The NewsRupt

The dot-com crash killed a lot of bad ideas, and history has been kind to it for that reason. But the period also buried a category of ideas that failed not because they were wrong but because they were early — and the current technology cycle is quietly resurrecting them. Home-delivered groceries, subscription everything, online used-goods markets, on-demand logistics, social commerce: nearly every fashionable model in the past five years has a direct ancestor with a 1999-era website and an obituary. What distinguishes the second life from the first is worth understanding, because the differences are also a checklist for which of today's resurrections will survive.

Consider grocery delivery, the model that consumed more capital per failed idea than perhaps any other. Webvan built automated warehouses in 1999 for cities that could not fill them, and collapsed. The 2010s wave — better routing software, smartphone penetration, gig logistics — got further and still burned billions finding the same truth: groceries are low-margin, dense-with-operations, and habit-driven. The companies finally reaching sustainability have done it by narrowing their promises — scheduled delivery windows, fewer SKUs, member economics rather than growth-at-all-costs — which is to say, by behaving like the boring grocery businesses they sit on top of. The idea was never wrong; the operating discipline was.

What has actually changed between eras falls into three buckets. The first is infrastructure that no longer needs building: payment processing, cloud compute, identity, logistics marketplaces and communications APIs are commodities now, which removes the capex that made 1999 models capital-intensive before earning anything. The second is behavioural: the customers who had to be convinced to buy online in 2000 are the ones paying online today without noticing. The third is data discipline — modern operators can measure unit economics per order from day one, whereas their predecessors were often flying blind by design, because the investor story mattered more than the margin.

Against these genuine improvements sits one persistent constant: human impatience with capital markets. The current cycle's AI-adjacent spending has recreated some of the old pathology in new clothing — infrastructure buildouts sized to projected demand rather than observed demand, revenue from related parties rather than customers, and valuations that price in years of flawless execution. The specific model matters less than the funding temperature. When capital is patient, second-life models get to find their real economics; when it turns, the discipline disappears exactly when it is most needed.

There is also a category of resurrection that deserves scepticism precisely because it is fashionable again: the model whose economics only ever worked at growth-stage scale. Marketplace businesses that lose money on every transaction in pursuit of network effects, subscription services priced below sustainable cost to build an audience for advertising that never arrives, delivery networks whose density assumptions require a city to be fully served before any route profits. A good test for any of today's resurrections is to ask which of the three era-changes above actually fixes the old model's fatal flaw. If the answer is none — if the flaw was structural rather than infrastructural — the second burial is only a matter of funding weather.

A limitation worth stating: comparisons across eras rest on press coverage and company disclosures, which over-represent headline failures and successes while missing the quiet middling outcomes. Capital-availability conditions are assessed qualitatively. Survivorship in the modern cohort is still too early to score; several models cited as successful remain venture-supported rather than independently profitable.

The NewsRupt files this from the London desk. Corrections and right of reply are handled under our published standards policy.