A patent is its examined, allowed claims — and a published application has neither been examined to conclusion nor allowed. That is the whole distinction, and it is the one most often blurred when a spacecraft "patent" is reported. Under U.S. law, the document the public usually sees first is a published application: a disclosure the statute forces into the open about eighteen months after filing, long before the Patent Office decides whether anything is patentable. Calling it a patent gets the legal status wrong and, with it, the competitive read. The application tells you what a company chose to disclose and pursue. Only the grant tells you what it actually secured.

The eighteen-month timer is statutory, not discretionary. Section 122 of the Patent Act sets the rule directly.

"...each application for a patent shall be published, in accordance with procedures determined by the Director, promptly after the expiration of a period of 18 months from the earliest filing date for which a benefit is sought under this title. At the request of the applicant, an application may be published earlier than the end of such 18-month period."— 35 U.S.C. 122(b)(1)(A), U.S. Code

That clock is why patent data is a delayed window. The earliest filing date "for which a benefit is sought" can reach back through a chain of earlier applications, so a publication that appears today commonly reflects research conceived eighteen months — or more — earlier. When a published application surfaces describing, say, a new thruster mount or antenna architecture, it is reporting a research decision the company made well before the publication date. The disclosure is real; the timing is lagged.

Reading the kind code

The fastest way to tell an application from a grant is the kind code appended to the document number. A U.S. application publication carries an A-series code — A1 for a first publication of an application filed without a prior publication. A granted patent carries a B-series code — B1 for a grant issued without a prior publication, or B2 for a grant that follows an earlier A1 publication. So US 2026/0123456 A1 is a published application; US 12,345,678 B2 is a granted patent whose application was published earlier as an A1. The numbering format differs too: applications use a year-sequence number, grants use a running patent number. These are not cosmetic. The A document confers no right to exclude; the B document does. A competitor can read and even practice the subject matter of a pending A1 publication without infringing anything, because there is nothing yet to infringe.

The exceptions to publication are equally informative. Section 122 states that an application shall not be published if it is no longer pending, if it is subject to a secrecy order under section 181, or if it is a provisional application. The first means abandoned or already granted applications drop out of the publication stream. The second is the national-security carve-out: defense-sensitive filings can be withheld from publication entirely, which is why parts of the space and defense record are absent. The third reflects that a provisional application is a placeholder that itself never publishes or matures into a patent — it only establishes a priority date for a later nonprovisional filing. There is also a non-publication option: an applicant who certifies the invention has not and will not be the subject of a foreign filing can request that the application not be published at all, trading public disclosure for continued secrecy until grant.

What each document licenses you to conclude

The practical discipline follows from the legal one. From a published application, you may conclude that the named applicant filed, on or before a stated priority date, disclosing the described subject matter and seeking the claims as filed. You may not conclude that those claims will issue, that they will issue in their filed scope, or that any enforceable right yet exists. Examination routinely narrows independent claims; some applications are rejected and abandoned outright. From a granted patent, you may conclude that the Office examined and allowed specific claims, and that the owner holds the right to exclude others from practicing what those allowed claims cover for the patent's term. Even then, the scope is the allowed claims — not the abstract, not the title, and not the broadest language the application started with.

One more consequence of the eighteen-month rule is worth naming: provisional rights. Section 154(d) of the Patent Act provides that, once a patent issues, the owner may in defined circumstances obtain a reasonable royalty for certain pre-grant activity that occurred after the application was published — but only if the published claims and the issued claims are substantially identical, and only after the patent actually grants. In other words, publication can begin to put competitors on notice, but the enforceable right still depends on the grant arriving and the claims surviving examination largely intact. This is the legal mirror of the analytical point: the published application matters, but its weight is contingent on a future grant. Until that grant exists, a published A1 is a disclosure with a notice function, not a property right a court will enforce. Treating the publication date as the moment exclusivity attached gets the sequence backwards.

For competitive intelligence in space and defense, this distinction sets the altitude of every claim you can make. A wave of published applications in electric propulsion or phased-array antennas is evidence of where companies are directing R&D and what they are willing to disclose — a forward-looking signal, appropriately hedged. A cluster of granted patents in the same area is evidence of secured, examined rights that can be asserted. Conflating the two inflates the story: it turns intent into entitlement. The cleaner reporting names the document for what it is — application or grant, A-series or B-series — and confines the conclusion to what that status actually supports.