[shittyanalogy]

Damn, one time $300 fee for "Up to" 5 Mbps download and 1 Mbps upload. That's like getting free internet for 6 and a half years[1] from current providers after 6 months. And if you want to upgrade you've already got the hardware.

I wonder what the minimum time of undercutting at a loss service they can offer is without it being considered anti-competitive.

[1] They guarantee 7 years of free service https://fiber.google.com/cities/austin/plans/#plan=free

[josu]

Is there any evidence that they are operating at a loss?

For 2013 Comcast reported 13.5 billion operating income on 64.6 revenue [1]. This means that they could potentially lower their prices 20% without incurring in losses. If we assume that Google is able to operate more efficiently, which isn't such a bold assumption, they could potentially undercut prices even more. On top of that, we could also assume that in terms of infraestructure the cost of offering 10mbps is simillar to offering 1 gbps.

So I'm not that sure that they are operating at a loss.

[1] http://files.shareholder.com/downloads/CMCSA/3686143613x0xS1...

[rayiner]

Comcast is reporting 20% operating income for a network that is largely already paid for. There's reasons to believe that Verizon isn't doing a great job recouping it's up-front costs on FiOS: http://www.businessweek.com/magazine/content/11_13/b42210461.... For the FiOS rollout, the capital costs alone are $1,000-2,000 per house: http://nypost.com/2010/04/01/why-i-think-verizon-had-to-pull....

There's no reason to believe that Google's per-household costs are any less. These companies all likely outsource to the same local companies that put fiber in the ground. To the extent that Google has an advantage, it's with uptake rates. Verizon and Comcast have to wire whole cities, pursuant to build-out clauses in local franchises. Google refuses to accept such clauses, and only builds in neighborhoods with demonstrated interest. The economics of building fiber make this a huge advantage. Most of the cost of building the network is in passing by the house, which is a cost you pay whether or not the house subscribes. If your uptake rate is 40% versus 20%, your per-house capital cost drops dramatically.

[pixl97]

How is that income figured exactly. I mean, doesn't every major movie seem to lose money even after raking in billions? I just don't trust any number from Comcast without a large volume of information on how those numbers were calculated.

And at that, that doesn't explain how Comcast even stayed in business till that point? Were they allowed to install their networks at a loss? Are there costs artificially high? Are they claiming TV programming costs in their figures?

[rayiner]

A better example is probably Charter, which doesn't have the noise of the media/movie business: http://phx.corporate-ir.net/phoenix.zhtml?c=112298&p=irol-re.... Look at page 54 of their 2013 10-K.

Their revenues were $8.15 billion (just over $100 per customer, for the whole year). $5.35 billion in operating costs, and $1.85 billion in depreciation and amortization (the loss in value of the capital in the ground). $925 million in net income on that revenue (11%).

What's really shocking is that their revenue per customer relationship is under $130 per year. That's the result of something else pundits ignore: cable company franchise agreements require them to build out to neighborhoods where many subscribers just get basic cable, which is usually subject to a regulated rate of just $10-20 dollars/month.

Looking at just the ISP business makes the numbers look worse. $4 billion of the revenue is from video, while $2 billion of the costs are for programming (see page F-10). If you take out the video revenue, you're down to $4.15 billion in revenue, with about $5.15 billion in expenses (i.e. you're losing a billion dollars a year).

Ultimately, of course, after paying interest on debt Charter lost money for the year. The idea that cable companies are wildly profitable is an HN myth.

[tanzam75]

> Their revenues were $8.15 billion (just over $100 per customer, for the whole year). ... What's really shocking is that their revenue per customer relationship is under $130 per year.

You seem to be an order of magnitude off on your ARPU calculations. $8.15 billion in revenue (p. F-4) / 5.9 million customers (p. 1) = $1,381 ARPU

Either that, or you're mixing up monthly with annual. On p. 4, they calculate their monthly revenue per residential customer relationship at $108.

However, you are generally correct that the cable business is a tough business to be in. Lots of labor costs, lots of capex, lots of interest costs. On an ROA basis, they're not doing that much better than the regulated utilities.

[rayiner]

You're right, I was looking at the Table on Page 4, which lists revenue per customer relationship at $107.97. That must be per month, not per year.

[digikata]

These numbers should also say something to free market advocates that dream of multiple companies installing redundant infrastructure to compete for customers.

[wmf]

I think the "myth" comes from attributing almost all costs to the video side which results in conclusions like $4.15B Internet revenue with minimal costs == all profit.

[tim333]

Google say they are not planning to operate at a loss. Re Kansas: "Several Google executives at the event were very clear that delivering gigabit internet access over fiber for $70 a month (and even free 5 Mbps fiber) is a business that will not only help advance Google’s consumer goals, but also make it money."

I daresay it depends on what assumptions you make about future sign up rates, cost of capital and the like.

https://gigaom.com/2012/07/26/the-economics-of-google-fiber-...

[gohrt]

Google may be factoring in their advertising revenue, averaged per website visited, or something like this. Not purely financed by connectioned fees.

[exelius]

Operating margin is a bad measure here. Net margin is a much better method because the telecom business is geographically constrained, meaning many of the tricks that technology companies use to avoid taxes (and thus boost profits) don't work in this industry. The revenue is generated in the US using assets that physically exist in the US, so it's a lot harder to dodge the taxes the way you can with IP assets.

Comcast's net margin is closer to 12-14%. And I would say that it's a pretty bold assumption that Google would be able to operate more efficiently than that: Comcast is a mature company in a mature market. Comcast's RF broadcast technology has some limitations, but it also has benefits: relatively low operational costs being a big one of them. Both Verizon and AT&T tried an architectural approach that is quite similar Google's, and their deployments stalled because of platform scalability problems.z

[wmf]

Everybody other than Google reports that installing fiber costs over $1,000 per home, but Google is selling it for $300 and saying "trust us".

[runeks]

You can't really generalize about per-installation costs. Digging fiber out to several houses is much more expensive than digging over to a 200-apartment complex, which lets you split the bill for laying fiber between 200 people.

In Copenhagen, Denmark, my apartment complex just got connected for 3800 DKK per apartment. And that's in a country with fairly high wages and 25% VAT. For similar apartment complexes (large ones, close to each other), in the US (where labor is cheaper and there isn't 25% VAT), I imagine Google can easily do less $500 per apartment, and possibly reach $300.

Provided Google doesn't have to pay a tonne in fees to get to lay fiber in the ground (which is probably why it's only available in select locations).

[adventured]

Labor is cheaper in the US? Can you provide some data on that, I understood labor was cheaper in Denmark. From what I've seen the median income in the US is about 20% higher than in Denmark.

[runeks]

Looks like the mean hourly wage for electricians is about 20% higher in Denmark compared to the US (avg. $25.75 vs 188.07 DKK for 2013) [1][2]

[1] http://www.bls.gov/oes/current/oes472111.htm

[2] http://www.def.dk/Loen/Loenstatistik/Loenstatistik.aspx (sorry, Danish)

[rhino369]

There are really two costs for installing fiber. The cost to drag it past your house, which is expensive, probably over 1000. And the cost to bring it from your curb to your house. I've heard 500 for that, but it could around 300. Or google could just think of it as an advertising cost.

[exelius]

The $500 figure is probably correct - I'm guessing Google is assuming a conversion rate of some of those customers at $300 turning into paying customers, and thus they'll get back part of the subsidy. 5mbit internet blows, and most people will probably want to upgrade.

[ericd]

I would happily pay that cost up front in exchange for a guarantee of being charged just the service cost afterwards. If only that was an option on the gigabit side.

[tinalumfoil]

> That's like getting free internet for 6 and a half years from current providers after 6 months.

Is internet in Kansas really that bad? I'm paying $50/month for 30/21 (I'm paying for 25/25 but oh well) here in New Jersey [0]. I can't imagine doing much with that internet besides basic web browsing. Even low-res video is choppy at that speed.

[0] http://www.speedtest.net/my-result/3952406450

[madengr]

Well, I'm in Overland Park, KS (KC suburb) and paying $120/month for 30/5 + static IP + phone. My provider (CCI) wired my area 10 years ago with hybrid fiber-coax, but within the last 5 years have used all fiber for up to 200/200. They'll never run run all fiber in my area unless pushed.

Both Google and AT&T are wiring Overland Park with 1000/1000, so that ought to put CCI and Time-Warner's underwear in a bind.

[mayneack]

Yes - My parents in mid-mo basically can't get anything faster than 5 down at any price. I know people who live outside city limits who can't even get that.

[JoshTriplett]

Once you've paid for the equipment, I suspect that price isn't actually a loss for Google. The bandwidth would be dirt cheap compared to gigabit ($0.35/month scaled down linearly from the gigabit rate of $70/month), and I wouldn't find it at all surprising if they make that much on the average Internet user.

[blueskin_]

The main issue is finding a way to make use of 5mbit. You get what you pay for, I guess, and if you really can't afford anything better, 5mbit is probably better than anything else other ISPs offer for a similar price.

[bluedino]

When AT&T offers 6mbs and calls it broadband, everyone complains and wants the FCC to change the definition of broadband.

[bnt]

As a person living and working with 600k-3mb/s - you can live with it fine.