> I can easily imagine it costing fractions of a trillion in startup costs, but these would be recouped easily if everyone switches over Hmm. Let's think about

> I can easily imagine it costing fractions of a trillion in startup costs, but these would be recouped easily if everyone switches over Hmm. Let's think about that. Let's say a "fraction" of a trillion is $240 Billion. There are ~100 million households who are potential broadband customers. If they pay $100/month, that's an income stream of $10 Billion/month or $120 billion per year. So you'd be repaid in 2 years! Well, unfortunately that's not all profit, so let's say 10 years. Still decent. But wait! There are already 100 million (actually 114 million) households with broadband in the U.S., and their average broadband speed is 124 Mbs. Yes, that's behind the world-leading Singapore at 200 Mbs (I wonder why that is..) but better than Sweden and about the same as France -- two other nations with population density closer to the US than to Singapore. So pretty much everyone who wants broadband has broadband with the exception of a small number of households in rural areas - you are talking about say, 20 million households. But there it's really expensive to lay fiber. Now those rural households will pay their $100 but that's an income stream of 24 Billion/month which will not justify spending fractions of a trillion. And for the other 120 million households, you are going to have a hard time getting them to pay more for faster broadband, as there are rapidly diminishing returns. Not many would be willing to pay $30/month more to go from 120 Mbs (US) to 200 Mbs (singapore). That convexity (diminishing returns) means when you begin your rollout, the incumbents can reduce their price by $20/month and you have to massively increase speeds to compete. Then they reduce their price by $10/month and you have to massively increase speeds even more. So you are really screwed now, with a trillion in debt, and at the complete mercy of competition that can steal your customers by spending a lot less money, in terms of reduced fees, than you have to spend in terms of increased infrastructure costs. That is why beyond a certain point, it doesn't make sense to compete on faster broadband. So I think the mistake you are making is thinking in terms of average revenue and average costs. But actually you need to think in marginal terms. Say it costs an extra $X to deliver a network that is 1 MB faster than the current network. But customers would be willing to pay $Y more for 1 MB of speed, or equivalently, would give up 1 MB in speed for savings of $Y. If X < Y, you have a winning proposition, but if X > Y, you have a losing proposition, and this is true regardless of whether the average costs and average revenue work out in your favor. And the problem is that Y is kinda small in a market that is close to broadband saturation, even if everyone always wants more speed and even if there are rural households that are left out.