Everyone probably knows by now that the European Union (EU) is likely to decide that “big Tech”, meaning the content and social media companies, has to somehow subsidize the telcos. (If not, check this Reuters report from September 30, 2022, "Analysis: European telcos set to win fight with Big Tech, could set global agenda.") The goal is to offset the cost of the infrastructure they’ve had to deploy to carry the traffic that makes big tech profitable, to raise telco ROI on infrastructure high enough to keep the party going.
How did it come to this? Why hasn’t the innovation engine that gave us smartphones, personal computers with browsers, social media, the Web, and streaming content been able to give us a better, profitable, model of carrying the traffic all that stuff generates? Can’t we fix it?
The innovation engine I’m talking about is the “startup”, and the reason why startup companies haven't fixed the network itself is complex, and therefore hard to fix.
Big companies like network equipment vendors are traditionally risk-adverse, and for good reason. They got big serving a specific market, they’ve achieved profitability and brand recognition based on past products, and their biggest risk is likely that the value of their products will be eroded if some next-big-thing comes along. Let’s face it, if it had been up to big companies, our idea of networking could still be wireline analog voice, and the Internet still an obscure government project. With big operators and big vendors taking root and becoming trees, our industry has relied on new players, those startups, to inject innovation into a marketplace that could otherwise stall out.
Back in 1984 when Cisco was founded, networking was telephone switches. The founders were two university pioneers in computer networking, and Cisco was an implementation of principles of what became the Internet. A Cisco router at the time cost perhaps a tenth of what IBM’s network devices, called “SNA” cost, so it wasn’t hard to see the opportunity. Through the 1990s, there was a lot of startup innovation in the network space, but today VC money is focused on creating the next Facebook or Twitter --creating traffic but not so much the stuff that carries it.
How much innovation do we really need in social media? On the other hand, the networking technology that underpins social media and everything on the Internet could sure use some innovation. Why has VC and startup focus shifted away from “being the next Cisco,” a goal so often stated it became a cliche? I believe that VCs have contributed by succumbing to easy-money thinking. Social media startups are a combination of some cloud software and cloud services; these require less investment than building actual physical products.
But remember that what these VCs and social media startup founders want is an exit strategy. Could part of the problem also be that the big network incumbents, the big operators, are too conservative to be seen even as buyers of innovation? Yes, but could there be another problem? Yes, there could.
The real problem here is that we have an ossified network model. Networks today still adhere to the same overall structure they had thirty years ago, when digital voice replaced analog and the dollar value of a single order by operators for “Class 5” voice switches was more than Cisco’s entire annual revenue. We build networks in a hierarchy, with each layer concentrating traffic for greater efficiency. In fact, the term “Class 5” was first applied to the edge elements in that hierarchy; Class 4 was the next level, and so forth. We did this because bandwidth was costly, and you needed to concentrate traffic to improve efficiency.
Today, bandwidth isn’t nearly as costly. We can deliver fiber broadband to consumers for a fraction of what a single digital trunk with a tenth the capacity would have cost 30 years ago. Given that reality, why haven’t we asked a very basic question whose answer could define our best path forward: “Is capacity now cheaper than capacity management?”
Think about that one. Suppose that fiber bandwidth was really cheap. Build a network, or a part of a network, with cheap bandwidth, and what happens? One thing that comes to mind is that most of the complexity of traffic routing goes away. How much router code is dedicated to finding the “best” route? In a network with cheap capacity, each of the routes is essentially as good as any other. Another thing is that device complexity can drop. Instead of doing a lot of electrical-layer packet examination and handling, you can just create a series of wavelength hops on different fiber strands to get from Point A to Point B.
Then think of the impact of this on costs. Simple wavelength hops are a lot cheaper to create than complex router networks. You could expect fewer, simpler, devices, and that means lower costs for equipment. It also means lower operations costs, because generally speaking opex is proportional to both the complexity of the individual devices and the number of devices.
Why, then, aren’t we seeing a bunch of wavelength-hops-capacity-versus-capacity-management startups? A big part of the reason is that old obsolete network model. How do you build a network based on sufficient capacity versus one based on conserved capacity? Another piece is that telcos tend to be resistant to buying from startups, and would today’s network equipment giants want to acquire a startup who wants to change the game? Absent real revenues or a ready buyer, there’s no way for startup investors to make money on their deals.
OK, I admit that this sort of grand transformation of networking wouldn’t be easy. You might even need tax credits for the telcos to encourage write-downs of gear. But subsidies paid to network operators to carry Internet traffic? I can think of a dozen ways that could go wrong, including the operators using the money to buff up their bottom line and shifting the profit problem from the network part of the Internet to the content and experience part. But if we can’t do something technically constructive to prevent it, we may be headed to that.
How does an Internet driven by subsidies change the role of the innovation-generating startup? Or the roles of VCs? What is the future of the Internet – or our futures, as people whose lives are touched by the Internet? Wouldn’t a little network innovation be a better option?