Google Fiber is dead.
The effort to bypass phone and cable monopolies with fiber cable, begun a decade ago amid great fanfare, has been quietly sold to a private equity group.
Yes, it’s a bummer. Despite sitting on a Google list in Atlanta for over a decade, I never got an invitation. (I think it’s because my house sits so close to Decatur, which declined Google’s terms.) I’m also looking at an AT&T bill that’s twice what it was back then, for broadband service that’s no better.
Om Malik says Google Fiber failed because Google is a big organization that doesn’t care about its customers. That may be true. The bigger problem is that, in the 21st century, wires represent a failed business model.
I have been following this story for decades. Almost 30 years ago I wrote part of a cable industry book about DOCSIS, a plan to use fiber to increase the capacity of copper wire for “a fraction of what it would cost AT&T” to upgrade their phone lines.
The trouble is that “a fraction of what it would cost AT&T” is still a lot of money. It’s a capital expense. You get that money by borrowing it. Paying back money takes time.
The Enemy of Change is Time

When you build with cash, what you buy is a one-time expense. If you’re renting what you buy with cash, the second year’s rent is nearly all profit.
Contrast this with the first years of your mortgage. Remember that huge interest expense, and how little equity you got in the first years of the loan? That’s how AT&T built the phone network.
When you build with cash, you can rebuild the next year. Build with loans and you’re on the hook for the duration. This is fine if the asset’s value increases. But what if the asset’s value falls? What if it depreciates faster than you can pay back the loan?
That’s what has happened with phone lines. Copper phone lines were made obsolete by cable. Cable was made obsolete by fiber. Even fiber’s value can deteriorate, and there are costs for upgrading to higher speeds. Yet a phone company may still be paying off the copper.
The problem isn’t the fiber itself so much as the labor costs of running it, and then of maintaining it. AT&T is only now paying out the bonds it took to replace the raw wire that ran by my house in 1984 with jacketed wiring they installed in 1998.
Google thought it could bypass this mess. It couldn’t. Just as the cost of wires must be paid for slowly, they only pay off slowly.
You can’t run a tech business on a 30-year mortgage.
The Future is Wireless

It’s not cheap. But you’re only renting the towers, and you’re only buying the radios and switches. These get better every few years and, if you can make loans that fit those terms, you’re good. The costs of radio and switches have also fallen over time.
The only borrowing you need is for licensing the frequencies you’re using, and while those licenses sell for enormous sums, you can make that work, because each generation of equipment makes the license more valuable.
Consider WiFi. At the turn of the century a WiFi signal ran at 10 Mpbs on frequency bands, at 2.4 GHz and 5 GHz. The latest version, called Wi-Fi 7 or 802.11be, uses a third frequency band at 6 GHz but should run at 46 Gbps. That’s 46,000 times faster, using the same unlicensed frequency.
These same improvements are being found in licensed frequencies. It’s why SpaceX recently paid EchoStar $17 billion for frequencies it can use with Starlink, its satellite phone service. When my wife and I cruise the Atlantic next week, we’ll have practically the same Internet speed we have at home.
Thanks to wireless technology you can now get home Internet service for $30/month. It’s not a Gigabit, but it’s plenty for the video and social services most homes need. You’ll note that while Google is dumping its fiber, it’s keeping Google Fi, which re-sells T-Mobile capacity.
Google didn’t dump its fiber because it was greedy. It dumped its fiber because it’s a tech company, and because wires can’t deal with Moore’s Law of Radios.







