Sunday, December 6, 2015

FTTP vs FTTN; when is spending $4,400 better value than spending $2,100?

NB: Update: 7 Jan 16. 
The Delimiter comment picked up a calc error** in the below post.
I did allocate the FTTN costs/revenue over 1 year not 1.3 years, which inflates the OPEX pa and the Revenue pa for FTTN. The total diff FTTN vs FTTP being about the same, as the Rev and OPEX changes offset. Makes the OPEX diff $5 per month, and the Revenue $25 per month in favour of FTTP over FTTN. So about the same difference overall. Great to see Delimiter readers paying so much attention to the detail. Revisions identified below with **.
Thanks to david 04/01/2016 at 3:55 pm at the Delimiter comments for the catch.
== Original post

The National Broadband Network (NBN) is Australia's plan to connect all Australians to faster broadband. The current Liberal/National government plans to connect about 25% of premises to cheaper to install ($2,100 per household), faster to roll out FTTN. The previous Labor government aimed to connect 93% of premises to more expensive to install ($4,400 per household) and slower to roll-out FTTP.

It costs $4,400 to install fibre to the premises (FTTP) but only $2,100 to install fibre to the node (FTTN). Recent figures from NBN Co suggest that when the OPEX (Operating Expenses) paid per year to run these services and Revenue generated from these services is taken into account, then FTTP may be better value than FTTN in as short a period as 6 1/2 years. See the Graph below, which shows Cash generated by FTTP exceeds that generated by FTTN after 6 1/2 years. The graph shows total costs for FTTP and FTTN (including CAPEX and OPEX) become equivalent ten years after installation. Coincidentally, the graph shows FTTP breaks even (total revenue exceeds total costs, including CAPEX), at the same time, after ten years. The equivalent breakeven time for FTTN is 30 years, using the numbers below. This graph does not take into account build time, which I will deal with in a second post. **Cash impact of graph unchanged.
Model online to download at DOI:

I did some analysis on the recently published HFC slides (pdf; from Delimiter's 25.11.15 page) which compare FTTP and FTTN Capex, Opex and Revenue. I tweeted them recently (on 2 Dec; first, second, third).

Click on image to enlarge

The HFC slide analysis indicates:  FTTP $20pm (per household) less OPEX; $10pm more REV (per household) than FTTN. With a $30 per month financial benefit, a $2,300 CAPEX diff pays for itself in 76mths (6 1/2 years). The  $2,300 CAPEX difference is FTTP $4,400 CAPEX/hh, less FTTN $2100 CAPEX/hh (per household). (NB: Calc error Revision**: FTTP $90pa per houshold less OPEX, $300pa per household more Revenue than FTTN).

The analysis is based on: 350k households in HFC slide (see image)
FTTP CAPEX $4400/hh | OPEX $125M pa | REV $275M pa
FTTN CAPEX $2100/hh | OPEX $200M pa | REV $225M pa (NB:** This should be over 1.3yrs.)

If you work out these total costs and revenues per household, that equates to:
FTTP CAPEX $4400 | OPEX $350 pa / hh | REV $785 pa / hh
FTTN CAPEX $2100 | OPEX $570 pa / hh | REV $640 pa / hh  (NB:** Opex $440 pa/hh Rev $490 pa/hh)

This data suggests FTTP is better value than FTTN when you take into account OPEX and Revenue after 6 1/2 years. FTTP is not only cheaper to run per annum (OPEX $220 less for FTTP; $570 vs $350 i.e. close to $20 per month), but generates more average revenue ($785 vs $640, close to $10 per month). This difference totals about $30 per month FTTP > FTTN greater margin per household for NBN Co. This is a total net benefit of $360 per household per year for using FTTP over using FTTN. The CAPEX difference of $2300 ($4400 -  $2100) would be earned back in 76 months i.e. $2300 / $30. Over 25% of say 10M households, that works out in total to be a benefit of $900M per year ($360pa * 10M * 25%). So over the say ten year life of FTTN, the benefit is approaching $9B total. The longer FTTN remains in place, the greater the foregone benefit for not switching to FTTP. (NB:** OPEX $90 less for FTTP; Revenue about $300 more for FTTP; diff about $30 per mth).

This annual and total benefit does not take into account extra CAPEX expense to replace the FTTN with FTTP. This would be an extra difference between FTTN and FTTP.

Thus, assuming these figures are correct FTTP would pay for itself, compared to FTTN within 6.5 years, according to NBN Co's numbers on the slide linked above, based on differences in Opex and Revenue, regardless of nearly 100% greater CAPEX on FTTP.

But this is the first time the HFC slide has brought all those numbers easily together. Such information could have been presented in the cost benefit analysis but these numbers tell such an interesting story, that surely I would have remembered that.

Summary: It makes sense to spend $4,400 CAPEX on FTTP (rather than $2,100 on FTTN CAPEX) when the OPEX and Revenue make the margins much more favourable for FTTP than for FTTN. The OPEX and Revenue figures from the HFC slide indicate about $30 per month margin benefit (OPEX plus revenue difference) for FTTP over FTTN or $385 per annum. This margin benefit makes FTTP more attractive than FTTN after 6 1/2 years of installation.

old/Revised **FTTNFTTP
CAPEX per HH$2,100$4,400
OPEX TOTAL pa$200M$125M
OPEX per HH pa$570/$440**$350
Rev TOTAL pa$225M$275M
Rev per HH pa$640/$490**$785
Margin per HH pa$70/$50**$435


Anonymous said...

Hey! I could have sworn I've been to this blog before but
after checking through some of the post I realized it's new to me.
Nonetheless, I'm definitely delighted I found it and I'll be book-marking and
checking back often!

Here is my webpage sdf

Aaron Titman said...


Would you be able to comment on the assumptions made by the following poster regarding your analysis.

Due to the specific posters tendancy to try and turn anything around as a negative, it would be awesome if you could address his comments, in a succinct and clear manner.



aj said...

I don't think the opex and revenue numbers are per annum -- the first row explicitly notes that $700M is 3y of revenue, and 1100M/350k/3/12 would be $87 per customer. $700M over 3y is just $233M pa, so planning on more than doubling your revenue (+$225M, +$260M, +$275M) when your customers are already paying over $80/month for internet just doesn't sound plausible.

As far as I can see the 225/260/275 numbers are just the difference between the additional revenue and the base revenue (ie, 225=700-475, 260=700-440, 275=700-425), and I can't see what use that would have, so I tend to think they're just mistakes when filling out the slide (copied the wrong formula in a spreadsheet?).

For the opex numbers, the footnote suggests the opex numbers go until FY2022, which would be more like 5 years.

If so, then normalising the revenue changes from 3 year to annually gives: 141M, 146M and 158M (ie 141M, +5M, +12M) for FTTP, FTTdp and FTTN. Normalising opex from 5 year to annually gives: -65M, -60M, -50M (ie, -65M, +5M, +10M).

If revenue for FTTP had been set as $470M rather than $475M these figures would've exactly matched, ie giving NBNCo the same profit independent of technology chosen. Which is probably a reasonable first pass at how NBNCo decides on how much to charge, given they're essentially a monopoly provider...

Anonymous said...

Dr. Ferrers, thank you for your informative and forthright analysis. Let us hope somehow, despite the future-hobbling mess that is the MTM, that some elements of the original FTTP NBN can be salvaged. Perhaps FTTdP can be a part of some solution. However, I fear that the special interests have won the day and the public is left holding the can yet again. Anyway, thanks again and keep up the excellent work. D.

Anonymous said...

Whilst your mathematics are sound you cant really draw any predictions as to the total average costs for the entire rollout due to two key reasons:

1) The 350k sample set is not even close to being representative of an average 350k within the entire country - rollout started in rural areas which skews things for metro predictions

2) The forecasted budget has continually blown out for both FTTP and FTTN but at differening rates, this was not factored in nor that the cost of technology and equipment required at both the provider and customer side (the parts owned and operated by NBN specifically) has chaned numerous times since the first connection til now - this further skews any form of prediction.

I'm a huge supporter of FTTP but taglines lines like "when is 4400 better 2100" from an so called analyst that misses some pretty obvious variances from which those numbers are derived from essentially deminishes the value of your message to anyone in-the-know and gives the opposing side ammunition to counter :(

It does make a great news article though..