An Airline Is Just a Software Stack Nobody Finished Building

From the Travel Tech Podcast newsletter for episode 27, An Airline Is Just a Software Stack Nobody Finished Building.

There's a moment in this conversation where my guest compares running an airline to selling hamburgers: your meat costs triple overnight, but you can't triple your menu price on the same day. That's not a metaphor he reached for casually; it's how he explains why a company that hit $350 million in revenue in three years still went bankrupt. I wanted that kind of unfiltered read on the business, not the usual airline talking points, in front of our readers across aviation, travel tech, and AI.

He started as CEO of one Icelandic carrier, moved to deputy CEO of another, then took the top job at a startup low-cost airline built on WOW air's old team and DNA, scaling it from zero revenue and no way to process a payment to roughly $350 million a year in three years flat, before it went under this past winter. Listen on Apple Podcasts, also streaming on Spotify and YouTube.

The tension running through the whole conversation: he did almost everything right by his own account, and the business still didn't survive one input cost spike.

That tension has a structural cause

That tension has a structural cause: airline distribution and inventory systems are far less standardized than they look from the outside. A new carrier can't plug into an existing seat-and-ancillary distribution stack the way a new e-commerce brand plugs into Shopify; it has to negotiate OTA and GDS integrations one relationship at a time, which is exactly why building "minimum viable" airline infrastructure from scratch ate months and real capital before PLAY sold a single ancillary product reliably.

AI is adding pressure here without yet closing the gap. My guest is skeptical that customer-facing AI agents will book flights end-to-end anytime soon, given that even basic seat-and-bag selection on major OTAs is a relatively recent capability; where he does see near-term value is internal: helping new commercial hires make sense of load factors, route economics, and fragmented operational data faster than tenured staff currently explain it to them.

This week's guest is Birgir Jónsson, former CEO of PLAY Airlines and former deputy CEO of WOW air, on why aviation resists new talent, why fuel price can erase good execution overnight, how PLAY's Iceland-hub model lost its footing after 2022, what it takes to train technical experts into managers during hypergrowth, and where AI genuinely helps an airline today.

Key Takeaways

• Aviation's technology gap is real, not perception: Selling a seat looks like a 30-year-old solved e-commerce problem, but distribution and ancillary integrations are still negotiated deal-by-deal, not standardized across the industry.
• A five-year-old airline already has legacy debt: Being new doesn't mean being unencumbered; early technical decisions calcify into constraints just as fast as at an incumbent carrier.
• Fuel price is the variable that ignores execution: PLAY's business case assumed roughly $660 per ton; after the invasion of Ukraine it rose toward $1,500, and no amount of on-time performance offsets that swing.
• Hub strategies expire when capacity shifts: PLAY's Iceland-as-a-stopover model between Europe and the US worked well historically but lost relevance once new aircraft types and geopolitical disruption pushed capacity onto direct Atlantic routes.
• Scaling breaks management before it breaks operations: Growing from a handful of staff to hundreds within a year requires training technical experts as managers, a distinct skillset that companies often assume comes for free with promotion.

🔧 Three to Build

Birgir scaled an airline from zero to $350 million in three years and still watched it go under. Here are three systems drawn straight from where he says the business actually broke.

Build a macro risk model that flags when your business case assumptions are breaking.

Airlines write business cases against a fuel price benchmark, then keep flying the plan even when the market has moved far past it. A model that continuously compares live commodity prices against the assumptions baked into your route P&L gives finance and ops teams an early signal rather than a post mortem. PLAY was still flying routes priced at $660 per ton when the real cost hit $1,500. That gap is detectable in near real time if you are watching for it.

Catch: the hard part is not the model. It is building the organisational trigger that actually grounds a route when the threshold is breached.

→ Financial risk modelling for airlines

Build the seat and ancillary inventory layer a new carrier otherwise assembles by hand.

Birgir's core point is that selling an airline seat looks like a solved online retail problem from the outside, yet every new airline still stitches its distribution and inventory stack together deal by deal. A system that automates inventory management and seat allocation against live demand is exactly the piece PLAY had to build from scratch before it could reliably sell a single ancillary. Get it right once and it becomes the infrastructure other carriers rent rather than rebuild.

Catch: allocation logic that is too aggressive protects yield on paper but strands seats that a looser model would have sold. The tuning is the whole game.

→ Automated inventory and seat allocation

Build a network model that tells you when a hub strategy has quietly expired.

PLAY's Iceland as a stopover between Europe and the US worked beautifully until new aircraft types and geopolitical shocks pushed capacity onto direct Atlantic routes. By the time that shift showed up in load factors, the strategy was already stale. A route network model that reads passenger flow patterns and market trends can surface that erosion while there is still time to redeploy aircraft rather than defend a dying route.

Catch: the model will keep recommending routes your brand and slot portfolio cannot actually fly. Network optimisation only pays off when it is constrained by the assets you really hold.

→ Route network planning from passenger flows

Alex Brooker, Host of the Travel Tech Podcast, Founder of Airside Labs