What does a GB of Internet service really cost? The worst case scenario (2011)
blogs.howstuffworks.com
blogs.howstuffworks.com
How about the core network? High capacity core routers? Firewalls? Datacentres? Power? Optics? Engineering? Ongoing upkeep? Monitoring? Upstream bandwidth? Interconnects? Peering LAN membership? LIR membership?
In many ways their naivety isn't too surprising, we started an ISP last year (https://telcom.io) and the shear scale of the undertaking and costs baffels us everyday!
Peering? Peering cost almost $0. I've had friends who had gigabit fiber brought in & they peered with people for next to nothing.
Having said that some ISPs (especially ones that own the infrastructure) have a habit of increasing prices (if they can from limited competition; i.e. roll out costs are high so not too many companies roll out to a region) and/or leaving customers to rot on old infrastructure (in order to maximise return on sunk cost; makes the CFO happy).
EDIT: clarification.
If anything, congratulations on living in one of the few areas in the US that actually has broadband competition.
Or we go with the simple answer and split last mile and transit into different services, but that requires government intervention.
If they switch to usage based billing, they can literally print money. They won't need to upgrade infrastructure for the foreseeable future because the stifling caps will keep people from fully utilizing what they pay for. Forcing video services/games/music/etc. to not create new services that require more bandwidth.
It's an awesome snowball effect for Comcast, and a giant middle finger to the users.
The reason not to is that the production plays a huge factor in those things, and people use similar amounts so folding infrastructure into cost doesn't distort prices very much. With bytes, neither of those are true and pricing gets weird and often bad.
...because the cost of running the network doesn't significantly change whether it's idle, or at full capacity?
...because -unlike electricity, water, and sewage- we have an infinite supply of bytes?
So is the marginal cost of each electron or H2O molecule.
> ...because the cost of running the network doesn't significantly change whether it's idle, or at full capacity?
You have to have extra capacity to deal with surges. Plus you probably have a normal distribution of bandwidth among your users, with some serious outliers. Why should they pay the same as someone who uses it less often?
> ...because -unlike electricity, water, and sewage- we have an infinite supply of bytes?
To start with, transferring data uses electricity, so I don't follow your logic. And there is certainly not infinite bandwidth.
Given that you totally misunderstood my second statement, I'm not surprised that you don't follow my logic. The difference in cost to run a wired network with a given capacity -whether that network is being utilized at 0% or 100%- is effectively zero.
> So is the marginal cost of each electron or H2O molecule.
Again, you misunderstood my second statement, so it's not too surprising that you'd say this.
You have to find and expend fuel to run electron pumping equipment. The cost of each unit of fuel is non-zero. You have to find and treat H2O. Most power plant fuels, and all H2O is a finite resource.
We don't have to find and refine byte caches buried deep within the Earth. Bytes are free. And -again- a network costs the same whether it's pumping 0 bytes, or all of the bytes it can.
The reality is privates companies want to increase their profit, they figure (and have convinced folks like you) that this is a fair way to increase prices.
The reality however is over time it enables price increases for everyone.
Of course, that's not what we're talking about. We're talking about private companies making up their own caps and their own cost structures so that they can push to the 30%+ profit margins seen in the wireless industry.
It's up to ISPs to opt into it. My ISP happens to and the streaming quality is quite good. However my costs do not decrease.
Problem is there are large swaths of the population that don't live anywhere near these metro areas and the communities and economies they participate in are nowhere near as dense.
Take the upper mid-west for example: Chicago, Minneapolis and Detroit are relatively easy (cheap) to wire up, but what about all the 15,000 population towns in between? Or the other 25% of the population that don't live in cities at all. In these places the options are most often between traditional 56k dialup, or satellite.
As everyone else in this thread is saying, the last mile is where the real costs are at, and in America we suffer from both the physical reality of being less dense and the political paralysis that allows organizations like Comcast to flourish. Even if we did fix the political problem, physics is still a bitch.
Or are you telling us that 25% of the US population doesn't have these things either?
Why does Chattanooga, TN have gigabit symmetric FTTP service for $70/month, [0] when San Francisco, CA, New York, NY, or those cities that you mentioned don't have anything comparable?
Hint: It's not a population density problem. It's a political problem. Internet service over municipally-owned infrastructure is so cost effective that the incumbent players have work very hard to make it illegal in many districts. [1]
[0] https://epbfi.com/internet/
[1] If you find yourself thinking that this is a reasonable move, ask yourself: "Why do we deliver electricity, water, and sewerage service over municipally-owned infrastructure, but balk at doing the same for Internet service?"
The community network remains self-funded and profitable, has repeatedly used annual profits to upgrade bandwidth, and remains significantly cheaper than the commercial cable and DSL broadband providers.
Remember there are a lot of paid shills. It's pretty cheap to wage a "information war" in order to prolong good gouging times.
Shame it has to happen at cost of rest of the economy.
Second, you forgot to factor in the costs of bad management (half kidding).
- We are an ISP, lets rebuild every major app in existence, put our name on it, and deploy it as bloatware.
- We are an ISP, lets acquire a bunch of content and IP so we can provide unique content to our customers.
- We are an ISP, lets acquire company x that has nothing to do with our core services and infrastructure.
Bottom line: ISPs love wasting effort, resources, and time on products and services their customers don't want, and continue neglecting their core services that customers do want improvement on.
Comcast, when I was a Cable TV subscriber offered Video On Demand. With it I could watch about 5 episodes of up to 10 different television series complete with commercials for free. Or I could elect to be able to watch a movie within 24-hours of purchasing for $4.00 or more.
They basically did half of what their internet based competition did for a higher price with more limited viewing options and a $10/mo fee for the box required to use it. So I plucked down $35 a piece for a few low-end Rokus, purchased a Netflix subscription, used my existing Amazon Prime and eliminated my cable TV service all together.
If there was competition in local broadband it makes you wonder if these companies would all go belly up when you look at how terribly their products stand up in markets where there is competition.
The last-mile is also the most expensive part of any cable. You aren't allowed to just tear up people's yards and run cables through them, you have to work with the local government (or Home-owners association) to get approval. Said approval process may cost money, etc. etc.
The most expensive, regulated, and complicated, part of network delivery is the last mile. And that's the real reason why local monopolies like Verizon and Comcast exist... except in the cases where Google buys out an entire municipality's fiber line and then gets cozy with the local government.
Thanks for that. I haven't laughed that hard in a week. :)
http://www.slashgear.com/google-buys-fiber-internet-system-i...
Google would not have bought an entire municipality's local fiber for so cheap unless some shady deal was going on. IIRC, Kansas City's deal required Google to wire up the schools with 1 GBPS internet for example (definitely a "local government deal").
I have no idea what backroom deal Google made with Provo, Utah, but something similar HAD to happen for Google to win the fiber for only $1.
Politics is politics, no matter the players.
Oh god. You're great. :)
"Google will take over Provo’s fiber network and improve upon it as soon as the acquisition is official ... which will reportedly cost Google about $30 million. But Google will also take over Provo’s remaining construction loans, which reportedly need to be paid for another dozen years at the very least, according to the official agreements released by city officials on Thursday.
Google will also be covering residents’ costs, as Provo households have reportedly been paying $5.35 a month on top of their utility bills regardless of whether they used iProvo." [0]
Another thing you're forgetting: iProvo was owned by Provo, UT. Public utilities have a different set of incentives and obligations than private corporations. This means that deals like "Hand over control of the municipal fiber network to a company in exchange for that company assuming all related debts, as well as performing massive upgrades to said network in short order, on their own dime." make lots of sense, if you can trust that the company will live up to their word. I would be shocked if there wasn't a provision in the agreement that would force Google to hand the network back over to Provo if they failed to meet their obligations.
Google has proven that they're serious about delivering gigabit symmetric connections at very reasonable prices to residents in the areas that they've been able to enter. The incumbent telcos have demonstrated time and time again that they're very committed to high-speed Fiber-To-The-Press-Release deployments, as well as token FTTP deployments in politically important areas... and nowhere else.
> ...Kansas City's deal required Google to wire up the schools with 1 GBPS internet...
Which is a perfectly reasonable stipulation. If you're not going to require universal service (just as ATT, Comcast, & etc. are not required to provide for data services), it's entirely reasonable to demand that a private operator provide service to things like schools and libraries to operate in an area.
> Politics is politics, no matter the players.
You really need to look at the vast gulf between the games that incumbent telcos play (e.g. banning municipally owned and operated fiber service in the name of creating a "level playing field") and Google's acceptance of public-good service requirements in exchange for reduced pole attachment fees.
[0] http://www.ibtimes.com/google-fiber-utah-why-provo-sold-its-...
I would say that accepting the debt of the local municipal government is by definition, "getting cozy with the local government".
Or would you say otherwise?
Do you say that the acquiring company is "getting cozy" with the company that's divesting itself of those assets? If not, why not?
[0] Say, a stagnating project that the divesting company can no longer fund at a level required to ensure the project's success in the marketplace.
Fishy. Unless I have complete and utter trust in the larger company.
Consider this: Google is the only company right now that has substantial web traffic, and can very easily violate Net Neutrality by making say... Youtube Google-fiber specific. Even if Google doesn't do this, the fact remains that if they continue to grow in the ISP space, their monetary power and political power will grow exponentially.
Or, Google can play the tactics of the previous ISPs by degrading their competitor's traffic (such as slower Bing, Yahoo, or DDG performance on their network).
Its innately fishy from my point of view. Google gains too much power for a very low monetary investment.
Okay. I guess you don't pay attention much. This sort of thing is not entirely uncommon and frequently entirely aboveboard. :)
> Google is the only company right now that has substantial web traffic...
What? Netflix pumps twice the traffic of YouTube. [0]
> Or, Google can play the tactics of the previous ISPs by degrading their competitor's traffic...
Backbone providers (of which ATT is a huge one) can facilitate this sort of thing right now. I wonder why they don't. ;)
> Its innately fishy from my point of view. Google gains too much power for a very low monetary investment.
You... uh... missed the part where Google takes on the obligation to perform $30 million in near-term improvement, the obligation to pay iProvo's outstanding construction loans for at least the next twelve years, the obligation to cover the $5.25 per month per Provo, UT resident currently paid by each Provo resident.
Google is putting a lot more than $1 into this deal.
[0] http://variety.com/2015/digital/news/netflix-bandwidth-usage...
http://archive.sltrib.com/story.php?ref=/sltrib/news/5620658...
> But Provo taxpayers will still have to pay off a $39 million bond that the city originally issued to build the network. With interest, taxpayers still have to pay $3.3 million in bond payments per year for the next 12 years.
> Provo Mayor John Curtis also revealed Tuesday that the city will have to front additional money that was not mentioned before.
> The city will have to pay about $722,000 for equipment in order to continue using the gigabit service for government operations already using the network, such as the operation of traffic lights and police and fire services. Meanwhile, Google will lease the network to Provo city for free for 15 years.
----
> CenturyLink, an Internet service provider, also sent a letter to Curtis asking for a 30-day delay on the vote, claiming the city did not give ample time or opportunity for other competitors to submit bids for the network.
-----------
Your information also contradicts the official Provo website: http://www.provo.org/about-us/current-issues/google-fiber
Looks like your standard no-bid municipal takeover to me. Socialize the costs, privatize the profits as they say. A serious offer would have had other ISPs at the table to counter-bid against Google.
Of course, politics never changes. Google got cozy with the municipality, shut out the competition, made the city pay for its own loans, and is going to privately profit off of the fiber the city built.
Uh. I'm not twisting anything, bro. My information came from the International Business Times article that I linked to earlier.
I also expect that you haven't read the franchise agreement entered into by Provo and Google [0] or the Fiber Lease agreement [1] or the Asset Purchase agreement [2] or the Network Services agreement. [3]
If we read through those (and I strongly encourage you to do so), we discover that the only thing that the International Business Times got wrong was the claim that Google was going to be taking on the financial liabilities of Veracity Networks LLC (the previous company that was in charge of running, managing, and maintaining iProvo) [4]. I could see why they mixed that up... the language in the Excluded Assets section is slightly tortured.
> A serious offer would have had other ISPs at the table to counter-bid against Google.
Are you basing this on more than just the single complaint from CenturyLink?
Look. Google has proven that they're both able to and serious about deploying gigabit municipal networks and pricing access to those networks at a rate that kicks the shit out of anything any other commercial ISP offers.
Read the official agreements. You'll see that they're all bog-standard. Actually dig in to Google Fiber's performance and track record (rather than just skimming the froth from some pundit's clickbait). Do some research and discover that what Google is doing today should have been done in the mid 1990's [5], except nationwide. Most of the entities who are scrambling to counter or stop Google (and -generally- municipally-operated fiber) now are the ones who reneged on those promises decades ago. (But were more than happy to retain the incentives and tax breaks.)
Another thing. Remember the fact that iProvo was sold to another private company five years before Veracity and Provo decided to let Google Fiber manage, maintain, and upgrade iProvo. [6] In your opinion, how does this situation [7] compare to a city changing the specialist company to whom they contract out power plant operation and maintenance? :)
[0] http://www.provo.org/home/showdocument?id=2298
[1] http://www.provo.org/home/showdocument?id=2300
[2] http://www.provo.org/home/showdocument?id=2296
[3] http://www.provo.org/home/showdocument?id=2304
[4] I bet you weren't aware that there was a commercial entity running Provo's network before Google Fiber came in. ;)
[5] Albeit at 40->100mbit symmetric.
[6] http://archive.sltrib.com/story.php?ref=/sltrib/news/5176637...
[7] Namely, Provo switching the manager, maintainer, and upgrader of iProvo from Veracity Networks to Google Fiber.
The city of Provo spent roughly $39 MILLION on building out their local, municipal fiber. And then they gave the network away to Google.
> You'll see that they're all bog-standard.
There is nothing standard about giving away MILLIONS OF DOLLARS worth of fiber away for $1.
The only thing "standard" about this, is that Google, a for-profit company, accepted the deal. Obviously, because gaining a huge amount of infrastructure, debt-free for a single dollar is massively beneficial to a company with loads of capital, and allows Google to expand into the ISP space.
> Another thing. Remember the fact that iProvo was sold to another private company five years before Veracity and Provo decided to let Google Fiber manage, maintain, and upgrade iProvo. [6] In your opinion, how does this situation [7] compare to a city changing the specialist company to whom they contract out power plant operation and maintenance?
Simple. Broadweave actually paid more than $1 for the deal. Much less fishy. In fact, Broadweave (later Veracity) paid a total of $40 Million on the deal.
Still fishy of course: Broadweave's debt was owned by the city of Provo, which is how Provo got into the mess in 2013 where they are still paying for the fiber they gave away to Google for $1. As far as I can tell, the city is just incompetent and grossly underestimates the value of their home-built network.
I can only imagine what Kansas City had to do to get noticed.
That’s why you put PVC pipes to every house and run the cables through them – to replace the cables you just have to tie the new one to the end of the old one, and pull the old one out.
In either case, the private resident doesn't own stuff that literally is being built under his front lawn. Such is local politics, and the rules and regulations change from city-to-city, county-to-county, and from state-to-state. And just studying the laws of one area may require a lawyer...
But there is so much money to be made by restreaming that video every time...with ads. Plus how could Google enforce deleted videos or marking then pirate if there was a local cache.
The price of bandwidth is cheaper than telecoms are letting on, yes. But there is so much that can be done to streamline the existing architecture as well.
Obviously the cost of bandwidth is a lot less than $50/month. But even at 1/10th that figure, it's a big chunk of the variable expense of the service.
No, because the $50 was almost entirely profit. You would have to subtract a much smaller number.
Also bandwidth costs drop dramatically over time as we build better fiber endpoints.
And in the real world Netflix will connect directly to your network at a cost of $0. They'll come all the way into your datacenter for free.
Edit: Okay, you changed your post around a bit, let me respond to that.
>Obviously the cost of bandwidth is a lot less than $50/month. But even at 1/10th that figure, it's a big chunk of the variable expense of the service.
The ISPs offer different plans. There's usually a minimum plan that's something like 1-2mbit. Just to reach the tier where you are capable of pulling down 10mbps, you have to pay an extra chunk of money, and that covers the cost of equipment and domestic bandwidth all by itself.
If those minimum plans are sold below cost, that's their problem, and they should stop gouging at the high end to make up for it. If those minimum plans are not sold below cost, then something isn't adding up.
What's going on? The nature of fiber deployment is that most of the cost is getting wires out to a neighborhood. Once you're there, hooking up additional households is relatively inexpensive. Moreover, most municipalities will not let you just wire up the neighborhoods you want. To get permission to wire any of them, you have to wire most or all of them. Once you've wired up a neighborhood, it makes sense to offer lower-priced plans, because a customer signing up even at below your average per-house cost is better than one not signing up at all.
Now, you could fix this by not having different prices for different service tiers. But that isn't what maximizes your revenue. And when you're working with a single-digit operating profit margin, you need to figure out ways to maximize your revenue. Services like FiOS wouldn't exist if it not for "gouging" on high-margin stuff like Triple Play packages, just like airlines wouldn't exist without "gouging" on first class seats and extra bags.
[1] http://www.verizon.com/about/file/3755/download?token=8QwymX...
I'd be much more interested in seeing what it really costs to build an operate an ISP with broadband service to the home (whether it's fiber or copper)
We also have good data from Veizon's public filings. Their wireline ARPU is $125 per month. Their EBITDA margin is 23%. Thus it costs about $96 per subscriber to operate their network, no including capital costs or interest.
Anecdotally, a few months ago, there was a transformer-caused fire a couple of blocks from my home that burned a telephone pole, disrupting power & internet service to a few city blocks of single-family homes. Crews were re-routing power as soon as the fire was out, and the ISP crew worked through the night to replace the pole and restore internet service.
So the service targeted at these price-insensitive customers gets them more regular status updates during outages, dedicated account managers, etc. that your residential customers really don't care about(their boss isn't going to chew their ass out when their personal Netflix goes down, but they will if the business and its 400 employees shuts down). The residential customers only care about the value/cost - they don't buy redundant backup internet, run their own automatic backup power generators, etc.
Businesses are like first class and consumers are coach on the same flight.
The biggest problem in much of the USA is duopolies (and the DSL side of that typically sucks so it is effectively a monopoly).
10/10 would be a customer again.
Obviously they're both pretty terrible compared to how other countries and companies do it (no caps at all), but I certainly prefer his solution over Comcast's if I'm forced to choose.
We didn't "plan which movie we were going to watch tomorrow", we just marked them for download (on locally-adapted forks of P2P software like eMule, with time/IP range filters) whenever we found an interesting one, and then we'd have 20-30 to choose from when we actually felt like watching a movie. And we could still watch a different one, it was just somewhat more "expensive".
It was far from a dystopian scenario, and frankly I wouldn't mind returning to that model if it could save me 20% on my bill.
I don't know about surge pricing, but I'd definitely join an ISP that just offered uncapped off peak speeds.
On the cost side, I suspect the cost of support, service and customer acquisition significantly affects the cost/GB. I used to switch between UVerse and Comcast every 6-12 months (the point at which the promotional prices ended) because it was very easy to do so and they didn't have a requirement that I remain for a contractual period of time.
However, even given all of this, "what we pay" is barely related to "what it costs", at least in most of the US. In the majority of markets there is often only one service provider that provides service at speeds in excess of 25Mbps. Sometimes, there is two, but in the state that I live in, the choice is often "Cable Provider" vs "Phone Provider". If you're lucky, your phone provider is AT&T on the newer network (with those giant boxes in each of our subdivisions -- no FIOS here) and they can deliver more than 25Mbps (inbound) and the other is Comcast at up to 100Mbps. Both have caps (unless you are a business subscriber) with AT&Ts at 150GB or 250GB depending on the service and Comcast at 250GB. A small number of areas have an additional cable provider (usually Wide Open West) that offer reasonable speeds and I think WoW doesn't do caps but I'm not sure.
The price you pay from each of these providers will be different depending on how many competing providers exist in the area. AT&T U-Verse at much higher speeds costs less in my house than AT&T U-Verse in my family's cottage up north (it tops out at 12Mbps with 150GB caps).
Until there is competition offering uncapped service (and customer demand for it), it doesn't really matter what the cost is. The folks at these companies charge what they can get away with given the market they're operating in, which in many markets is either an absolute monopoly (me, up north with AT&T) or them sharing the market with a single competitor who is able to deliver faster or slower service than them with identical data caps. The cost for entering a residential market is high, riddled with regulation and practical concerns, and the companies have lobbied hard for restrictive state laws that prohibit municipalities from providing a competing service.
Prior to them setting up this whole "pay for unlimited service" arrangement, I ran afoul of the 250GB cap and immediately switched to Comcast Business. The difference is stark. I had some service done with AT&T that resulted in one of the AT&T guys putting a shovel through the Comcast Business wire at 4:30 in the afternoon. I called them up and they sent someone to my house at 6:30 PM. I also get a special phone number, not open 24-hours, but I rarely talk to more than one person about the problem I'm experiencing and the hold times are minimal. He fixed the service (at no charge -- despite it being a competitor that caused the problem). That was one experience, but I've called support a few times and I've always had my issues addressed quickly. When I was a residential customer, waiting several days for a service call was normal.
I pay more than twice as much as the residential service at $130/mo for the same speeds, but I have no data cap and routinely use 2-3 TB/mo according to my router. I'm on the same wires, using the same company, paying a lot more, and getting no caps, and what I'd characterize as exceptional customer service. This almost irritates me more. It's not an issue of being unable to provide this kind of service to everyone, or an issue of network capacity, it's an issue of being unwilling to provide this kind of service to everyone because the market dictates that they don't need to in order to be (very) profitable.
Now, I suspect that the small businesses that they target this service at probably don't use the kind of bandwidth or have the performance requirements that I do (I'm in the 2-3TB/mo range). My dad's 9-person company was happy with IDSL at 1Mbps bidirectional until two years ago and would probably still be on it if Comcast hadn't wired up the park and undercut the other providers by half, so they're competing for a different kind of customer in small business. And this may bean that the small business market actually has more competition since many are willing to accept slower service offerings as identical products (or maybe the manufacturing non-tech heavy small businesses in my area are just that way).