Fiber's operations and infrastructure will be sold to one of the major telecoms within the year (just like Google did with Motorola once it ceased to be useful). My money is on CenturyLink since Google's approach meshes well with their overbuilding strategy and footprint.
It's a 0-5% margin business for the foreseeable future; and best-case scaled scenarios it's a 10% margin business with huge capex. Google shouldn't be interested in anything with less than a 20% margin at scale. So from a portfolio standpoint, it was a stupid bet.
My guess is that Google entered the business without knowing all the nuance of the access world. They made an assumption that there was some magic bullet of architecture and comp sci that could lead to game-changing economies of scale. But it turns out the telecoms are pretty solid on the engineering / design side, and customer service and network maintenance are the big cost centers. Neither of those can be automated with algorithms or AI (well, network maintenance sort of can be, but the telecoms are already on top of it), so there's not a ton of opportunity for Google to leverage its strengths.
Running a datacenter is a very different business from running access networks. Operations that are cheap in a datacenter are expensive in the field (cable runs, equipment swaps, etc.) Likewise, there are different considerations around who does the work -- it makes sense to hire full-time people in a datacenter; less so when your equipment is scattered over a few million square miles. All of that leads to a very different operating model, so it's no wonder that Google couldn't figure it out.
> It's a 0-5% margin business for the foreseeable future; and best-case scaled scenarios it's a 10% margin business with huge capex. Google shouldn't be interested in anything with less than a 20% margin at scale. So from a portfolio standpoint, it was a stupid bet.
I know that's standard financial thinking, because a business is assumed to be able to absorb pretty much any amount of capital and invest it at a rate of it's core rate, which for Google is indeed 20%. However, obviously, Google cannot do that. It can invest maybe 10-15 billion, but no more (presumably unless you want to make Google a low-margin business). So that doesn't appear to be true.
Is it better for Google to "invest" their 80 billion or so at 3% in liquid assets versus investing it in a 10% margin business ? Here's the thing: they'll make more profit with the money in Fiber than they'll make with the money in bonds and stocks. Definitely. If fiber provides a 10% return on investment, definitely.
Sadly I know: for the GOOG and GOOGL stock prices, yes it is a better idea to keep cash on the sidelines, as then you get the benefit of investors assuming Google can deploy cash at 20%, despite it being obviously wrong. But it offends my sense of efficiency.
> Definitely. If fiber provides a 10% return on investment, definitely.
Telecoms return about 10% at scale. Startup stage of a telco is incredibly expensive; and you typically won't see positive cash flow until a decade or so later (which is why it's usually financed by debt offerings; not equity or cash). Risk is also high; because if you enter the wrong market as an overbuilder, you can ensure nobody in the market (including yourself) will ever turn a profit.
All in all, overbuilding a telecom is a high-risk, low-reward investment with a very long payoff period. There's a good reason there aren't more companies lining up to do it.
It's a 0-5% margin business for the foreseeable future; and best-case scaled scenarios it's a 10% margin business with huge capex. Google shouldn't be interested in anything with less than a 20% margin at scale. So from a portfolio standpoint, it was a stupid bet.
My guess is that Google entered the business without knowing all the nuance of the access world. They made an assumption that there was some magic bullet of architecture and comp sci that could lead to game-changing economies of scale. But it turns out the telecoms are pretty solid on the engineering / design side, and customer service and network maintenance are the big cost centers. Neither of those can be automated with algorithms or AI (well, network maintenance sort of can be, but the telecoms are already on top of it), so there's not a ton of opportunity for Google to leverage its strengths.
Running a datacenter is a very different business from running access networks. Operations that are cheap in a datacenter are expensive in the field (cable runs, equipment swaps, etc.) Likewise, there are different considerations around who does the work -- it makes sense to hire full-time people in a datacenter; less so when your equipment is scattered over a few million square miles. All of that leads to a very different operating model, so it's no wonder that Google couldn't figure it out.