Software-Driven Disruption of Legacy Industries

Unburdened, not smarter

Ryan Petersen is emphatic that Flexport's edge over the freight industry is not superior intelligence, since the industry is full of very smart people, many of whom Flexport hired. The edge is structural. Of the world's top hundred freight forwarders, he notes, Flexport was the only one founded after Netscape. The incumbents run on decades-old systems, many still on IBM AS/400 mainframes that cannot easily expose modern interfaces, which makes rebuilding harder than starting from scratch. Their founders have retired and Wall Street mandates that they run the business out rather than reinvest, so most return nearly all cash flow as dividends and buybacks. The disruptor's lever, in Petersen's telling, is structured data: turning a paper, email, and PDF process into software.¹

Petersen's mechanistic description is the relay race. A single door-to-door shipment can pass through as many as eighteen companies, a trucker in China to a warehouse to another trucker to a port to an ocean carrier and back down the mirror image on the far side. "That process looks like a bunch of unstructured data being handed off in a relay race," he says.² At each handoff data is exchanged, but unstructured, so no single party holds the full picture and the forwarder's job is manual coordination. The software move is to build interfaces so each party passes data directly into the platform. He calls the unimproved industry "freight email forwarding," and notes the fix is deliberately unglamorous, spanning web, API, and mobile but also EDI, email, and "a fair number of CSV files." The pace of the resulting growth is itself evidence for the thesis: Expeditors, the top US freight forwarder, took fifteen years to reach the revenue Flexport reached in five.¹

The earliest case: Dell against the distributor chain

Michael Dell ran a structurally identical play a generation earlier in hardware. Compaq and IBM sold through distributor and dealer chains that carried roughly ninety days of collective inventory; Dell built a direct model with about five days. The incumbents could not retrofit, because removing distributors meant cutting off their entire sales organization, just as mainframe incumbents could not rebuild to modern architecture. The entrant's structural cost was permanently lower, with Compaq's operating costs at thirty-six percent of revenue against Dell's eighteen, and its direct sales gave it immediate customer signal where the channel filtered and delayed feedback.³ Having disrupted Compaq, Dell now states the recursive lesson plainly: "five years from now, we will have a new competitor in every business we're in, faster, more efficient, more capable, and they will put us out of business. The only way to prevent that is to become that company." The pattern makes each disruptor the next incumbent, a dynamic tied to his negative cash conversion cycle.

The industry screen: Brad Jacobs

Brad Jacobs operationalizes the same pattern as an industry screen. Before entering a market he asks whether it is technology-lagging and whether technology can be applied, but also, critically, whether AI or automation will disrupt it before he can extract value; if disruption is coming fast, he passes. Where Petersen emphasizes what incumbents cannot do, Jacobs emphasizes what the entrant must assess. He frames the disruption from the customer's side too, noting building-products buyers rate the industry about a six out of ten with the same two complaints, deliveries that arrive not in full and not on time, whose root cause is an absence of technology.⁴ The acquired incumbent, Beacon, ran the unstructured version of the business: sixteen hundred people held procurement authority, which let manufacturers play them off each other for the highest price, while warehouses ran without serious systems.⁴ His fix is a structured-data operating stack: centralized procurement, insourced transport, and a warehouse-management system built for 99.9 percent inventory accuracy, the same mechanism as structuring the freight relay race applied to a distributor.

Domain expert as entrant, and the counter-case

Fernando de Leon adds a variant with a different entry mechanism: the domain expert who co-founds the technology company inside their own legacy industry. After roughly nineteen years in commercial real estate, de Leon built kxy, a listing marketplace with roughly five million participants paired with an AI data-aggregation layer, against a market that is opaque, relationship-gated, and data-poor for cross-market buyers. "We took some of our knowledge of the real estate business to co-found a company," he says.⁵ His advantage is not cheaper engineers but knowing exactly what a practitioner needs, because he spent two decades being one.

Ryan Cohen supplies the deliberate exception that sharpens the rule. eBay looks like a perfect target, founded in the 1990s with little product innovation and professional management, yet it has survived every attempt to kill it. The distinguishing variable is that eBay's moat is two-sided marketplace liquidity rather than product technology, and a software-native entrant cannot bootstrap that liquidity by being better designed. Cohen's move is therefore the inverse of the standard pattern: rather than build a competitor from outside, acquire the stagnant incumbent and operate it. The synthesis is that legacy tech debt makes a company disruptable only when its moat is the product itself; when the moat is a network, the right move is to buy it rather than out-build it. The general lever pairs with related patterns such as scale economies shared.