Yeah, because the only cost in broadband delivery is the undersea fibre??
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!
I hate comments like these because for how frustrating someone is saying they know so much more, you don't find any articles written by them, a cost breakdown (because no one wants to share how they do things) and/or no real data. Yes, from the ground up starting an ISP, the trenching, the datacenters & everything in-between cost monies, but relative to a company that has all of this in place and doesn't upgrade anything, it's a moot point. The reason why people complain about caps on anything (cellphone, land data or anything else) is that these companies are making _BILLIONS_ off of us & refuse to upgrade their core network because it's beyond 5 or 10 years old.
Peering? Peering cost almost $0. I've had friends who had gigabit fiber brought in & they peered with people for next to nothing.
Indeed, the naivety of this article is breathtaking. I speak as someone who has worked in telecoms, ISP, DSL and webhoster environments, this person has no idea what he's talking about.
Of course the missing item in that calculation is the last mile. Netflix could set up a cage at each landing point and the savings would be nearly 99%. And that adequately explains why international transit is rarely the biggest cost issue (obviously there are corner cases around the world where this is indeed the biggest issue). The biggest part of the cost in the infrastructure (systems, staff, maintenance) is from PoP to customer.
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).
The last mile is the vast majority of the cost, but it's not what ISPs talk about when they try to justify price increases and refusals to peer. They talk about the usage, even as that gets cheaper and cheaper. If we shut that rhetoric down, and get them to be honest, we're a lot closer to reasonable policies.
Or we go with the simple answer and split last mile and transit into different services, but that requires government intervention.
The only way to prop up their profit margins, in their eyes, is to talk usage. They could very easily provide everyone gigabit service, but then there would be no up-charge for the next 5-10 years as all the services catch up.
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.
> ...because the marginal cost of each byte is effectively zero?
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.
> To start with, transferring data uses electricity, so I don't follow your logic.
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.
By your logic we should also have tiered rates. Ie. Data in prime time costs more than off peak.
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.
It's not the worst idea, as long as the price is connected to the cost.
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.
If you're going to support regulating the shit out of internet and capping their profit margins, sure. We'll all be paying about $40/month for gigabit internet, I'm all about it.
Of course, that's not what we're talking about. We're talking about private companies making up their own caps and their own cost structures so that they can push to the 30%+ profit margins seen in the wireless industry.
It’s incredibly annoying when a big multinational ISP asks you to pay 120$/month for a connection with throttling, data caps, and blocked services of competitors, while a small, local business manages to provide the same for 30$.
This entirely ignores the part where we have 1000x the landmass. Providing internet in Moldova or Estonia is an entirely different physical prospect than providing it to Alaska or Montana.
Yea, if you want to treat all the major metro areas as islands this is a comfy little world view to hold against the evil cable corps, however, American cities are still generally less dense than their Europe or Asian counterparts.
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.
> Take the upper mid-west for example: Chicago, Minneapolis and Detroit are relatively easy (cheap) to wire up...
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]
[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?"
We have a largely suburban town, 20K population when we moved in 15 years ago, primarily single-family dwellings. The commercial ISPs wouldn't even consider providing broadband. Community network plans were made and implemented, then the commercial ISPs spoke up and protested. We had a community cable and broadband internet network built from scratch, fully deployed and running for a couple of years before the incumbent cable company with a pre-existing network (Comcast) started to provide broadband 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.
Yes, feel free to come up with excuses for your shitty Time Warner/Comcast monopolized market. What about Google Fiber? "Oh, they just want to sell more ads, that's why it's so cheap, yadda yadda". You Americans deserve your fate. Instead of starting a revolution, you come up with bullshit excuses for your oppressors. Instead of complaining about the NSA, you complain about businesses not going overboard to escape their surveillance.
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.
Too true. Every time I've had a guy out to install service I've given them one laptop to get connected to the service. When they're done, I hook up my router and put the laptop back in the cupboard. I've been using the same Core Duo Acer laptop for this purpose for years.
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.
What about the costs of network equipment? For all of the ports? What about the price of internet from all of the middlemen involved? If I created an ISP today, I would need to buy bandwidth from a Tier1 provider (like Hurricane Electric, or Verizon) and then resell said bandwidth to my customers.
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.
> You aren't allowed to just tear up people's yards and run cables through them,
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.
Who owns the PVC pipes? Does Verizon own them, or are they property of the Home Owner's Association?
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...
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.
When the original Google fiber city selection was going on I was attending school in one of the smaller sized competitor cities. It doesn't even take back room shady deals for this to be a thing; the administrators of all sorts of local government institutions were more than happy to publicly subjugate themselves to the will of Google just on the hope that it would sway their favor.
I can only imagine what Kansas City had to do to get noticed.
> 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.
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.
> What do you call it when one company acquires the assets and obligations of a portion of another company [0] for a token amount of money?
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.
> 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.
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.
> I don't know why you are twisting the story around, but I'm not inclined to trust you or your data, or your arguments anymore.
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? :)
Dude, we're looking at a deal where a city gave away MILLIONs of dollars worth of fiber cable away to a private company for $1.
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.
Centralized services are such bad models too. Imagine if YouTube came out with a Miro type player that allowed everyone to cache videos and sever then to others. No more waiting to buffer a video you've watched ten times. Google's sever usage would drop.
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.
So his bottom line is $50/month per customer for a constant 10 mbps. To put that into perspective, the operating profit margin of wireline ISPs is 5-20%. The low-end of that scale is for traditional phone/DSL/fiber. Assuming an average subscription of ~$100/month, that bandwidth is coming out of an operating profit of $5-20/month.
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.
I'll use Verizon as an example because their numbers are easily available. Verizon's ARPU for wireline was $125.32 per month in Q3 2014.[1] But their operating profit margin on wireline was only 2.3%. So if you've got a $70/month FiOS subscription, is that costing Verizon money? Probably!
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.
Too bad this analysis has nothing to do with delivering internet access to the home since the ISP typically just leases a few miles (or 10's of miles) of fiber to the local internet exchange point, rarely do they lay an 8700 mile undersea cable... and even if they do, they still need to build all of the infrastructure to get the network to your house.
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 have some good data on how much it cost Chatanooga to build their network. About $330 million total. For roughly 71,000 subscribers, that's $5,000 per household.
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.
How do you compute that? What the numbers show is that most of the revenue goes to ongoing operation and maintenance. They have to pay for the original build with what is left over. With respect to Verizon, analysts are skeptical they'll ever pay off the build-out of fiber.
This thread might be a good place to ask: Why is enterprise bandwidth an order of magnitude more expensive than residential bandwidth from the same company, delivered in the same way?
There are generally two ways you can approach some cost: Maximizing expected value or minimizing risk. Larger companies especially will focus most of their thought on their core product which they 'maximize value' on, and lump everything else into the 'mitigating risk' bucket.
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.
Because if the enterprise service stops working you'll have someone working on it and giving status reports on progress in hours. In residential you'll have someone look at it in 2-3 days if you are lucky.
Yes, though "days" is hyperbole. 2-3 hours would be the max, even if it were a single home. If many homes are out, the ISPs make it a high priority (or lose customers to providers that do).
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.
From what I've read and heard, the key is reliability, consumer internet service is not as stringent on reliability, but enterprise internet lines usually come with a guaranteed reliability.
What ISP do you have? Mine is Time Warner and it started with $35/month (after applying their first-time subscriber discount) >4 years ago. The first 3 years, it was 15Mbps, then it was increased to 20Mbps. But the cost eventually increased to $55/month in 2015 and by then, I cannot justify spending that much for an online access as a grad student (usually am not home), so I told them to downgrade me back to 15Mbps for $35/month just a few months ago. Overall, I do not notice much of a difference in speed decrease, but I wish there is more ISP available in my area other than Time Warner, who seems to have a monopoly here...
Moved to the Bay Area ~6 months ago, heard about Sonic and even though their service is slower and slightly more expensive than Comcast, they are an excellent company that I have 0 regrets about signing up for.
So in the last paragraph the author imagines a dystopian future where you have to plan which movie you are going to watch tomorrow so Netflix can pre-download it for you during the night, and for some reason he thinks that it would be a desirable scenario. LOL.
A decade or so ago, back when 256kbps was the average cable speed and there were caps for international traffic here in Portugal, our ISP had something called Happy Hours, which meant that all usage between 1-9am didn't count for the cap.
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.
Actually, he said that if you were to download during peak hours, it should cost ten cents per gigabyte. You'd have the option to download at no cost during off-peak hours. The alternative, what is really happening in some parts of the US today, is that you will be charged a dollar per gigabyte whether or not the line is currently congested.
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.
There's more than a few problems with this particular assessment as it relates to "what we pay" and what kind of service we can expect for that price.
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).
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!