Build vs Buy (Own the Rails)

The mantra and its source

Alex Bouaziz, co-founder of Deel, credits Nikolay Storonsky of Revolut with the mantra behind this bet: "build everything, if you own everything, you do everything well."¹ The claim is that past a certain scale a company should build and own its core infrastructure rather than assemble it from third-party SaaS, because off-the-shelf software "is not tailored enough to the different applications you want to build," and ownership compounds into control, quality, gross margin, and a moat. Storonsky's Revolut builds its own stack; Bouaziz says Deel under-invested in internal tools early and now treats that as a mistake of timing rather than direction.

When to switch from buy to build

Bouaziz is precise about the trigger, and it is a gate rather than a preference. The condition is "a couple hundred percent year-on-year growth plus profitability." At that point, he argues, a company has the confidence and the runway to make longer investments and longer bets. He frames the deeper fork as a year-four decision: whether to be "an acquisition outcome with a great 1-3 products," or to "own the rails, be as close to the metal as possible in a way no one has ever done before."¹

The discipline matters as much as the ambition. Bouaziz notes that building everything "is the dumbest thing to do when you're small but the smartest thing to do as you scale." The prescription is explicitly gated on growth and profitability; premature vertical integration is a known failure mode, which is what keeps this from being a universal rule.

What Deel built

Bouaziz enumerates what owning the rails meant in practice. Deel built its own payroll engine, which he frames as "a little crazy" given that single-country payroll companies became unicorns while Deel targeted more than 100 countries. Its own employer-of-record infrastructure and local entities, built over his father's risk-averse objection, is described as "the best decision we ever made" and generates $25 to $30 million per month. It built an internal ticketing system with an AI layer that routes tickets automatically, and an owned knowledge base with AI on top. Part of that speed came from acquisition: Deel bought a company that had been forced to build multi-country payroll from a tiny home market, an instance of vertical integration from necessity supplying rails Deel would otherwise have built itself.

The second act: rails become a product

Bouaziz teases the natural sequel to owning the rails: externalizing the internal tooling as a product, framed as thinking about the relationship "the same way you think about Amazon and AWS."² The deep-operations tools Deel built for itself, which he says saved tens of thousands of hours, could be sold to other operations-heavy companies that genuinely need to function across 150 countries. In that telling, owning the rails is not only a cost and moat play; at scale the rails become a new revenue line and a new addressable market.

Tensions

Bouaziz presents the bet as Deel's structural answer to the argument that software moats are disappearing because money and compute now solve software. If code and data are replicable, he reasons, the durable edge moves to owned infrastructure, owned data, regulatory entities, long contracts, and operational depth across geographies, which is exactly what owning the rails accumulates. The tension worth tracking is the same one the gate is meant to manage: the trade of speed for control only pays once the company is large and profitable enough to afford the slower path, and the boundary between disciplined vertical integration and premature over-building is judgment, not rule.