GPS is a single point of failure. The Army just paid a startup to prove it.

An Austin company you cannot buy has just won an $11.26m contract from the U.S. Army. Tern AI, a private firm, received a firm-fixed-price award under the SBIR Phase III programme to develop an "independently derived positioning system"—a polite way of saying navigation that works without GPS. Retail feeds have carried the news as a defence-win headline. It is worth reading, but not, as it happens, for the reason the headline implies.

Start with what the contract actually is. SBIR Phase III is the point at which the Pentagon stops funding research and starts buying a product. Where the earlier SBIR phases are grants for feasibility and prototyping, Phase III is an ordinary procurement contract: with no ceiling on value or duration. For a firm the size of Tern, which only a year ago closed a $7.5m seed-extension round, $11.3m is close to a whole corporate lifetime's worth of capital. The award is a rounding error in defence budgets and a step-change for the company.

The reason the Army is paying is that a fifty-year-old assumption is coming apart. GPS—the free, government-run satellite signal buried in everything from phone maps to precision farming—was designed to be robust, and for decades it was simply trusted. That trust is now a strategic liability. The live experiment is Ukraine, where Russian forces have jammed and spoofed GPS continuously since 2022, degrading Western-supplied munitions and forcing autonomous systems to operate blinded. The commercial world has felt the same fragility: airlines in the Middle East logged roughly 1,500 GPS-spoofing incidents a day in 2024, a 62% jump on the year before, and jammers that can blot out the signal cost as little as $100 online.

Tern's answer is software rather than hardware. Its "independently derived positioning system" (IDPS) reads the road network with onboard sensors and a proprietary algorithm, the way facial recognition reads a face, and derives position without any external signal at all. The company says it has run on four classes of military vehicles under live jamming with less than 0.2% drift. The military makes an ideal first customer not because it is sentimental about startups but because its cost of a lost signal is highest—a cockpit or a munition has no "recalculate route" icon.

Here the honest calibration begins. The assured-PNT category Tern is selling into is real but small. One research firm puts the market at $1.6bn in 2024, rising to $2.7bn by 2030—single-digit billions in an industry where a single fighter programme is worth more. It is also an incumbent's market: inertial-navigation houses and defence primes have spent decades building the safety cases and supply chains that militaries demand. The software challenger's edge is that it displaces a ubiquitous, free incumbent, and its weakness is that it is asking large, conservative institutions to abandon a habit of fifty years. Tern's own founder concedes the resistance. Government contracts are the wedge that lets it in, which is precisely why this one matters.

Which brings the reader to the uncomfortable truth of the headline. Tern AI is private. There is no ticker, no prospectus, no way for an ordinary investor to own a share of this particular bet. The news is therefore best read as evidence about a category and a lesson in parsing defence-contract reporting, not as a tip. When a private firm announces "awarded $11.26m," the useful question is never the dollar sign; it is which phase of the SBIR ladder the money stands on. A research award is colour. A Phase III contract is a customer with a budget and a deadline, and that is a different species of fact.

What the investor can take from this story is a re-ranking of a quiet risk. For five decades markets treated GPS as a given, like gravity or the postal service—free infrastructure with no price to discipline its use. The jamming in Ukraine, the spoofing over the Middle East, and now a stream of small Army contracts tell a different story: the assumption is being repriced as fragile, and the first movers are startups the investor cannot own by name. The category, though, is larger than any one firm, and it will be felt wherever public trusts, drones, and precision dependents trade. That is the structural change. The cheque to a private Austin company is just the receipt.