Technology discontinuities can be very disruptive of business models, in areas such as capital investment burdens, for example. For cloud data centers, we might ponder how accelerated spending on graphics processor units could affect more-routine updates of servers supporting traditional operations.
Researchers at Omdia estimate that AI server capex will be 66 percent of total data center capex in 2024, for example. And cloud computing giants tend to increase capex spending substantially, year over year, in any case.
Between 2018 and 2022, annual cloud computing hyperscaler capex has grown about 30 percent a year, according to researchers at Costar. Omdia thinks that rate will be exceeded in 2024.
Of course, not all observers think the increases will be that high. Some think the increase could be closer to 10 percent, overall, with the caveat that some spending could be shifted from other areas of infrastructure to server purchases.
Also, since the spring of 2024, most of the hyperscalers have announced increases above the level noted by Dgtl Infra at that time.
Still, most observers would tend to agree that hyperscale cloud computing capex will increase in 2024 between 15 percent and 35 percent.
The point is that capex as a percentage of revenue is growing.
U.S. cable TV operators also face a discontinuity, namely a switch of platform from hybrid fiber coax to fiber-to-home networks, driven primarily by home broadband requirements.
And cable operators are likely to find the cost of switching from the existing access platform to fiber-to-home to be as costly as telcos have faced, as it means switching from one type of platform to another (copper to fiber for telcos; hybrid fiber coax to FTTH for cable operators).
For that reason, cable operators are likely to use any number of targeted upgrades whenever possible. The objective will be to delay a full “rip and replace” as long as possible. Near term, HFC upgrades will continue as long as possible, with incremental FTTH overlays in some cases.
But a full transition away from HFC for most households will be quite disruptive, on the order of what telcos have faced moving from copper to fiber access. Where upgrading HFC has cost something on the order of $200 per location, an FTTH upgrade might cost between $1,200 and $2,500 per passing.
That is an order of magnitude increase in capex.
So the cable operator transition from HFC to FTTH will be much more disruptive than the cloud computing hyperscaler investment in AI capex.
No comments:
Post a Comment