MRLN · Merlin Labs $2.02 -0.04 (-1.94%) 2026-SEP-18 12:49 EST

Merlin is building software that flies existing aeroplanes without a pilot — described here as "the Waymo of the skies." Mart is unusually blunt about what that means for position sizing: 50–100x upside if it works, 100% downside if it does not, so size it such that a zero does not change your life.

The news is a regulatory certificate, which sounds dull and is not. New Zealand's aviation authority granted Merlin a Part 146 design-organisation approval, meaning its own engineers can now sign off certain aircraft design changes internally rather than routing every one through an outside firm. Combined with the operating and maintenance certificates it already holds, Merlin now has the complete set of permissions needed to fly its own certified aircraft on revenue routes there — assembled over four years, one audit at a time. The moat argument is that a competitor can copy software but cannot copy a regulator's delegated trust.

He also relays an independent investor's visit to the flight-test site, where the same two components that passed a formal certification stage were the ones physically flying the aeroplane — the check you make a trip to run. One structural point worth holding: none of Merlin's target aircraft are modern fly-by-wire designs, and neither are the world's cargo and military fleets, so a company that can retrofit reaches thousands of aircraft already flying rather than only future deliveries. Near-term risk is mechanical: in mid-September ~44m insider shares unlock on the same day a preferred instrument resets its conversion price.

Merlin is building AI that flies aircraft with reduced or no crew, starting with the military C-130J. It's the highest-variance name in the book: ~$1m of quarterly revenue trading at more than 50x sales on a tight float — "about as far out on that curve as you can stand." The stock printed a new low at $4.20, down over 30% from a $6 open and far below the ~$9.50 spike on the CDR headline.

Mart went looking for the usual cause of a drawdown this size — a short report, a lost contract, pulled guidance, a resignation — and found nothing. The move is mechanical: when rates back up, money exits speculative growth violently regardless of company news. "Merlin was not sold for anything Merlin did, it was sold for what Merlin still is" — an idea on the verge of implementation, which trades badly when money gets expensive. Meanwhile the Critical Design Review passed, retiring the biggest engineering risk, and analysts still carry Buy ratings (Cantor initiated at $11 vs ~$13 consensus). So he bought more and made it a full position: "If I liked it at $6, I am not going to talk myself out of it at $4.40." His model's bull case is a 48x from here — with the flight demonstration and production orders still the whole trade, and the risk that implies.

Merlin Labs is a post-SPAC company building "Merlin Pilot," an AI system that flies aircraft with reduced or no human crew — starting with the military C-130J transport. It is a classic high-risk, high-reward micro-cap: tiny current revenue, a huge potential market if the technology certifies, and a stock that swings violently on news and on the broader appetite for speculative growth.

The stock has "taken a massive hit" in the rates-driven selloff — not because anything at the company broke, but because expensive money crushes exactly this kind of pre-revenue, high-multiple name. Crucially, the Critical Design Review (the single biggest engineering risk in the early thesis) has already passed. So Mart is buying the same asymmetric bet at a lower price with less risk attached, and has sized it up to a full position.

Merlin builds software that lets aircraft fly with fewer (eventually zero) human pilots — think "self-driving for planes." It bolts onto existing aircraft rather than building new ones, starting by letting one pilot do the work of two. The military is the first customer (a C-130J autonomy program just cleared a big engineering milestone), with the much larger commercial-aviation market as the long-term prize. Uzo compares the whole setup to AST SpaceMobile, where he made a fortune on the warrants as it grew from a ~$1bn startup into a ~$40–50bn company.

He treats it like a venture-capital bet that happens to trade on the stock market: expect dilution, big swings and delays, but the payoff could be enormous. What he likes are the stacked protective "moats" — years-long safety regulation that keeps rivals out, the fact that once your software is certified into a fleet nobody rips it out, data gathered from planes already flying, and recurring software-style licensing fees. Big names (GE, General Dynamics, Honeywell) partner with it instead of building their own, and insiders are buying rather than selling. The catch: it's almost binary — a 30–50x (even 100x) winner if it works, but a genuine zero if it doesn't, and there's some messy financing (convertible preferred shares that could swell the share count). Size it small and be patient for 3–5 years.

Think of CDR as the government's final exam on the blueprints. Before yesterday, there was a real chance the US military could have looked at Merlin's design for flying a C-130J autonomously and said "go back and redo this." They didn't — they said "the design is sound, now go build it." That's a big deal: a lot of defence startups never make it past this gate.

What it doesn't change: the price targets ($16 bear / $77 base / $216 bull) and the fact that the real downside is still $0 if execution fails. What it does change is the odds. Mart moves his probability from roughly one-in-three chance of a bad outcome down to one-in-four, and shifts more weight toward the middle and upper cases. The company still has to physically integrate the software into the aircraft, run ground tests, and then actually fly it — that's all still ahead, targeted for later this year — but those are engineering hurdles, not "does the design even work" hurdles.

The stock is already up ~60% from the $6 entry because the market is pricing in exactly this shift. Two Wall Street analysts now cover it (both Buy-rated, consensus ~$13). There are also a lot of short-sellers caught offside — over 4.6 million shares betting the stock goes down, at expensive borrow rates — which adds fuel to near-term moves. Mart isn't calling a squeeze, but he notes the setup is as favourable as it gets for people who are already long.

Merlin Labs makes the AI "brain" that flies a plane. It is a software platform you install into existing aircraft so it can take off, cruise, and land with a reduced or eventually zero human crew. The genuine differentiator: it can communicate with air traffic control over the radio the way a human pilot does, understanding accents and complex instructions — something almost no competitor has achieved — and every flight it completes adds to a proprietary dataset that makes that skill progressively harder to replicate. It has been flown on 7 different airframes from a small Long-EZ up to the C-130J military transport, and it holds the most stringent aviation-software certification class (DO-178C Level A).

Mart added it at $6 as the first non-commodity name in the Codex portfolio. His valuation work (covered in detail in the separate deep-dive newsletter): bear ~$16 if execution is minimal (and $0 if it fails entirely — so the position is sized for that possibility), base ~$77, bull ~$216 (a 35x from entry). The catalyst calendar: a C-130J Critical Design Review expected in 2H 2026, New Zealand civil aviation certification progress, and likely additional sell-side analyst coverage coming. The bull case is enormous; the honest caveat is that the company is burning through cash, the field has credible competition, and a de-SPAC stigma weighs on sentiment.

Merlin sells the "brain" that flies a plane. It's a software box you bolt into a normal aircraft so it can take off, fly and land with fewer human pilots — or none. The clever part is that it can talk to air traffic control over the radio the way a person does, understanding accents and instructions, which almost no rival can do; every flight it makes adds to a private dataset that makes that skill harder to copy.

Why now: airlines and the military are running out of pilots (thousands retiring every year, pilot pay now bigger than fuel), so anything that cuts a two-pilot crew to one is valuable. Merlin has a real US special-operations contract on the C-130J military transport — but Mart is careful that it's a spending ceiling ($105m), not cash in hand, and the company is still burning ~$90m a year against ~$183m of cash, so it's a 2–3-year bet, not a this-quarter story.

The maths: if it just keeps executing, his "bear" case is ~$16 (≈3x the $6 entry); the middle case ~$77; the dream case ~$216 — about 35 times. The honest catch is the real bear case is $0 if the demo fails and the cash runs out, so he sized the position small. It earned a spot because the downside still pays ~3x while the upside is enormous — the kind of asymmetry he wants, with conviction backers (Baillie Gifford) and full insider lock-up behind it.