There are many comparisons being made between the current AI buildout and the Internet and dotcom buildout that started almost 30 years ago and ended in a bust in the early 2000s. Ethan Mollick recently reminded me that Amara’s Law is a useful way to think about this comparison. That is, we tend to overestimate the effects of a new technology in the short term and underestimate its effects in the long term.
The dotcom and network buildouts that occurred in the late 1990s were significant. It was the start of my career, I witnessed the exuberance, and I saw many companies flourish and then disappear. In hindsight, the peak of that exuberance was being invited to a private Elton John concert for Nortel customers in a downtown Atlanta ballroom.
At the time, new online companies were using the web to set up new businesses that would completely disrupt the status quo. Telecom companies raced to build new networks using optical networking technologies that would move massive amounts of data around the world. Most of the companies that were part of this “gold rush” no longer exist or were acquired for “cents on the dollar.” But the investments that were made facilitated how the web is part of our lives today.
The birth of e-commerce
One example, often highlighted in these debates, is Pets.com, which launched in 1998 with a business plan for selling pet supplies online. Great idea, but this business model was not ready for the time, with the company losing money on every sale due to high costs of shipping, especially the heaviest products. As is also well documented, Internet penetration was low, advertising costs were high, and customers were not yet accustomed to buying online.
In terms of logistics and fulfillment, Amazon, which owned 54% of Pets.com at one point, was one of the companies that successfully solved that problem. In the mid-2000s the company launched Prime for customers and Fulfillment by Amazon for sellers, both leveraging a national warehouse and logistics platform it had built. A whole logistics industry for online purchases had flourished beyond the postal and courier companies of the dotcom era, but too late for Pets.com.
The infrastructure that followed the dotcom bust eventually enabled Uber, launching within a decade of the dotcom bust, with that business built on web connectivity, and online mapping and payments, all delivered from a mobile app. Uber was one of the companies that helped normalize app-based logistics, payments, and delivery coordination. Uber, DoorDash, and Instacart are further examples of how much the logistics economy matured after the dotcom bust, and the kind of maturation missing in the Pets.com era.
What these situations demonstrate is that transformational technologies need many other facets to be successful, since they are literally transforming the market. Chewy, launched in 2011, was often compared to Pets.com, but unlike its predecessor it entered into a mature logistics and e-commerce industry that did not exist a decade earlier. There are other factors, such as increased Internet penetration, e-commerce maturity and consumer acceptance, which have also been important for Chewy and the growth of the e-commerce industry.
Value creation
In hindsight, the contrast between Pets.com and Chewy highlights the ultimate value-add of the Internet and dotcom boom. The value was the modernization of the logistics industry, the efficiency of complete e-commerce transactions including returns, and the migration and creation of many online businesses.
For AI, there is significant discussion about how the massive spending will generate a return. At the centre of the debate is a basic question of where the real value is. Significant levels of automation and efficiencies are at the core of the potential return, but there are still questions about how these efficiencies will be implemented and adopted by organizations. Also, people are concerned about what displacement will occur, and vague references to re-skilling and future employment-producing innovations will not alleviate those concerns.
Where are we going?
Personally, I am already taking advantage of these efficiencies, I can point to concrete benefits, and I know more are coming. But I think large parts of society and the market are still figuring out how AI will be implemented and used productively.
There is no doubt that AI will be part of our world in the future, just as the Internet and dotcoms promised their future almost 30 years ago. But aside from the many dotcoms that failed, many of the network builders and equipment providers also failed. The fiber buildout from that time created multi-decade value, while the equipment and vendor-financing arrangements lost value in a few years and sometimes almost immediately, and the innovations, such as Uber, took many years to come to maturity.
The lesson of Amara’s Law is not that the AI buildout is fake. It is that timing, winners, business models, and useful innovations are much harder to predict than the existence of technological change. If the dotcom era is a guide, the value will accrue to those who best leverage the AI infrastructure to deliver real benefits to the largest number of users.
In closing, two pop-culture references come to mind: a Wendy’s commercial and the Jurassic Park movie. For the short-term AI buildout, the question is still “Where’s the beef?” while for the long-term impact, the warning may be: “Objects in mirror are closer than they appear.”
Sources
Mollick, E. (2024). Co-intelligence: Living and working with AI. Portfolio/Penguin.
Mollick, E. (2024, January 6). Signs and portents. One Useful Thing. https://www.oneusefulthing.org/p/signs-and-portents
Pets.com. (n.d.). In Wikipedia. Retrieved June 10, 2026, from https://en.wikipedia.org/wiki/Pets.com
Roy Amara. (n.d.). In Wikipedia. Retrieved June 10, 2026, from https://en.wikipedia.org/wiki/Roy_Amara
Inferential Loop 