NFV (Network functions virtualization)

SDN, NFV Not Slashing Wireless Capex Yet

SDN and NFV have operators planning for lower capital expenditures, but that's not yet being reflected in their projections for 2014.

One of the big promises of NFV is the ability to buy lower-cost, off-the-shelf hardware and manage it with software. It's a money saver for the operators, but a potential threat to their equipment suppliers that rely on them to constantly upgrade their hardware. So far, however, it appears that lower capex is just a goal of the operators. (See Defining SDN & NFV.)

AT&T Inc. (NYSE: T), Verizon Wireless , Sprint Corp. (NYSE: S), and T-Mobile US Inc. have all wrapped up their fourth-quarter earnings, and on average, their wireless capex projections for this year are flat, none significantly decreased.

Table 1: Tier-One US Wireless Operators' Capex

2012 Capex 2013 Capex Projected 2014 Capex
AT&T $19.7 billion $21.2 billion (wireless: $11.19 billion) $21 billion
Verizon Wireless $16.2 billion  $16.6 billion (wireless: $9.4 billion) $16.5 - $17.0 billion
Sprint  $5.37 billion $7.45 billion $8 billion
T-Mobile $4.2 billion $4.3 - $4.6 billion
Source: operators' Q4 earnings reports

AT&T was the only exception, but with an expected capex decline of less than 1%. Last year saw it spend more money as it began its Project Velocity IP (Project VIP) build, but it's lowered projection for 2014 comes as it shifts its spending from hardware to software.

AT&T CEO Randall Stephenson said on the carrier's fourth-quarter earnings call that 2014 would be the peak year for Project VIP investment. But, according to Raymond James Financial Inc. (NYSE: RJF) analysts, the bulk of AT&T's spend will now actually go towards Project Agile, its plan to streamline its operations and simplifying its processes to improve customer interaction. (See AT&T Reports Q4 Leap in Revenues, Profits.)

"As such, we expect the incremental capex favors labor and software and not equipment in 2014," the analysts wrote in a research note. Raymond James expects this shift to hurt some of its wireless equipment vendors like Alcatel-Lucent (NYSE: ALU), CommScope Inc. , JDSU (Nasdaq: JDSU; Toronto: JDU), Dycom Industries Inc. , and Synchronoss Technologies Inc. (Nasdaq: SNCR).

LTE spend continues
Verizon's focus for 2014 is continuing to deploy capital to increase the capacity and density of its LTE network. As such, it is forecasting a range that could see it slightly decrease spending or increase it by up to 1%. (See Verizon's 4G Strength Keeps It Above the Fray.)

AT&T and Verizon were both able to lower their capex spend expectations because they are significantly ahead of Sprint and T-Mobile in their LTE network build outs. For Sprint, most of its heavy lifting came in 2013 when its spending increased 13%, but it's far from over. This year it's expecting to spend 9% more as it brings Spark to more markets and plans to cover 250 million people with LTE by mid-year. (See Sprint Adds 58K Postpaid Subs in Q4 and Sprint Eyes SDN to Re-Craft Its Core.)

For its part, T-Mobile CEO John Legere said he is happy with the carrier's "good, balanced capex portfolio." The scrappy smallest operator plans to spend $4.3 to $4.6 billion in 2014, and Legere jibed his competitors that with significant capital it would be able to close the gap on the larger tier ones quite easily. How might it get that extra capital? Legere suggested consolidation in the industry could help there. (See Is SoftBank Ready to Reunite With Legere?)

"And it also isn't hard to understand why we have a position that we've espoused consistently that, over time, this industry is ripe for the impact of further consolidation, which is one of the ways to have significant capital exploited to try to close that gap," Legere said on the earnings call. "But our own growth right now and our capital profile, we feel is balanced and allows us to do the things that we've outlined."

— Sarah Reedy, Senior Editor, Light Reading

Want to learn more about SDN and NFV? Check out the agenda for Light Reading's Big Telecom Event (BTE), which will take place on June 17 and 18 at the Sheraton Chicago Hotel and Towers. The event combines the educational power of interactive conference sessions devised and hosted by Heavy Reading's experienced industry analysts with multi-vendor interoperability and proof-of-concept networking and application showcases. For more on the event, the topics, and the stellar service provider speaker lineup, see Telecommunication Luminaries to Discuss the Hottest Industry Trends at Light Reading's Big Telecom Event in June.

Page 1 / 3   >   >>
brookseven 3/7/2014 | 4:46:41 PM
Re: It's not About Capex  


But of course, those units are already bought and paid for.  So doing Virtualization on them will only cost money.  :)

I am not denying that there is a place for this stuff.  I am just saying that this is likely a much smaller win than is being shown.


[email protected] 3/7/2014 | 2:00:33 PM
Re: It's not About Capex Agreed and when telco makes a different bet than the web or enterprise IT, it generally ends up loosing.  Remeber the old ATM versus IP arguments.  The work on ATM25 for home networks.  The use of CORBA when the world had gone SOAP and them REST, RCS, IMS, and the belief system on QoS.  One thing I've learned is copy what's working, just copy it better.
Dan Warren 3/7/2014 | 1:55:15 PM
The elephant in the room on virtualization is latency So much I would like to add, but it is dinner time and I am getting harrassment, so a slight change of direction based on the quote in the title lifted from Alan's post.

If latency is the issue, how does NFV fit with 5G and 1ms end-to-end latency?  I pushed vendors on this (and the impact of 1ms latnecy on interconnect) at MWC.  No one has an answer.

If you virtualise the network for 4G (and 3G and 2G maybe), what happens when you go to '5G' whatever that is, and the latnecy requirement runs in to physics, as well as physical topology again?
TomNolle 3/7/2014 | 1:49:25 PM
Re: It's not About Capex I'm no fan of formal standardization processes either; I believe in prototype implementations.  I started and ran a project for open implementation of NFV (CloudNFV, website http://www.cloudnfv.com/) through the end of January, getting it launched both as an ETSI PoC and a TMF Catalyst.  There's considerable documentation on it, and a tutorial on NFV overall, on the website.  It runs today.
[email protected] 3/7/2014 | 1:45:54 PM
Re: It's not About Capex Agreed, but I bet many telcos SMSCs are now running at <10% utilization thanks to internet based messaging services.  Likley the CRBT infrastructure is lightly used. How about their API management platforms, their multiple LBS platforms, their tens of voice VAS platforms, their multiple enterprise UC platforms? If they've deployed IMS, given the delays in VoLTE roll-out I bet there's a lot of spare capacity waiting to be turned on.  Telcos have lots of lots of service elements and many are lightly used given their conservative buying policies.  Its more than SBCs.
[email protected] 3/7/2014 | 1:39:35 PM
Re: It's not About Capex Don't get me started on telco standardization and its inability to deliver business value to telcos.  In the wider world virtualization is separate from cloud.  But you are correct the standards appear to be lumping it all together for better or more likely worse from an implementation perspective.

Glad we agree on the service agility issue, whenever I'm told that I push for specifics and none stand up todate, but I'm sure there will be some cases that make sense, the question will then become is if enough to justify the invstment.
brookseven 3/7/2014 | 1:08:53 PM
Re: It's not About Capex Alan,


That makes an assumption which is not necessarily true in this case.  Imagine that you need to buy $10M of servers and they are all working at 100% capacity at all times.

Server consolidation only works when you run multiple machines at well below capacity.  Which might or might not be the case in any place in telecom.  It is in Enterprises because most functions that an Enterprise does is a one off server (how many Sharepoint Servers does a company need?).  In telecom (on the network side) they run 100s of the same applications - high volume, low mix (How many Session Border Controllers does an operator need?).  

That does not mean that there are no small volume or high mix components.  Just again, reducing the value of virtualization.


TomNolle 3/7/2014 | 12:55:58 PM
Re: It's not About Capex I'm not dodging any question, Alan, and I'm a long-standing member of the ETSI ISG and a contributor to the process.  Leaving the cloud out of this is not possible because the current standard implementation for NFV based on the proof-of-concept submissions to date is OpenStack.

I also said earlier that NFV does not IMHO do anything for service agility, so there's no point in continuing to ask me for examples of where that might happen.
[email protected] 3/7/2014 | 12:54:50 PM
Re: It's not About Capex Hi Brookseven,

The main justification of virtualization is server consolidation - capex savings.  A VMWare sale person goes to the CIO and says, spend $1M on my software and I'll save you $10M in server costs.  So within the financial year there is a capex reduction and the CIO get a big bonus for efficiency gains. 

The elephant in the room on virtualization is latency, that is why no financial exchange has been virtualized, why United booking system is not virtualized, the list goes on.  I agree with you on the limitations of NFV.

We're agreed on the people and process issues that make opex claims look silly.
[email protected] 3/7/2014 | 12:42:28 PM
Re: It's not About Capex Hi Tom,

Let's leave cloud out of it, that is separate to virtualization.  Virtualization is NFV, the V stands for Virtualization of network functions.  So no dodging the question.  An SMSC is software, a HSS is software, an EPC is software.  Today running in their silo'ed implementations with expensive custom systems.  All NFV (according to the telcos that put the NFV whitepaper together that began all this) is doing is putting the software onto commodity servers, the same as all other industries when they implemented virtualization.

Telco is not more complex that is a fallacy, healthcare, military, financial exchanges, etc.  are as complex if not more complex given such requirements as zero down time thanks to Sarbox.  So the excuse of telco is special doesn't cut it.  That's an excuse used by vendors to justify over charging their telco customers.  So please try again in explaining why telcos can not achieve capex savings as every other industry.

And please just give me one specific service agility usecase that is made possible through NFV.  We've got to stop talking marketese and have plain and simple explanations.  Else we're just repeating the IMS fiasco with NFV.
Page 1 / 3   >   >>
Sign In