Merlin's New Zealand Retreat Is Less About Geography and More About Where Cockpit Autonomy Software Actually Scales
This is an update to a story I covered back in July, when Merlin’s autonomous Caravan landing at Oshkosh read as a compelling demonstration that the certification path was moving. A lot has changed since then.
On September 9, Merlin, Inc. announced plans to pursue a civil certification strategy that sets it on a path to bring autonomy to large aircraft. The headline framing was about ambition — 7,000 commercial transports, a looming pilot shortage — but the more revealing detail was buried underneath: to concentrate its efforts on large aircraft, the company is withdrawing its small aircraft Supplemental Type Certificate application with the Civil Aviation Authority of New Zealand, and will apply the technical foundation and regulatory experience it built there to the new program.
That’s not a pivot. That’s a full reset of the certification strategy.
The Real Signal Isn’t the Target Market — It’s the Certification Logic
Small aircraft STC work and large aircraft type certification are not the same regulatory exercise done at different scales. They involve different FAA offices, different software assurance standards, different levels of scrutiny on AI and ML systems, and fundamentally different risk tolerances from the certification authority. What Merlin is really announcing is that it’s abandoning one certification playbook and betting its future on a harder one.
The strategic rationale they’ve laid out leans heavily on the defense side: the company plans a phased approach to certifying its Merlin Pilot autonomy system on large, high-assurance aircraft, leveraging experience from its New Zealand program and its C-130J autonomy work with U.S. Special Operations Command. And the regulatory artifacts from New Zealand aren’t being discarded — Merlin’s New Zealand certification work achieved SOI 3 and a Stage 4 issue paper on AI/ML-based natural language processing, which will inform future collaboration with the FAA and other regulators. That’s a meaningful data point: they’re carrying forward IP and regulatory paperwork, not starting from zero.
The argument is that military high-assurance work and a partial New Zealand regulatory record give Merlin a faster path than starting from scratch. I think the certification gap between where they are and where they’re going is being underestimated. Defense assurance frameworks and FAA or EASA commercial certification aren’t just different in degree — they’re built on different assumptions about who operates the system, under what oversight, and what the acceptable failure modes look like. The regulatory environment compounds that gap: EASA, which also prohibits single-pilot operations in commercial airliners, recently performed a study on eMCO and concluded that it would be unsafe with current cockpit technologies. That was particularly notable because Airbus was pushing for eMCO with its upcoming A350F freighter. EASA has shut the door on this for now, pivoting focus instead toward boosting safety in existing two-pilot cockpits. I covered the EASA eMCO delay in depth back in August, and that regulatory signal hasn’t softened since.
The Pilot Shortage Argument Is Real, But It’s Not a Certification Argument
Merlin has specifically called out that there are over 7,000 large commercial transports in the US, and that the country could be short of over 30,000 pilots by 2032. That workforce math is genuine and the addressable market framing is understandable — but it conflates a demand signal with a regulatory pathway. The FAA and EASA don’t accelerate certification timelines because an industry is short-staffed. They might accelerate standard-setting conversations, but the assurance requirements on AI in a transport-category cockpit don’t bend because of a hiring crisis.
What the pilot shortage argument actually does is strengthen the case for a different category of cockpit software: not full autonomy, but AI co-pilot assistance tools that reduce workload within existing two-crew regulatory frameworks. That’s the nearer-term market, and it’s the story I’d be watching with Merlin. Their NLP work — specifically the Stage 4 issue paper it reached in New Zealand — is the piece of their technical foundation that could find a commercial expression without requiring eMCO or single-pilot certification at all. Airlines can actually plan around a timeline like that.
Merlin and its autonomous systems won’t result in crew size reductions for the foreseeable future, and that’s a fair read of where things stand. Whether Merlin is genuinely committed to the full-autonomy endpoint or whether the assistance layer becomes the real product is still an open question — but the defense-to-commercial translation story is usually more complicated than the press release suggests, and the regulatory artifacts they’re carrying forward are worth watching closely.