Pip: Dave once crawled the entire internet in an afternoon — printed list, tape drive, white screen — and hit the end. That story is doing a lot of work in today’s episode.
Mara: It is, and it connects directly to where we’re headed: the parallels between the dot-com infrastructure collapse and the AI investment panic happening right now, and what history actually says about where this goes.
Pip: Let’s start with the bubble that wasn’t quite what everyone thinks it was.
Lessons the Dot-Com Era Left for AI
Mara: The central question here is whether today’s AI spending panic is a replay of dot-com recklessness — or whether the people spending the money actually learned something the first time around.
Pip: The post sets up that answer with a direct memory from the trading floor, where a young investment banker dropped a business plan on the desk and asked not whether her idea was possible, but how to build it — food delivery ordered online, delivered to offices, mid-nineties.
Mara: And the response was honest: “There’s one major speed-bump you are going to run into: All these restaurants are going to need dedicated data lines and a dedicated data line, like the ones we use here at the bank, costs roughly $10,000 a month to maintain.”
Pip: The idea was real. The infrastructure wasn’t there yet. Grubhub didn’t arrive until 2004, which is exactly the point — the concept was sound, the roads just hadn’t been paved.
Mara: That’s the core argument applied to AI today. The dot-com crash wasn’t fundamentally a financial failure; it was an infrastructure gap that spooked investors into pulling out before the roads got built. The post is direct: “The crash was not because of the finances of these companies, it was created by the pullout of finances to those companies, now without highways leading to the Disney Lands of technology.”
Pip: And AI leadership, the argument goes, actually remembers the nineties — which is why the massive capital expenditures on data centers, fiber, energy infrastructure, and low-earth-orbit satellites look like recklessness to Wall Street but look like foundation-pouring to anyone who watched what happened when the foundation was missing.
Mara: The post lists what that foundation requires now: purpose-built data centers, energy capacity, high-bandwidth fiber, fixed wireless, 5G and 6G, and satellite coverage for remote access. The upshot is that until most of that gap closes, scaling is impossible — but unlike the nineties, enough infrastructure already exists for significant near-term growth.
Pip: There’s also a forward-looking section on agents — software built in plain language, running autonomously — and a genuinely striking claim: that we may be approaching the first single-employee company worth a trillion dollars.
Mara: The post frames that not as hype but as a consequence of Moore’s Law applied broadly — every disruptive technology has shortened the distance between idea and product, and AI compresses it further still.
Pip: So the panic on CNBC every morning is, by this reading, just investors who haven’t learned to read infrastructure timelines — same leopard seal, different penguin colony.
Mara: The post closes with a reframe worth noting: a personal preference to replace “Artificial Intelligence” with “Assisted Intelligence” — as a way to accurately describe what the technology does now and, perhaps, calm some of the noise.
Mara: The through-line is infrastructure as the real story — not the money, not the hype, but whether the roads exist yet.
Pip: Next time, maybe we find out whether the roads are being built fast enough. Same feed, same ideas — one step ahead of reality.