Capitalism is defined by booms and busts. However, I do think the dot.com and now the AI booms prove that humans learn some things from the past. The railroad boom after the civil war had a similar dynamic in overbuilding the supply but a lot of those routes were poorly built and made no sense for rail networks. Hence, a lot of those tracks were removed or needed to be redone. Dot.com was built on bad business models/assumptions but atleast the infrastructure didn’t need to be redone. AI is now being led by the strongest companies in world history with massive investments in infrastructure with the demand for that infrastructure still outpacing the supply. This can be seen in pre-leasing of new developments by both co-location and neocloud providers. Makes one wonder how early we are in the boom phase during this cycle
Demand in this new phase is a tricky thing to measure. A lot of the "demand" is from investors, not necessarily downstream users (even companies). I think we have a situation where companies are deliberately overinvesting as a competitive tactic. The aim is to "win" the so-called AI arms race and be the one company that gets to provide models while most other companies will "lose." We may have a pretty painful correction if this pans out like that. Not sure what that may look like or when it will happen, but the most crucial infrastructure is not just chips or data centers but energy infrastructure right now. Those dependencies---with the utility companies and investors---are where the real risks to the public lie in my estimation.
I think most is owned in some way---probably consolidated within Zayo Group, Lumen Technologies, and Crown Castle, although I don't think there's a lot of available data on this. I have read that there's a lot of dark fiber from the 1990s and 2000s that is used for private networks. Location matters though. A lot of fiber is not in prime locations.
In my personal opinion, I think that something very important was happening layers up in software and operating systems that actually laid and defined that pathway. During the 1980s, 1990s were when mass software and then the internet were being digested into the structure of civilization, but that digestion was through hidden, exclusionary politics. The only public-facing arguments were about civilizationally inconsequential technicalities, like, is MSDOS really an OS or middleware? is IE part of Windows or a separate product? While the real decisions about distribution, licensing, and lock-in were hammered out by IBM lawyers, Microsoft executives, and regulators behind closed doors. The result was a kind of private constitution for the digital age, written without democratic input. And not just Microsoft specific, everything from infrastructures layout decisions to all types of legal and regulatory matters as well.
Had this digestion happened under the older American democratic republican paradigm, with mass member political parties, civic structures, public deliberation, policy variability, deliberate redundancy, etc the internet might have become a federated digital commons instead of a set of corporate silos. Local ISPs and software guilds could have flourished like earlier cooperative utilities and credit unions, embedding the network into civic life. The Old Academe and various types of civic institutions might have treated the web as a public knowledge infrastructure, not a private advertising platform. Instead, as you show and as the software story shows institutionally, the advent of the digital age was absorbed through narrow interest group politics. In my reckoning, hat’s why by the 2000s the paradigm was already locked, Microsoft on the desktop, a handful of firms defining the web, and later Google, Apple, Facebook, and Amazon consolidating the internet. We were left not with a civic project, but with a special interest groups carve up
Excellent piece, very informative.
Capitalism is defined by booms and busts. However, I do think the dot.com and now the AI booms prove that humans learn some things from the past. The railroad boom after the civil war had a similar dynamic in overbuilding the supply but a lot of those routes were poorly built and made no sense for rail networks. Hence, a lot of those tracks were removed or needed to be redone. Dot.com was built on bad business models/assumptions but atleast the infrastructure didn’t need to be redone. AI is now being led by the strongest companies in world history with massive investments in infrastructure with the demand for that infrastructure still outpacing the supply. This can be seen in pre-leasing of new developments by both co-location and neocloud providers. Makes one wonder how early we are in the boom phase during this cycle
Demand in this new phase is a tricky thing to measure. A lot of the "demand" is from investors, not necessarily downstream users (even companies). I think we have a situation where companies are deliberately overinvesting as a competitive tactic. The aim is to "win" the so-called AI arms race and be the one company that gets to provide models while most other companies will "lose." We may have a pretty painful correction if this pans out like that. Not sure what that may look like or when it will happen, but the most crucial infrastructure is not just chips or data centers but energy infrastructure right now. Those dependencies---with the utility companies and investors---are where the real risks to the public lie in my estimation.
Fascinating piece of map ! Do you know if all dark fiber invested during this time was bought and is being used ?
I think most is owned in some way---probably consolidated within Zayo Group, Lumen Technologies, and Crown Castle, although I don't think there's a lot of available data on this. I have read that there's a lot of dark fiber from the 1990s and 2000s that is used for private networks. Location matters though. A lot of fiber is not in prime locations.
What would have been have we seen a nationalization of the fiber 😏
In my personal opinion, I think that something very important was happening layers up in software and operating systems that actually laid and defined that pathway. During the 1980s, 1990s were when mass software and then the internet were being digested into the structure of civilization, but that digestion was through hidden, exclusionary politics. The only public-facing arguments were about civilizationally inconsequential technicalities, like, is MSDOS really an OS or middleware? is IE part of Windows or a separate product? While the real decisions about distribution, licensing, and lock-in were hammered out by IBM lawyers, Microsoft executives, and regulators behind closed doors. The result was a kind of private constitution for the digital age, written without democratic input. And not just Microsoft specific, everything from infrastructures layout decisions to all types of legal and regulatory matters as well.
Had this digestion happened under the older American democratic republican paradigm, with mass member political parties, civic structures, public deliberation, policy variability, deliberate redundancy, etc the internet might have become a federated digital commons instead of a set of corporate silos. Local ISPs and software guilds could have flourished like earlier cooperative utilities and credit unions, embedding the network into civic life. The Old Academe and various types of civic institutions might have treated the web as a public knowledge infrastructure, not a private advertising platform. Instead, as you show and as the software story shows institutionally, the advent of the digital age was absorbed through narrow interest group politics. In my reckoning, hat’s why by the 2000s the paradigm was already locked, Microsoft on the desktop, a handful of firms defining the web, and later Google, Apple, Facebook, and Amazon consolidating the internet. We were left not with a civic project, but with a special interest groups carve up