Codifying Tacit Knowledge: What a SpaceX Spinout's Test Software Means for Aerospace's Barriers to Entry

Aviation Week reported on September 14, 2026 that Sift, a company founded by former SpaceX engineers including CEO Austin Spiegel, supplied the data-unification platform behind Heart Aerospace's electric X1 demonstrator, whose first flight took place in August 2026. The platform consolidates telemetry from ground testing, preflight checks, high-speed taxi runs and flight operations into a single system.

The interesting economic question is not what Sift's software does — telemetry unification is a familiar category — but where the underlying capability came from. At SpaceX, Spiegel spent five years building internal tools that centralized data from simulations, component tests, integrated vehicle testing and operations, letting engineers reuse analytical routines across programs. That capability was not written down in a manual; it accumulated as an organizational routine — the kind of practical, partly tacit know-how that evolutionary economists Richard Nelson and Sidney Winter (1982) argued is the real unit of a firm's competitive advantage, because routines resist easy imitation or transfer.

Eric von Hippel's concept of 'sticky information' (1994) offers the mechanism: information is costly to move from where it is generated to where it could be used, and that cost — not any legal protection — is often what keeps valuable know-how locked inside the firm that produced it. Aerospace testing discipline has historically been sticky in exactly this sense: a small number of firms with decades of high-cadence hardware iteration (SpaceX, and before it the legacy primes) built up testing rigor that a new entrant could not simply buy, only accumulate slowly and expensively through its own trial and error.

Sift's product is what innovation economists Robin Cowan, Paul David and Dominique Foray (2000) called codification: converting tacit, experience-based knowledge into an explicit, transferable artifact — in this case, software rather than a manual or a patent. Codification does not just describe knowledge, it changes its economics: once encoded, the marginal cost of transferring it to a new user collapses, even though the original accumulation was slow and costly. That is precisely what has happened here — years of SpaceX's internal testing discipline, now available to any aerospace developer willing to pay a subscription.

For smaller aerospace developers — electric and urban air-mobility programs like Heart Aerospace's chief among them — this matters strategically more than it might appear. A meaningful part of the capability gap between incumbents and new entrants in hardware-intensive aerospace has rested on exactly this kind of accumulated, sticky testing know-how, not only on capital or certification experience. Tools that codify it compress that gap, in the same way flight simulators once compressed the gap in pilot training, or cloud infrastructure compressed the gap in enterprise computing capability between large and small firms.

Why would this capability leave SpaceX at all, if it was a source of advantage? David Teece's work on the 'appropriability regime' of innovation (1986) helps explain: some forms of tacit knowledge are, in practice, impossible to fully contain once the engineers who built them are free to leave — a dynamic AnnaLee Saxenian documented in her account of Silicon Valley's job-hopping engineers diffusing know-how between competing firms. Rather than let that diffusion happen informally and uncompensated, Sift's founders productized it — turning an inevitable spillover into a monetizable asset, a pattern seen before when veteran chip-design engineers spun out the first commercial EDA (electronic design automation) software companies from what had been internal semiconductor-house tooling.

The forward-looking piece is what Sift is adding on top: AI features that let engineers query telemetry in natural language, with an explicit ambition toward AI agents performing root-cause analysis of flight anomalies. If codification so far has mainly transferred the data-organization layer, AI agents would begin codifying a second, harder layer — the interpretive judgment senior test engineers apply when diagnosing an anomaly. That is a materially bigger shift: it would mean the experience gap eroding is not just about having clean, unified data, but about having the judgment to read it — historically the least transferable part of aerospace testing expertise of all.

For aerospace and defense executives, the signal to track is not this one vendor but the trend it exemplifies: internal test-data infrastructure, long treated as a durable and largely invisible source of competitive advantage, is becoming a purchasable commodity. Firms that built their edge mainly on accumulated testing discipline should assume that edge has a shrinking half-life, and that the next differentiator will be the judgment layer AI agents are only beginning to approach.