An optimistic brief historical perspective
As a close observer of and longtime participant in the tech industry, I, like many others, am skeptical of the current rush to build datacenters. Tech industry executives love huge, expensive infrastructure projects. Not only do the executives make personal fortunes on these projects, they also get to stroke their egos with the wealth and power they control as huge physical infrastructures are built out. The temptation to launch these projects is near irresistible to these mere humans and their financial backers are dazzled by the flashy jargon these salesmen sling.
I’m skeptical, but also optimistic. Investment in infrastructure can be serendipitous.
In the late 1990s and early 2000s, huge investments were made in expanding the computer network. The web of buried optical cables, satellite connections, and network routing equipment from Cisco, Comcast, old Bell Telephone System derivatives, and other network providers grew rapidly. At the time, infrastructure investors were betting on profits from the new World Wide Web without thinking enough about business models to extract profits from their investments that eventually would be written off when the “dot-com bubble” burst.
However, after the burst violence, business models—like targeted advertising and online sales—appeared that were profitable and the Internet boomed, all enabled by the dot-com investment in network infrastructure.
The computer cloud of today began as a computing power waste reclamation project by Jeff Bezos and his team at Amazon. For Amazon to become a retail success, even in the early days when they sold only books, their sales website had to respond quickly to retain the attention of their impatient and convenience-seeking customers, especially during peak usage hours, usually mornings early in the week.
For customer-satisfying response times, Amazon had to deploy server capacity to support rapid response during peak hours, but this extra capacity idly accumulated costs and yielded little return. Amazon faced increasing wastes while scaling to ever larger numbers of customers.
The Amazon team were brilliant: they sold their idle computer time to businesses that could use it during off-peak hours without investing capital in their own computers. This business and technical model, cloud computing, was born when Amazon Web Services (AWS) was launched to recoup some of Amazon’s investment in peak-hour servers.
AWS eventually became a substantial revenue center in the Bezos empire. The cloud made the rapid expansion of online business possible. Capital investment for a computing startup sank and innovation soared.
Cloud computing requires a robust computer network, a barrier that the massive dot-com bust investment in physical network infrastructure had already removed. Fortunes were made, including for Zuckerberg, Bezos, and Musk. Technology executives have been on the prowl for similar opportunities, perhaps to the point of obsession with AI datacenters.
Predicting the future is more like conjuring a tempest than directing a laser scalpel, but I am optimistic that the current investment may prove useful in ways that we cannot imagine today.