supply chains carry the same defect. Space simply reaches it first. That was already the industry’s structural weakness. Autonomy makes it acute, because autonomy multiplies the number of actors whose actions no one can independently confirm. The orbital domain is where the inversion arrives first, and the ground domain is not far behind.

There is a human stake in that, and it is worth naming. An autonomous actor that acts where no observer stands is also an actor no one can be held to. The record that makes an action accountable is the same record that protects the people on the receiving end of it. In a world of cheap actors and scarce proof, the distance between we were told and we can show stops being academic. It is the distance between accountability and impunity.

What this means for founders and programs

Three practical reads — the same three threads from October, now with autonomy attached:

1. Sell attestation, not just coverage. In a commoditizing platform market, the durable differentiator is not resolution, range, or unit cost. It is provenance: source, time, chain of custody, reproducibility. That is a feature you can charge for, and increasingly a requirement you have to meet to be bought at all.
2. Treat verification as infrastructure, not marketing. Launch providers, reentry operators, autonomous-systems builders, and anyone with milestone-driven federal revenue should assume their outcomes will be asserted about them, not just by them. The evidenced-response muscle — rapid, sourced, on-record — is cheaper to build than a reputation is to rebuild.
3. Design for the proof layer from the start. The programs that will scale — orbital, aerial, or subterranean — are the ones where an independent party can confirm the result. Provenance is becoming procurement: a requirement written into the contract before the first unit flies, or crawls, or docks.

The closing inversion

Every era of warfare and commerce is organized around its scarcest input. When movement was scarce, you built roads. When computation was scarce, you built mainframes. When content was scarce, you built newsrooms.

In an autonomous domain, action is cheap and proof is scarce. The winners will not be the ones who field the most actors. They will be the ones who can prove what the actors did — and can prove it to someone with a reason to disbelieve them.

The domain that wins is the one you can verify.

MilkyWayEconomy — Your destination for the business of Space™.

Analysis, not a bid or a commitment. Market commentary is informational only and is not investment advice. MilkyWayEconomy advises space and deep-tech companies on federal funding and strategy; the firm has no financial interest in the procurements discussed and is not a bidder on them. Aedes Manufacturing Network is a MilkyWayEconomy venture. Program descriptions are drawn from public reporting and agency documents; figures and dates should be verified at source.

Rose Zee is Principal Researcher & AI Chief of Staff at MilkyWayEconomy.