US researchers see Capex fall in cable ahead – We say probably not

Those with short memories tend to believe what executives from cable firms tell them, and this week there was the repeat of an earlier forecast from New Street Research, which said all the US cable giants would all drop their capex over the coming year or years and so will many in Europe.

It is true that their spend currently – especially the highly visible public big three, Comcast, Charter and Altice – has been inflated by the rollout of DOCSIS 3.1, and that this was far higher than had initially been expected.

DOCSIS 3.1 was introduced around 2013 as 10 Gbps up and 2 Gbps down technology compared to what had turned out to be the 100 Mbps up DOCSIS 3.0. The point 1 denomination was simply to suggest that it was hardly an upgrade at all, just going from 3.0 to 3.1. Some charlatans at the time suggested it was “just software,” which of course it was not, as not only did it require an entirely new concept of CCAP – a merged cable modem termination system with QAM – but it also needed CPE that was far more expensive and available from fewer suppliers.

The truth is that 3.1 was so far from DOCSIS 3.0 as to be unrecognizable. It didn’t help that the 100 Mbps speeds of DOCSIS 3.0 were so under represented by reality – the spec allows for 1.2 Gbps up, and 200 Mbps down – but in every instance of DOCSIS the headline speed is NOT what you get, the CPE allows a fraction of that speed, and what you get is what the cable operator can afford in its CPE.

So all the talk now of DOCSIS 3.1 upgrades being completed is nonsense. The CPE in almost all US homes will have to be upgraded to get more than the initial 300 Mbps speeds out of DOCSIS 3.1.

The specification of DOCSIS 3.0 allows for traffic from 54 MHz to 1,002 MHz, but no CPE was ever built which used all that bandwidth. And so too in DOCSIS 3.1 the available frequency in the standard goes from 54 MHz to 1,794 MHz, but once again CPE that can tune all those frequencies will only be rolled out over a very, very long time. The start point is to get to 1 Gbps. It also has to be understood that channel arrangements are totally different in this technology design, with 50 KHz channels multiplexed into 3,840 sub-carriers using OFDM – an arrangement totally alien to the remaining DOCSIS world. A CPE theoretically can speak to up to 4 of these clusters of sub-carriers, resulting in around 7 Gbps of real speed, and the theoretical upper limit of 10 Gbps. We shall of course never see this.

Comcast is the only operator pushing this at 1 Gbps, and this is only recently, and on a small number of customers, so the idea that capex for all this is virtually over, is pulling the wool over investors’ eyes. At best equipment pricing is coming down.

So there is some truth in it. Once you have built out the CCAP, to support either DOCSIS 3.0 or 3.1, you have perhaps eliminated 33% of the capex for the time being, but the CPE will always, in any network, make up more than 50% of the total build out spend. And as fresh CPEs are needed for faster speeds, this will be a constant drain on the cable firms for the foreseeable future until DOCSIS 3.1 is maxed out.

The truth is that already, as 1 Gbps offerings are fairly common in the US, built mostly around fiber to the home or fiber to the distribution point, offered by both telcos and cable firms. So this next round of more advanced CPE will be on the way fairly soon. Cable has always ruled the roost in boasting about broadband speeds, but AT&T in particular with, and both Verizon and others with fiber and with fixed wireless connections, will all overtake anything but fully configured DOSCIS 3.1 and in two directions, not asynchronously dominant in one.

What else is on the way? Well at both Comcast and AT&T new WiFi regimes which require multi-Access Points (APs) are arriving. Comcast plans to make it a retail experience to extend WiFi to run off approximately 3 APs per home, but even then the home gateway will need at the very least a software upgrade. However, Comcast has just changed the WiFi provider from Qualcomm to Quantenna, which offers far more expensive chips, and this will have to run through the system for quite a few years yet.

DOCSIS itself is about to go through yet another upgrade, which is to the Full Duplex version of DOCSIS – so once again, new CPEs and then both faster fiber backhaul for the CCAPs, and in fact faster CCAP, oh and did we mention Virtual CCAP. That is a ten-year cycle right there.

The truth is that Comcast in particular has spent a lot of money lately on X1 rollout and on its Homespot WiFi for cellular. It did each of these fast because it felt it was at a disadvantage not to have them. X1 meant that OTT delivery to the home TV was possible alongside QAM delivery and it did that because it quite rightly anticipated that skinny bundles would begin to eat its lunch unless it could offer them directly to the TV. The Homespots WiFi was to avoid being sidelined by a quad play and to add wireless. So it spends either to prevent revenues evaporating or to lift revenues.

That should be absolutely no problem, in that the key thing investment analysts are concerned with is not the absolute number for capex spent, but the ratio of revenue earned against capex.

So when Comcast spends more on its cellular network next year, this will not unduly worry critics, as it should bring in more revenues, to offset it.

What Comcast was saying at its results is that if its revenue goes up by almost 5%, as it did this year, that its “capital intensity” will be down 50 basis points or roughly 0.5%. It is this expression that has got everyone re-doing profit forecasts for Comcast, and then for all US and some European operators. It means it will spend half of one percent less on capex, as a percentage of revenue.

The table shows that if cable revenues grow 5% and the % of revenue capex spend falls to 14.6% from the current 15.1%, then Comcast will still spend 8.07 billion on cable capex, but this is lower per penny than it brings in, and so much (some) of this will fall to the bottom line.

This is assuming that opex on things like content does not rise too rapidly, something that Comcast has some control over through ownership of NBC, but companies like Charter and Altice have less control.

The truth is that X1 was accelerated because although it does not stem the flow of viewers leaving Comcast’s pay TV services, it did slow it down. And so that spend was vital. But there is more work to do here – Comcast will need fresh investments in deepening voice control experience on the X1, but this is not capital intensive and then it is duty bound to try to extend that voice control to its home automation strategy – something that will likely be left out of its cable revenues and put somewhere else.

Executives confirmed this as the Q4 results meeting, saying “CPE equipment including X1 and wireless gateways remained the largest components of our capital expenditures, but declined for the full year.”

And “For 2018, spending on customer premise equipment is expected to continue to decline. With X1 now deployed to nearly 60% of our residential video base, the pace of our rollout is started to slow. On the other hand, our spending on our network will continue to increase. As a result, we expect cable capital expenditures overall to increase in 2018.”

All of this is a drop in the ocean compared to how much the company plans to return to shareholders – it returned $7.9 billion in 2017 made up of $2.9 billion in dividends and $5 billion in share repurchases.

Charter in its results announcement said very clearly that its capex will continue to grow until past 2019, because it is in a very different place to Comcast, and it has no X1 up its sleeve.

In Q4 its capex was $2.6 billion up by $682 million year on year and again it cited higher spending on DOCSIS 3.1 CPE, scalable infrastructure and support. But it says it bought ahead for 2018 and has lots of CPE in stock and that will slow capex this year.

Altice is also in a very different place from Comcast, with its capex spend in 2017 $36 million more than its $955.7 million in 2016, but 2018 is going to be a huge step up to $1.3 billion.

Neither Charter nor Altice has made anything similar to the X1 investment, nor the voice remote control progress, that Comcast has made, and we suspect they are 1 year behind on 1 Gbps broadband from DOCSIS 3.1.

The simple truth is that the real gains for Comcast are to be found in content ownership, and its foray into international business, aiming to buy Sky in Europe, will consolidate further content ownership, and gain Europe-wide distribution and brand overnight. Its future is about broadband in the US, but about content everywhere else, and that doesn’t show up anywhere as capex.

But for the rest of cable, skinny bundles will dilute cable pay TV revenues, and spending on accelerating broadband becomes their only value add.