Brian Roberts & cablers ready to guarantee 100+ Mbps upstream to > 85% of US in 2-3 years

Brian Roberts has made damn he has an answer to the 12 million more lines of fiber home AT&T is building. Comcast covers almost half of the US. It will offer upstream 100-150 Mbps (85 MHz spectrum) or 300-450 Mbps (204 MHz.)

Fast upstream DOCSIS 3 is already shipping for field trials and will be widely deployed (85-90%) in a few years.

DOCSIS 4, which will offer a gigabit upstream, will come later but is irrelevant to this discussion unless you believe people need a full gig upstream in the next few years. That’s absurd.

The cost to upgrade is about $200/home, about $22 billion for the country. That’s about 4% of cable sales in the four years. It will probably be recouped just in increased fees for the higher speeds. The customer churn reduction will be even more significant.

It is a clear decision for the cablecos to upgrade and does not require a subsidy. They will move quickly.

#2 Charter & #3 Cox have similar plans, They represent about 89% of the 90-95% of the US covered by cable, over 80% of the nation. #4 Altice, about half of the remaining cable homes, will also upgrade to 100 Mbps upstream. including some fiber.

The smaller cablecos face similar economic and competitive factors and are likely to make similar decisions. Cable will almost certainly offer 100+ Mbps to about 90% of the country.

An additional 3-6% of homes are already built or funded for better than 100/100. including through RDOF and NTIA. 5G wireless will bring 100/20 to some of the remaining homes. T-Mobile committed to 99% 5G coverage as part of the Sprint merger agreement. Most, but not all, can achieve 100/20.

The result is that 93-96% of American will be offered at least 100 Mbps down, 100 Mbps up in a few years, not including a new infrastructure program. Many of the rest will have 100/20 through wireless. LEO satellite will pick up a few more but has capacity for fewer than ~ 3 million and possibly < 1 million. (More data needed.

It has not escaped my notice that the projected broadband coverage means that no infrastructure spending can significantly improve the broadband available to more than 4-7% of American homes.

DOCSIS 4, in the labs, can deliver upstream over 1 gigabit and downstream to a nominal 10 gigabits. It has other features that will drive deployment but its unlikely many customers will require more than 100-500 Mbps upstream.

Understanding the Mbps & MHz on cable.

Common coaxial cable reliably supports about a gigahertz of spectrum, divided between broadband and video. Current systems allocate about 25 MHz to upstream, running on low frequencies that were unused before cable modems were invented.

The higher upstream speeds of DOCSIS 3 use 85 Mhz and 204 MHz. That requires improved equipment and changing how the video signals are carried. But it does not require digging and new connections so is much cheaper than fiber. About $200/home compared to about $500-1500/home for the majority of fiber customers.

That much spectrum can carry a certain amount of data, which is shared. The predictable upstream is 100-150 Mbps (85 MHz spectrum) or 300-450 Mbps (204 MHz.)


Astute readers will notice I have not provided a primary source for this figure. It is from one of the most respected cable engineers. I;ve known him to be reliable over 15 years. His company does not allow him to go on the record.

I am seeking out senior sources whom I can name. I believe I understand enough of the technology to confirm $200 is in the ballpark

Talk to me

Scroll to top