How have the mighty cablecos fallen? That’s a critical question for the cable giants, of course, but it’s also a critical question for telcos and even for enterprises. Comcast, in their 3Q25 earnings, reported a loss of over a quarter of a million video customers, over a hundred thousand broadband customers. Charter lost about 70,000 video customers. In Massachusetts, cable subscriptions have fallen almost in half from their peak level in 2013. Remember that there was a time when telcos envied the cable companies, their CATV infrastructure that beat twisted-pair for broadband, and their stellar video revenue.
Video, streaming video over broadband is the problem, of course. Once you have an Internet connection that can comfortably carry multiple streaming video channels, the old model of linear TV is threatened. In particular, streaming attacks the “what’s on TV?” position, offering personalized content on demand. That means the linear RF model that cable was designed to support is far less valuable. It also means that broadband is far more valuable, and cable isn’t ideal as a broadband vehicle because of its shared-media model.
For the cable companies, one solution is to reduce the number of customers on a CATV span, so more bandwidth per customer is available. Another is to deploy FTTH in areas of high demand. A more cynical one, often adopted, is to cut deals with developers and/or communities to install cable infrastructure exclusively. But now, with FWA, those deals are meaningless. In any event, cable companies have a disadvantage here, in the form of “IRR”.
IRR stands for “internal rate of return”, and it’s a measure of how well a project can pay off for the initiating company, based on factors like the cost, benefit, cost of money, and so forth. IRR is related to a Wall Street metric that’s generally available, called “ROIC” for “return on invested capital”. If a company has a high IRR, it means that projects that have a constrained ROI will weaken it financially. If they have a low IRR, the very same project can actually improve their financials. If we look at Comcast, compared to AT&T and Verizon, we see that its ROIC is roughly 20% higher than the latter, and over 40% higher than the former. That means that anything Comcast, and in fact cable companies in general, do to improve their infrastructure is more likely to weaken their financial position. Anything, and that “anything” is important for two reasons; cost management and new services. That makes it important for AI.
When you apply AI to cost management, you have to reduce costs enough to offset the investment in AI. Your IRR/ROIC determines how much that is. Telcos, then, have a lower bar to meet than cable companies in harnessing AI or any other technology to lower operations costs. Is this related to the fact that, in general, customer satisfaction with cable companies is lower? Sure seems likely.
Same thing on the new-services side. Telcos can invest in infrastructure to support new services when ROI is lower than cable companies could tolerate. Even if both groups can make an investment, the telco will typically gain more from it, because the same return will boost the company’s overall ROIC more.
All of this is important when the easy add-on services like linear TV are falling away, and not just for cable companies. Some telcos used to offer their own channelized TV, but that’s almost never done today. Cable companies, as already noted, are losing linear RF video customers, and their infrastructure isn’t ideal for broadband, so many are losing broadband customers too. Cable has a lower “pass cost” than FTTH, but higher than FWA.
Is the fact that Comcast gained mobile customers a good sign? I don’t think so. Comcast offers very competitive mobile prices by bundling with their CATV broadband, and most of their services are MVNO rather than home-built. What I think you’re seeing with their last quarter is a sign that consumers in general aren’t happy with cable value. Some leave, and others bundle in mobile to make the service more cost-efficient overall.
Pushing bits, friends, sucks as a business for cable companies. It also sucks for telcos, just not quite as much, yet. Because, in no small part, bits are tough to differentiate in a competitive market.
One thing that this means is that there is enormous price pressure on consumer broadband, and a lot of work done to lower the cost per bit to the operator. Enterprises can benefit from this by using more consumer-broadband-related services to replace traditional business services like MPLS VPNs. I think that SD-WAN and SASE will ultimately utterly dominate branch-small-site networking. I also think that even for larger sites, we’ll see more and more consumer technology, like PON in fiber, applied to business services, so things like MPLS-router IP VPNs may become a backwater technology.
Enterprises can expect to see their network service costs fall as this shift in technology for business networks advances. That’s particularly true if you consider the cost per bit and not just the cost. It is already practical to give a branch office bandwidth that dwarfs what a headquarters site of the 1980s could have obtained, at one one-hundredth the cost. Given the available remote-site bandwidth, we can expect to see applications that could never have been afforded in those glory days of business networks service to thrive.
But who offers them? What technology will they be based on? What, overall, will the network of the future look like? Many of the new applications that enterprises want to run don’t look like today’s transaction processing at all. Meetings are often done not in rooms but via Teams or Zoom, and over the Internet. The applications whose potential excite enterprises the most relate to real-time, IoT-linked, holistic process management. The early example of these are hosted in the facility where the process is used. AI, in at least the most common current applications, is run in a data center and uses a WAN only to accept prompts (questions or instructions) and deliver limited-sized results.
Cable companies and telcos both face a harsh reality here, which is that on the one hand, consumers really don’t value anything they do except push bits, and on the other hand anything new that does just push bits is likely to stress the cost side of their balance sheet before it boosts the revenue side. Neither cable companies nor telcos seem very excited about offering any non-bit-pushing services, including AI hosting. Have video experiences bitten them enough to make them service-shy for things other than pure bit-pushing?
Telcos have whined about being disintermediated for at least thirty years, but if they insist in diving to the bottom of the value chain and taking root there, what do they expect? Business as usual means results as usual.
