Sentinel tripped the law that kills runaway weapons programs. The Pentagon certified itself and kept building — without a price.

The finding, in one paragraph: Congress long ago built an automatic brake for weapons programs whose costs run away — the Nunn-McCurdy Act, under which a "critical" breach means the program dies unless the Pentagon's acquisition chief personally certifies five statutory findings to Congress.▸ The Sentinel intercontinental ballistic missile tripped that brake in January 2024 with the largest breach in the law's history, 81 percent over baseline. The certification came six months later; every finding, including "no alternatives at less cost," was made by the department building the missile. What the brake actually stopped was the program's paperwork: its approval milestone was rescinded, and two years into the ordered restructuring, the program is being built and funded without an approved baseline, without a final design for its software, and without a price — the new estimate is due at the end of this year.

Five documents, read directly.▸ The trigger record is the Defense Department's July 8, 2024 announcement of the Sentinel Nunn-McCurdy review (the release text, including the five certification criteria verbatim). The current status is GAO's National Security Snapshot: Intercontinental Ballistic Missile Modernization Faces Critical Risks and Opportunities (GAO-26-108755, February 18, 2026), and the portfolio context is GAO's 2026 Weapon Systems Annual Assessment (July 2, 2026 — 104 programs, $2.4 trillion, average delivery now over 12 years). The statute is 10 U.S.C. §§ 4371–4376 as applied in the release. The dissent is the July 15, 2024 statement of the congressional Nuclear Weapons and Arms Control Working Group.

The September 2020 Milestone B decision priced the program at $77.7 billion and started engineering development on a $13.3 billion Northrop Grumman contract — the only bid received.▸ The July 2024 review put a "reasonably modified" program at $140.9 billion, 81 percent over. The driver is not primarily the missile: it is the ground. The program was priced on renovating the Minuteman's 450 launch silos and 600-plus facilities across 40,000 square miles of six states; the Air Force has since concluded the 50-year-old infrastructure largely cannot be reused and must be built new — which is why the $140.9 billion carries the label "reasonably modified" rather than final, and why the real number arrives with the new baseline at the end of 2026, two and a half years after the certification that relied on it.

Every year of Sentinel slip is a year added to Minuteman III — a missile fielded in 1970, already the beneficiary of every practical life extension, now facing service through 2050 with what GAO calls growing sustainment risks.▸ Meanwhile the software that runs the new system, per the same snapshot, has neither a finalized design nor agreed development metrics, and GAO's central recommendation is elemental for a project this size: the Air Force has not developed a transition risk-management plan for what the service itself describes as the most complex undertaking in its history.

Read the statute against the certification.▸ Nunn-McCurdy's premise is that a critical breach kills the program unless the Under Secretary of Defense certifies, among five findings: that the program is essential to national security; that "there are no alternatives to the program which will provide acceptable capability to meet the joint requirements at less cost"; that the new cost estimates are reasonable; that the program outranks whatever must be cut to pay for it; and that the management structure is adequate. Each finding was made by the department whose program was on the line — the statute's check is self-administered. The congressional working group's July 2024 statement pressed the soft spot: the review's "failure to truly consider viable alternatives," and cost growth it attributed to "underestimated costs, poor assumptions, incomplete data, and a lack of oversight." Two of the five findings are directly testable against what followed. "The new estimates... are reasonable": the estimate certified as reasonable in July 2024 was provisional, and its replacement still does not exist. "The management structure is adequate": the same certification rescinded the program's Milestone B — the formal judgment that its management baseline could not stand. The certification and the rescission are the same document arguing with itself.

The restructured program's new Milestone B — with its new baseline, schedule, and the first authoritative post-breach price — is due around the end of 2026.▸ Until then, appropriators fund Sentinel annually against the $140.9 billion placeholder; GAO's July assessment keeps it among the programs it tracks for the committees. The first flight in March 2028 is the next hard, public test. And the Minuteman III extension decisions — which silos, what parts, at what cost through 2050 — are a second program growing inside the first one's delays, so far unpriced in public documents.

All figures are from the documents cited in-line. The DOD release text is quoted from a contemporaneous full reproduction; the release itself is no longer directly retrievable from the department's website. The $77.7 billion baseline is implied exactly by the released figures ($140.9B at +81%) and widely documented.