A $677M "operating system for autonomous flight" with $1M of quarterly revenue — generational asymmetry, or a SPAC story priced on arithmetic that isn't contracted yet?

**Quick background**: Merlin (NASDAQ: MRLN) went public via de-SPAC in March 2026. It retrofits existing military aircraft with an AI "digital pilot" — anchor program is a $105M ceiling IDIQ with USSOCOM for C-130J autonomy, plus KC-135 work. Q1 2026: $1.0M revenue, $23.3M Adjusted EBITDA burn, \~$183M cash after an $80M PIPE, no debt. Stock is at $7.02 (11 June 2026), market cap \~$670M, 52-week range $5.78–$17.00.

The interesting part of the bull case: management's stated unit economics are $3M one-time integration + $2M/year recurring license per aircraft, across 800+ military aircraft described as under contract — a claimed line of sight to $1.6B of annual recurring software revenue. The variant perception: autonomous flight is a categorically easier problem than self-driving cars (cooperative airspace, procedural actors, minutes not milliseconds), and the market is applying a SPAC discount to a real defense software annuity.

Where I kept poking holes:

* "Under contract" does heavy lifting. The C-130J award is an IDIQ with a $105M *ceiling* — maximum authorised spend, not committed revenue. Task orders are at the customer's discretion. TTM revenue is $7.7M.
* The "no dilution needed" claim lasted 44 days. The March thesis said \~$146M of cash eliminated the need to raise. On April 29 the company raised an $80M PIPE (8M shares + 4M warrants at $6.67). At $23.3M/quarter burn, post-PIPE cash covers \~8 quarters.
* 2026 guidance of $32M needs $8M/quarter. Q1 delivered $1.0M. Roth cut its target from $25 to $15 after the print (still Buy).
* SBC plus equity payments to non-employees ran at 193% of Q1 revenue.
* Lock-ups on the overwhelming majority of shares start expiring around September 2026, against a public float of roughly 5M shares.
* The civil TAM behind the big scenarios sits behind a political gate: ALPA and allied unions formally oppose reduced-crew operations, and the "against" side only needs one incident, anywhere in the industry.

**One genuinely interesting angle**: the certification race doubles as a referendum on architecture. Merlin's ML-based stack has no established FAA certification pathway; Reliable Robotics is pursuing the same prize with a deterministic, non-ML design the FAA's framework was built to evaluate. Whoever certifies first doesn't just win a milestone — they reveal the regulator's preference. Meanwhile Merlin did just complete its C-130J Critical Design Review with USSOCOM (June 4), ahead of at least one broker's expectation, so the defense side is executing.

**Question for the sub**: when a thesis's downside case depends on contracted-fleet arithmetic ($53/share "bear case") but the contracts are IDIQ ceilings and the company has already raised twice in three months — what would you need to see before treating that floor as real?

This is quite a challenging "territory" for me, it is against all core value concepts, so interesting to see your opinion.