Today we’ll look back at the dot-com bubble, and forward to a world of AI:
Pets.com And The Dot-Com Bubble
In the wake of the dot-com bubble that burst in 2000-2001, Pets.com became something close to a poster child for the silliness of that era. If you view a CNN slideshow from 2010 (when people were still making slideshows), Pets.com is listed as the biggest of the dot-com flops. That reputation was never quite shaken: in 2021, a YouTuber called it “the biggest stock market failure ever”, and two years later another said that Pets.com was “the face of the dot-com bubble”. Those are just results from the first page of Google; there is much more coverage and commentary available with a similar tone.
To be blunt, nearly all of that coverage is wrong. Not a single one of those linked characterizations is anywhere close to true. Pets.com was not the biggest flop in dot-com. By the standards of the era, Pets.com didn’t even raise that much money: its February 2000 initial public offering brought in just $82.5 million. To name just one example, Buy.com’s IPO that same year generated twice as much capital, and that company went under nearly as fast. Peapod, Webvan, TheGlobe, Commerce One and many others had far higher valuations before all eventually hitting zero.
So the company isn’t one of the biggest failures of 2000, let alone in the entire history of the market, nor was it ever considered representative of the dot-com bubble. Its IPO was just three weeks before the bubble’s peak; shares barely even cleared their offering price before collapsing.
Why Pets.com became, in the public mind’s anyway, the epitome of dot-com-era foolishness is not entirely clear. Perhaps ironically, part of the reason is likely that the company had a tremendously effective marketing strategy that left an imprint on the public; its sock puppet mascot was interviewed on Good Morning America (back when that was a big deal) and starred in a $2 million Super Bowl commercial. That marketing was effective enough to make the company remembered, albeit not effective enough to make the company successful.
What is funny in retrospect is that the image of Pets.com immediately after the bust was as a frivolous company (granted, the sock puppet may not have helped here) with no real chance of success. Again, in the public’s mind Pets.com was part of a ‘bubble’ — and a bubble occurs when buyers (whether of shares of stock or Beanie Babies) pay prices that are completely untethered from reality.
But Pets.com — which as the name suggests, sold pet supplies to retail customers — actually had a decent business model. That business model was so effective, in fact, that a decade or so later Chewy was founded on essentially the same exact model. After fourteen years, Chewy now is worth just shy of $20 billion; its annual sales this year should clear $12 billion, meaning it accounts for something like 10-12% of the entire U.S. pet supply market.
This same dynamic plays out over and over again when looking back at the late 1990s and early 2000s. Dot-com darlings Webvan and Peapod (both online grocers) went under; Instacart is now worth $12 billion, and Uber and DoorDash (combined valuation: $260 billion) are in the grocery delivery business as well. Back then, there was Living.com and Furniture.com; now there is Wayfair and Overstock.com (its corporate name is now ‘Beyond’). TheGlobe.com was one of the biggest bubble winners and then losers; it was a social media site along the lines of Myspace and Facebook, both of which would later be valued in the tens of billions of dollars, and it launched an Internet calling business that presaged Skype (acquired for $8.5 billion) and Zoom (current market capitalization over $24 billion).
Yahoo! at the peak was one of the 100 most valuable companies in the world; that would not necessarily have been that outlandish had the company executed better (or succeeded in subsequent attempts to buy both Google and Facebook). Online currencies Beenz and Flooz look like Bitcoin if you squint. There were a lot of businesses that, like Pets.com, had good business models — and went bankrupt in a hurry.
Why Dot-Coms Failed
Each of those businesses had specific reasons for failing; none executed particularly well. Other dot-coms did survive and thrive (most obviously Amazon.com, but also eBay), even if their share prices cratered after March 2000. ($1,000 put in AMZN stock at its lowest point, during the third quarter of 2001, would be worth nearly $700,000 today.)
But there is one broad reason why these business models failed in 2000 and worked in 2015 or 2022: these dot-coms were simply too early. The technology existed for Pets.com to run the same business then as does Chewy now. But the infrastructure of the early Internet was not sufficient (no standardized systems, no interoperability, etc.) and more importantly customers weren’t ready. Here’s how Matt Stamski, analyst at Gomez Advisors, an early e-commerce research firm, explained the demise of Pets.com right after it shut down in November 2000 [emphasis ours]:
…the e-tail pet stores have not offered a compelling reason to shop online. Although delivering pet food and supplies directly to consumers is a convenience, that benefit is outweighed by the fact that the consumer has to wait days to receive their orders, Stamski said. Considering that pet food is available at just about any neighborhood grocery, few people have a reason to shop online, he said.
In 2000, the very idea that items had to be shipped — that, perish the thought, a customer would have to wait days for cat food — was seen as a hindrance to shopping online. And not by some coupon-clipping grandmother, but by an analyst whose job was to cover a nascent industry that would eventually transform global commerce.
There is always a gap between what technology can do and what its customers will allow it do. Consumers were not ready in 2000 (or 2005) to shop online for groceries or pet food, or to transfer their own cash into digital money, or to make phone calls over the computer (that’s what phones were for). The same was true of the automobile (see this piece from the Saturday Evening Post in 1930 mocking the skepticism toward the car), and the telephone, and probably fire too. It takes time for people to change.
What Can Artificial Intelligence Do?
"Cancer is cured, the economy grows at 10% a year, the budget is balanced — and 20% of people don't have jobs." That's one very possible scenario rattling in his mind as AI power expands exponentially.
That quote comes from an Axios interview with Dario Amodei, the chief executive officer of Anthropic. The artificial intelligence startup has developed a number of large language models named Claude, competing with the likes of OpenAI (where Amodei previously worked) and ‘Big Tech’ players.
Amodei expressed that prediction (which he believes will arrive in a matter of years) as part of a broader interview with Axios, in which the CEO expressed concern that lawmakers and citizens aren’t quite grappling with the potential of genAI. “Most of them are unaware that this is about to happen," he told Axios. “It sounds crazy, and people just don't believe it."
For those of us that saw the dot-com bubble first-hand, that sentiment sounds awfully familiar. 1999-2000 didn’t have quite the same sense of possible dread (or doom) around it; no one believed that Pets.com would eventually lead to the extinction of mankind. (There was the Y2K bug, but in my personal experience I don’t recall that being a big deal even before New Year 2000 passed without disruption.) But there were similarly huge predictions about massive, rapid, structural change across the society.
In e-commerce, respected advisory firm Forrester Research in 1999 projected $184 billion in online spending in the U.S. in 2004. That would have been a ninefold increase in five years; the actual figure turned out to be $65 billion, barely one-third Forrester’s estimate. The following year, Amazon founder Jeff Bezos told 60 Minutes that clothing, medicine bottles, and dinner plates would have embedded computer chips, with your plate even guiding you away from unhealthy meals. Bill Gates, the co-founder of Microsoft, that telecommuting would be widespread in five years1 (in fact, it took 20-plus years and a global pandemic, and now many companies are going the opposite way).
Generally speaking, the people within tech are going to make overly optimistic predictions, because their world centers on the capabilities of the technology, rather than the willingness of the consumer. That in turn suggests very real reason for skepticism toward Amodei’s prediction of a near-term apocalypse in white-collar jobs.
Who Is Hiring AI?
It is possible that Claude or ChatGPT will be able to replace white-collar workers in , say, 2028, particularly for certain tasks and/or in certain categories. But the decision to actually replace those workers will have to be made by people, not artificial intelligence — and those people are going to be much slower than the models will be.
Is a CEO really going to go all-in on AI the instant it’s ready? Or will she be more cautious? After all, once widespread cuts are installed, it’s very difficult to go back. Buy now, pay later startup Klarna has learned this lesson: it admitted last month that a shift toward AI-driven customer service had gone too far, and it’s now racing to hire human replacements.
In this case, the quality of the models seems to have been an issue, but the nature of the interaction clearly is also a factor. It’s worth noting Klarna’s CEO, Sebastian Siemiatkowski, is a co-founder of the company; the average corporate leader won’t have quite that level of job security. She will be in some cases risking her job if a shift to AI fails and leads to a loss of investor confidence and a lower stock price.
Similar calculations play out within the corporate structure. Will a middle manager push to remove the people he manages? Will he take the risk of AI making a major mistake — think accounting errors that screw up company-wide financials, or purchasing decisions that snarl the broader supply chain — that he will bear responsibility for? At some point, the answer is ‘yes’; but, here, too, that point only arrives after the technology is proven.
Customers will have to adapt as well. Even if AI on average is quicker in resolving problems, the tail of unfavorable (and brand-damaging) outcomes is likely worse; going in circles with a robot seems worse than doing the same with a human. (Think of how many complaints there are about customer service executed by offshore workers; they will be multipled in an AI world.) If a human is a moron, well, some humans are stupid. If the chatbot fails, than the company is stupid (and cheap). And for many customers, the very idea of a ‘chatbot’ is insulting and/or prima facie poor service; it will take a good deal of time for that to change.
And that will take longer than the likes of Amodei believe, because most companies (particularly outside the Silicon Valley bubble) are going to move more slowly than perhaps they should. The risk/reward of artificial intelligence might be positive for the company as a whole, but at the individual level the risks far outweigh the rewards. A salaried employee who successfully installs an AI model that displaces her colleagues gets a promotion and maybe a bonus and also the instantaneous, dreadful, realization that she is next. One who fails at any step of the process gets fired.
So the good news is that history suggests that AI is likely to not move as quickly as its proponents (and their media coverage) would have you believe. The bad news is that it’s still going to move somewhat quickly — and there will be consequences. The Internet ‘bubble’ wasn’t wrong — it was just early. AI, too, is coming.
As of this writing, Vince Martin has no positions in any companies or securities mentioned.
If you enjoyed this piece, give us a ‘like’ to both steer future content and to help us spread the word. Thanks for reading!
I’ve used this fact before, but the link used then is broken; readers will have to trust me on this one.