Most of the non-SpaceX blocks are areas that have had terabits/sec of fiber running through them for the last decade as part of long-haul intercity links. Since the last mile(s) of rural fiber is so much cheaper than the last mile of urban fiber [1] this is a massive windfall for the telephone companies.
The FCC created some very perverse incentives with this auction. They actually made it more profitable for telcos to leave intercity fiber unused than to splice a strand to serve the neighborhood it passes through, no matter how little the splice would cost. This led many providers (cough cough CenturyLink cough) to refuse to use or even license their fiber until they could scoop up this cash first. If even one residence in the block has service above 25mbit/sec the whole block is taken out of the auction.
I previously submitted a much-less-useful version that was colored by speed tier (there are only four) here:
https://news.ycombinator.com/item?id=25368927
This updated map has all sorts of goodies if you click on the region... stuff like the dollar amount of the winning bid and so forth.
[1] 100% aerial, no trenching, no manholes, no ducts, uncongested utility poles, simple (two-dimensional!) property boundaries, terminate at the property line not deep inside a multitenant/condoized building. The conventional wisdom "rural internet is expensive" got turned on its head by the counterintuitive economics of ADSS optical cabling.
IDK in the US, but in my country rural internet is indeed expensive. Using old telephone poles makes it easier, but you still need network elements, pay people for deployment and it's very low density.
The most expensive part is paying construction workers if you have to bury cable and all the stuff that involves that.
Fiber to the home is only buried when the electrical lines are underground (i.e. no poles). That is incredibly rare in rural areas. And gigabit switches ("network elements") are stupidly cheap now.
People have been totally bamboozled about what this stuff really costs.
B4RN is a pretty typical British rural broadband offer, in a rural area there's no real problem with just digging a trench, dropping the ducts into the trench, then blow fibre through a completed duct. In a city if you dig a trench that's a huge pain, because the city is paved or built on everywhere, but in rural areas it's no hassle. With modern agricultural practices a relatively shallow trench is enough that the fibre won't be disturbed by subsequent farming activity, these days nobody churns up half the planet to grow stuff because we know that destroying soil structure reduces long term productivity.
And as to those "network elements" it's GPON, the elements are passive splitters, basically a fancy prism, completely passive. There's only bit of actual electronics in each region but it's quite a bit fancier than a "gigabit switch" because it's multiplexing the light for all the nodes behind the GPON splitters.
Post starts with a non-question followed by a question mark?
This map is a map of the USA. Underground electrical service is extremely rare in rural areas of the USA. We have these things called utility poles. And we use them.
On what planet do GPON prisms cost more than switches? The whole reason it got popular was that it was even cheaper than active optical. Here's a 32-way prism for $14: https://www.fs.com/products/11528.html . That's under $0.50/customer. You pay a bit more for the weatherproof enclosure for it that mounts on a pole.
Frankly, GPON is for losers. 144-count ADSS is so cheap now it's stupid to build GPON/EPON/BPON these days. Most of central Washington State (Grant County) has a dedicated strand of fiber to each household -- it's future-proof. Just do it. GPON is a scam by large telcos who are afraid that unless they have the whole neighborhood sharing something (i.e. a fiber) they'll be forced to let each customer choose a different provider for that customer's strand. I predict that we will see stupid telcos keep rolling out GPON even after it becomes way more expensive than dedicated dark fiber. It's greed and bamboozlement.
Not sure what urban fiber costs, but in rural the overall cost of Fiber-to-the-home comes out to at least $30-40k a mile for aerial construction, which is legitimately hard to make work if you have less than 10 homes per mile without some subsidy.
The cost is ~30% cheaper if you are a telephone company though and can overlash the fiber on your existing telephone lines.
Think about it: why would overlashing be 30% cheaper? It involves exactly the same materials (the fiber) and labor (one bucket-truck visit to each pole) save for a $5 bracket.
The only thing you save is an extra pole attachment fee. But that's a monthly running cost, not an upfront construction cost. Attachments are ~$1/pole/month in the 30 states with regulated pole attachment rates and in the same ballpark for most (though not all) jurisdictions in the other 20. Dunno how you got from that to $10k construction cost savings.
Sounds like you got quoted some fiber contractor's "I get to buy a new car" price. ISPs buy their own bucket trucks and hire linemen.
Even so, $250/home setup cost for fiber is a bargain. CenturyLink charges $200 setup for crappy 10mbit copper DSL and people still buy it.
> why the total RDOF grant closed at only $9 billion instead of the anticipated $16 billion
It closed low because SpaceX placed a bid in a ton of territorries that were expected to be monopoly single-bid areas. In the ones SpaceX didn't win, the winning bid was nonetheless forced lower by their competitive bid. The ILECs are furious. No pity, they can eat a bag.
Over the course of the next year the winners need to submit detailed plans for their networks to the FCC, including (for the first time ever) the "oversell" ratio (i.e. how many customers with "1gbit service" are sharing a 10gbit backhaul). So it could be up to a year before we find out exactly what sort of service each ISP is claiming fits into the performance+latency tiers they bid for.
The FCC did lay down some hard rules though: (1) no DSL technologies allowed in the "gigabit" tier and (2) no geosynchronous or medium-earth-orbit satellites in the "low latency" tier. SpaceX is classified as low-earth-orbit.
> In the ones SpaceX didn't win, the winning bid was nonetheless forced lower
This sounds like a really satisfying result. Do you have any sources on specific areas where that happened? I was kind of surprised that Starlink wins weren't more broadly distributed through more states, do you think the FCC was trying to spread the funds around between multiple ISPs?
If you read the rules carefully, it does not require fiber all the way to the home, so of course the telcos will use HDSL, ADSL etc. for the last kilometer. They lobbied hard for the 25Mbps broadband definition so that they can use the old copper plant. 25/3 is pretty easy to do with bonded DSL and they have plenty of pairs. I would not expect them to do a log of new build, most likely some fiber to a hut perhaps and then bDSL from there to the customer. Even doing it that way they will need subsidies for that to be profitable at the $30 price point. Cost to build in rural is a lot less than urban but there are more customers in urban, so rural needs government support for anyone to do it profitably.
Everwhere I've read said that if you bid on gigabit it has to be gigabit to each customer and there's a blog post linked in this thread saying that the FCC isn't even allowing gigabit DSL. Citation needed.
"we expect bidders seeking support for the 25/3 Mbps tier will win support only in areas where higher speeds
are not economical, and that a five-year term .."
Does anyone know why Hughes, a OneWeb investor, placed and won bids solely within the state of Rhode Island? Is this some sort of squatting play to qualify them for future solicitations?
Given the state of OneWeb, it seems very unlikely they will be able to deliver even across such a tiny geographic area.
Yeah there is a scheme where a bidder can "assign" their winning bid to a subsidiary. A lot of telcos are required to keep their local exchanges as unmerged separate corporations for regulatory reasons.
26 comments
[ 4.4 ms ] story [ 68.4 ms ] threadThe FCC created some very perverse incentives with this auction. They actually made it more profitable for telcos to leave intercity fiber unused than to splice a strand to serve the neighborhood it passes through, no matter how little the splice would cost. This led many providers (cough cough CenturyLink cough) to refuse to use or even license their fiber until they could scoop up this cash first. If even one residence in the block has service above 25mbit/sec the whole block is taken out of the auction.
I previously submitted a much-less-useful version that was colored by speed tier (there are only four) here:
This updated map has all sorts of goodies if you click on the region... stuff like the dollar amount of the winning bid and so forth.[1] 100% aerial, no trenching, no manholes, no ducts, uncongested utility poles, simple (two-dimensional!) property boundaries, terminate at the property line not deep inside a multitenant/condoized building. The conventional wisdom "rural internet is expensive" got turned on its head by the counterintuitive economics of ADSS optical cabling.
The most expensive part is paying construction workers if you have to bury cable and all the stuff that involves that.
People have been totally bamboozled about what this stuff really costs.
https://b4rn.org.uk/about-us/our-network/
B4RN is a pretty typical British rural broadband offer, in a rural area there's no real problem with just digging a trench, dropping the ducts into the trench, then blow fibre through a completed duct. In a city if you dig a trench that's a huge pain, because the city is paved or built on everywhere, but in rural areas it's no hassle. With modern agricultural practices a relatively shallow trench is enough that the fibre won't be disturbed by subsequent farming activity, these days nobody churns up half the planet to grow stuff because we know that destroying soil structure reduces long term productivity.
And as to those "network elements" it's GPON, the elements are passive splitters, basically a fancy prism, completely passive. There's only bit of actual electronics in each region but it's quite a bit fancier than a "gigabit switch" because it's multiplexing the light for all the nodes behind the GPON splitters.
This map is a map of the USA. Underground electrical service is extremely rare in rural areas of the USA. We have these things called utility poles. And we use them.
On what planet do GPON prisms cost more than switches? The whole reason it got popular was that it was even cheaper than active optical. Here's a 32-way prism for $14: https://www.fs.com/products/11528.html . That's under $0.50/customer. You pay a bit more for the weatherproof enclosure for it that mounts on a pole.
Frankly, GPON is for losers. 144-count ADSS is so cheap now it's stupid to build GPON/EPON/BPON these days. Most of central Washington State (Grant County) has a dedicated strand of fiber to each household -- it's future-proof. Just do it. GPON is a scam by large telcos who are afraid that unless they have the whole neighborhood sharing something (i.e. a fiber) they'll be forced to let each customer choose a different provider for that customer's strand. I predict that we will see stupid telcos keep rolling out GPON even after it becomes way more expensive than dedicated dark fiber. It's greed and bamboozlement.
Bamboozlers gonna bamboozle.
The cost is ~30% cheaper if you are a telephone company though and can overlash the fiber on your existing telephone lines.
Think about it: why would overlashing be 30% cheaper? It involves exactly the same materials (the fiber) and labor (one bucket-truck visit to each pole) save for a $5 bracket.
The only thing you save is an extra pole attachment fee. But that's a monthly running cost, not an upfront construction cost. Attachments are ~$1/pole/month in the 30 states with regulated pole attachment rates and in the same ballpark for most (though not all) jurisdictions in the other 20. Dunno how you got from that to $10k construction cost savings.
Sounds like you got quoted some fiber contractor's "I get to buy a new car" price. ISPs buy their own bucket trucks and hire linemen.
Even so, $250/home setup cost for fiber is a bargain. CenturyLink charges $200 setup for crappy 10mbit copper DSL and people still buy it.
It closed low because SpaceX placed a bid in a ton of territorries that were expected to be monopoly single-bid areas. In the ones SpaceX didn't win, the winning bid was nonetheless forced lower by their competitive bid. The ILECs are furious. No pity, they can eat a bag.
Over the course of the next year the winners need to submit detailed plans for their networks to the FCC, including (for the first time ever) the "oversell" ratio (i.e. how many customers with "1gbit service" are sharing a 10gbit backhaul). So it could be up to a year before we find out exactly what sort of service each ISP is claiming fits into the performance+latency tiers they bid for.
The FCC did lay down some hard rules though: (1) no DSL technologies allowed in the "gigabit" tier and (2) no geosynchronous or medium-earth-orbit satellites in the "low latency" tier. SpaceX is classified as low-earth-orbit.
This sounds like a really satisfying result. Do you have any sources on specific areas where that happened? I was kind of surprised that Starlink wins weren't more broadly distributed through more states, do you think the FCC was trying to spread the funds around between multiple ISPs?
"we expect bidders seeking support for the 25/3 Mbps tier will win support only in areas where higher speeds are not economical, and that a five-year term .."
support for rural telcos IMO
That's not the gigabit tier. Read it again, paragraph 39 (it's a nice table).
Then look at the map. Hardly any money went to the "minimum tier" you're referring to (25/3).
That is called defaulting. There is an entire section about how the FCC claws back all the money and then some more for penalties.
It does for the gigabit tier.
Given the state of OneWeb, it seems very unlikely they will be able to deliver even across such a tiny geographic area.
Only the OneWeb sats at LEO can deliver on the latency requirements.
It will be interesting to see what happens as the FCC reviews the "long form" applications.