What we can learn from bonfire of the dotcom companies: David Richards
I spent last Sunday clearing out my garage, which doubles as a museum of poor decisions, and found myself holding a stress ball with the Webvan logo on it.
There was a BlackBerry too, some Yahoo! mousepads (remember those?), an AOL trial CD, and a handful of other tchotchkes I had shipped back from California years ago and never opened. It was like uncovering a pharaoh’s tomb, except the pharaoh had burned through $375 million in an IPO and filed for bankruptcy inside two years.
Because that was the reality behind those logos. Webvan, Ariba, AOL: these were not garage start-ups trading on hope. Ariba hit a market capitalisation north of $40 billion. AOL merged with Time Warner in a deal valued at $165 billion. These companies had armies of staff, spreads in the FT, and the unwavering conviction of every analyst who mattered.
The clearest signal that things had gone badly wrong came at the Super Bowl in January 2000, where 14 separate dot-com companies each paid around $2.2 million for 30 seconds of airtime. Pets.com sent a sock puppet. The company was bankrupt nine months later. The sock puppet, I am fairly sure, survived.
That game is remembered now as the peak of something ugly: capital deployed not to build anything real but to buy attention, not investment but theatre. The revenues were not there, the margins were not there, and in many cases every transaction lost money. The business model was not merely deferred; it had never existed.
Take that Ariba t-shirt. At its peak the company was trading at something north of a hundred times its annual revenue, with investors paying for cash flows that never materialised. Cisco, the infrastructure darling of the era, peaked at twenty-six times sales. Now consider Anthropic: founded in 2021, already trending to £30 billion in annualised revenue, and trading at roughly 13 times that figure. I know which one looks like a bubble.
Lower multiple, real contracts, and customers who cannot easily walk away because if those systems stop, work stops. That is not hype. It is dependency.
Will there be casualties? Absolutely, and probably more than the optimists expect. Some companies riding this wave are burning capital at a rate they will not survive, and when they fold, the commentary will be brutal. But the infrastructure beneath them will outlast them, just as the fibre and server racks of the dot-com era eventually became the plumbing for Amazon and Google.
That is the real divergence. In 2000, when the money ran out, what remained was debt and office furniture. Today, the capital is going into data centres, TPU clusters, and sovereign compute. The dot-com companies were buying visibility. The AI companies are buying power, in the most literal sense, and the contracts wrapped around that infrastructure are not easily unwound.
An enterprise that walks away from its AI stack does not just lose a tool; it loses the operational capacity those systems have quietly replaced. That is a very different kind of lock-in from anything we saw a quarter century ago.
British businesses should be sitting with that distinction rather than dismissing it. The question is not whether some AI companies will crash and whether the press will call it a bubble. The question is whether you are woven into this infrastructure or priced out of it when it matters.
The dot-com era punished the speculators and rewarded those who understood what was actually being built beneath all the noise.
I dropped the whole lot in the recycling bin, the Webvan stress ball, the Yahoo! mousepad, the Ariba swag, all of it. Those companies are long gone. The lesson, though, is not. This time the capital is not chasing attention. It is building control, and whether Britain ends up on the right side of that is, as ever, entirely down to us.
David Richards MBE is managing partner of Yorkshire AI Labs