Fibre Broadband Shares

[Snoopy]

A look at the numerical Buffett Tests 2,3 and 4 will tell you why it is the Australian Federal Government, and not share investors, that are funding this. As at EOFY2020, the debt ratio is

$34,750m / $36,850m = 94%

Only positive cashflow can save NBN from here!

From AR2020 Section E

"As at 30 June 2020 the total committed equity funding of $29.5 billion from the Commonwealth had been provided to NBN Co under the terms of the EFA."

Translation: $29.5 billion of equity is all you are going to get management - deal with it

"On 22 December 2016, a $19.5 billion loan agreement with the Commonwealth Government was signed for the period from 1 July 2017 to 30 June 2021. On 26 March 2019, the tenor of this loan was extended by three years to 30 June 2024."

Translation: You have screwed up your revenue projections. So to stop this thing going belly up, we will give you three more years to sort out some private funding.

"During the year NBN Co entered into facility agreements with a number of financial institutions to secure $6.1 billion of private sector debt for a period of five years. There have been no drawdowns from these facilities as at 30 June 2020."

Translation: $6.1billion found. Only $13.4billion to go. Keep an eye on that positive cashflow, managers.....

This blog from Gary Mclaren highlights some of the issues as the changing forecasts for planned future cashflows unfolded.

https://www.mclarenwilliams.com.au/2...#disqus_thread

The Kevin Rudd Labor plan from 2013 was forecasting $19.4billion in revenue over the 11 year planning period from 2010 to 2021. The latest ScoMo Coalition plan from 2019 had planned revenue slashed to $15.1billion, even as Capital Expenditure costs blew out from $32.9billion to $37billion.

The big difference between NBN in Australia and Chorus in New Zealand is that Chorus owns the legacy copper network. Chorus in New Zealand can harvest profits from the legacy copper network as their own fibre broadband roll out progresses. Contrast that to NBN which must buy out the existing networks run by the incumbent Telstra and Optus networks (the buyout being recorded in the accounts as 'subscriber costs') as NBN expands. P56 of AR2020 explains what is happening in this regard:

"Subscriber costs of $2.4 billion continue to reflect payments to Telstra for the disconnection of existing services and to Optus for the migration of subscribers to services over the nbnTM access network. These costs are expected to virtually cease by FY22 and, therefore do not reflect ongoing activities."

This means that those huge annual losses being posted by NBN are not quite as dire as they appear, given a longer term view. That is why 'EBITDA before subscriber costs' is a much better measure of debt servicing ability than net profit. I don't think Warren Buffett would be rushing to invest in this business. But it might not be quite as dead as those bare Buffett test figures make it appear.

SNOOPY

[Snoopy]

Also most people stream movies, not download them first and then watch. I really don't see many people being able to make use of such high speeds at the moment. Maybe if everything was streaming at 4K and you had a busy household, you might use those higher speeds, but otherwise it'd be wasted for the majority.

There is an interesting exercise in forecasting broadband demand in Australia up until 2026 here:

https://www.communications.gov.au/fi...token=xuAA8KO-

They look at:

1/ Total monthly data required ( Unit: GB/month ): Depends on household composition and changes in technology
2/ Bandwidth Required: (This is the rate of data transferred) (Unit: 'bits per second'): Depends on the number and composition of households accessing the net at any one time - typically peaks outside of working hours, during evenings or weekends). Bandwidth is normally only an issue at peak periods.

These two factors are further analysed by age, income and family type, The total 'internet population' is then 'meshed together' to get a 'population view' of 'Total data' and 'bandwidth' use.

The three technological changes forecast to change internet demand in the next five years are:

1/ Video On Demand (VOD): Evolving in resolution from 'SD' (Standard Definition) -> 'HD' (High Definition), -> 4k -> 8k
2/ The "Internet Of Things' (IoT): Homes are expected to have up to 50 smart devices by 2026 (including smart lights and heating).
3/ Use of Virtual Reality (VR) devices.

Against these growth drivers, file compression technology is expected to result in a 9% reduction in transmission load instantaneously (affects bandwidth) and collectively (affects data) every year. Note that the bandwidth and data requirements of 2/ are almost negligible compared to 1/ and 3/. But low latency may be crucial (you don't want to trip up waiting for the lights to be turned on). VR has similar bandwidth requirements to 8k TV (19.5Mbps) , both of which are ten times that of HD TV (1.9Mbps).

The narrative of the highest impact family scenario is

"While one adult makes a video call, the other multi-screens by streaming VOD and browsing the web. The two children are watching YouTube and gaming online. An update to software is downloading in the background."

The significant drivers of increased internet use are higher definition video (from HD to 4k) , both to watch and as conference software, and the use of more virtual reality in games.

The future 2026 scenario as outlined will require 50Mbps for high use households but less than half that for smaller households with fewer people.

Breaking through the bounds of my own limited imagination of a more internet based future, 'the pros' are predicting that up until 2026, a streaming rate of 50Mbps from your internet provider will be more than enough. Only 2% of households are forecast to need more bandwidth than this. These examples are peak loads. We cannot assume that high use households would pay more to operate at peak load for only a short time of the day. They may select a lower speed and cheaper plan that nevertheless satisfies their needs 95% of the time.

SNOOPY

[Zaphod]

Bear in mind that those bitrates advertised in the fibre plans represent EIR's (excess information rate), with the CIR's (confirmed information rate) being in the vicinity of 2.5Mbps - 10Mbps. In other words, you're not guaranteed to receive 100Mbps 24/7/365 on the UFB network, instead you can burst to that level for a period of time. Obviously that doesn't take into account the weakest link in the chain rule, so depending upon how saturated the upstream content delivery networks are, you may see even less.

Netflix for example, requires 25Mbps for 4K but we actually see usage around 16Mbps at peak on our router, cycling down to a few Mbps.

At this point, 1Gbps+ residential plans are really only good for bragging rights or releasing pressure on the wallet.

[Snoopy]

Bear in mind that those bitrates advertised in the fibre plans represent EIR's (excess information rate), with the CIR's (confirmed information rate) being in the vicinity of 2.5Mbps - 10Mbps. In other words, you're not guaranteed to receive 100Mbps 24/7/365 on the UFB network, instead you can burst to that level for a period of time. Obviously that doesn't take into account the weakest link in the chain rule, so depending upon how saturated the upstream content delivery networks are, you may see even less.

Quite right. The iPrimus website:

https://www.iprimus.com.au/index.html#!/internet.

lists the nominal speed of a customer plan PLUS the expected performance at peak time (from 7pm to 11pm). For 50Mbps there is no difference between the headline bit rate and peak time bit rate. For 100Mbps the expected peak time rate has been measured at 95Mbps. But for the so called 'Home Superfast' plan, which the fine print says is based on 250Mbps, no measured peak time rate is given. At first I was thinking, this must be because there probably aren't enough 'high end' users out there to get a statistically representative sample. But then I began to wonder if these actual speeds were based on line tests from individual houses, or bitrate tests from the exit node of the ISP headquarters. If the latter then it maybe that the 'choke point' is upstream of the ISP in the NBN network somewhere. The other reason for not releasing actual 250Mbps system performance could be that it is embarrassing ;-(

Netflix for example, requires 25Mbps for 4K but we actually see usage around 16Mbps at peak on our router, cycling down to a few Mbps.

At this point, 1Gbps+ residential plans are really only good for bragging rights or releasing pressure on the wallet.

Isn't the likes of NZ's Chorus broadband FTTP technology good for 900Mbps (download)/ 400Mbps (upload) straight off the bat? Sure they sell slower bitstream plans, but these would be artifically restricted by Chorus? I am surprised your peak bitstream speed is only 16Mbps Zaphod. Or are you on ADSL or something?

Reading between the lines, it sounds like you are happy with your set up. But presumably if Netflix says you need 25Mbps and you find 'a few' Mbps to 16Mbps is OK, you are able to buffer your incoming bitstream to provide an acceptable viewing experience?

SNOOPY

[Snoopy]

Quite right. The iPrimus website:

https://www.iprimus.com.au/index.html#!/internet.

lists the nominal speed of a customer plan PLUS the expected performance at peak time (from 7pm to 11pm). For 50Mbps there is no difference between the headline bit rate and peak time bit rate. For 100Mbps the expected peak time rate has been measured at 95Mbps. But for the so called 'Home Superfast' plan, which the fine print says is based on 250Mbps, no measured peak time rate is given.

Below is the best bit of writing I have seen on the interplay between 'Fibre Broadband Wholesalers' and the 'Retail Service Providers' in a deregulated telecommunications market.

https://www.nbnco.com.au/content/dam...per-170731.pdf

Before you read the article you need to know about the two components of any wholesale charge:

1/ Access Virtual Circuit (AVC): An access charge determined by the maximum PIR (Peak Information Rate) bit rate requested, and the
2/ Connectivity Virtual Circuit (CVC): The collective amount of bits to flow between the wholesaler and the retailer at any given time.

After I read the article, I suddenly understood how the whole system works. But when I tried to translate this experience back to New Zealand, and 'Chorus' with 'NZ retailers' I had a problem.

Appendix 2 in the reference below is the Chorus wholesale pricing list:

https://company.chorus.co.nz/sites/d...st-2020-10.pdf

But here is what does not translate. All of those many connection plans seemed to have one price, regardless of how much data our Retailer's potential end line customer might need (IOW no variable CVC charge). Isn't Chorus in trouble without a CVC charge? By not having a CVC charge, are Chorus not in effect writing contracts to supply an unlimited amount of data?

SNOOPY

[Zaphod]

Isn't the likes of NZ's Chorus broadband FTTP technology good for 900Mbps (download)/ 400Mbps (upload) straight off the bat? Sure they sell slower bitstream plans, but these would be artifically restricted by Chorus? I am surprised your peak bitstream speed is only 16Mbps Zaphod. Or are you on ADSL or something?

Yes you're correct, the speed of plan is artificially restricted, generally by having the core network hardware drop any packets that appear after exceeding the PIR. A software change is usually all that's required to shift a customer from one bitrate plan to another, but the complicating factor seems to be the contractual arrangements between the LFC & ISP. This latter point accounts for the restrictions on the end customer changing their plans regularly.

Reading between the lines, it sounds like you are happy with your set up. But presumably if Netflix says you need 25Mbps and you find 'a few' Mbps to 16Mbps is OK, you are able to buffer your incoming bitstream to provide an acceptable viewing experience?

We have a 100/20 plan, and are heavy users given my use of the connection for video conferencing, remote network administration etc. from my home office, and our propensity to stream videos & general use of the internet. While we do saturate that connection from time to time, there isn't enough justification IMO to increase the bitrate of the overall plan.

Netflix (just as an example) provides a general guideline for the total bandwidth you should have available so as to enjoy a seamless (non-buffering) experience. In this case I think they've padded the recommendation out to 25Mbps to ensure that your experience isn't affected by other equipment or people in the premises also using the same connection. The real-world requirement seems to be around 16Mbps for the 4K stream using their selected codec. Personally, I think their 4K image quality is good, but could be better with a different codec and higher bitrate requirements, but that's a whole different kettle of fish.

[Zaphod]

But here is what does not translate. All of those many connection plans seemed to have one price, regardless of how much data our Retailer's potential end line customer might need (IOW no variable CVC charge). Isn't Chorus in trouble without a CVC charge? By not having a CVC charge, are Chorus not in effect writing contracts to supply an unlimited amount of data?
SNOOPY

You can think of Chorus and the other LFC's as providing the network infrastructure to transport the data packets. The LFC's ensure that their pipes and switching/routing infrastructure are capable of handling the throughput to adhere to any CIR/PIR agreements they have, but they aren't overly worried about the total volume of data transported. The ISP's have peering relationships both locally and internationally, and it is these entities that keep track of the total data throughput and charge the ISP accordingly.

The best comparison I can make would be between the LFC's and the electricity network providers (Vector, PowerCo etc.) The lines infrastructure are sized according to demand, which admittedly is a function of the total throughput, but overall the network provider isn't interested in tallying the exact number of Kw/h being supplied to each retailer or consumer, which is the responsibility of other parties (peers in the above example).

Unfortunately, the landscape is becoming even more complicated, as some ISP's and peering network's seek to charge companies such as Netflix for traffic traversing their networks. This is regarded by some as double dipping, as customers have paid for a certain bandwidth and data cap, but yet the ISP/Peer is seeking to also charge another party for the same traffic.

I've perhaps oversimplified, glossed over, and committed other technical atrocities in the last two posts (for which I'm certain I'll be vilified for) in order to make it appropriate for those not in the IT industry, however the gist is correct.

[Snoopy]

The speed of plan is artificially restricted, generally by having the core network hardware drop any packets that appear after exceeding the PIR (Peak Information Rate). A software change is usually all that's required to shift a customer from one bitrate plan to another, but the complicating factor seems to be the contractual arrangements between the LFC (Local Fibre Company) & ISP (Internet Service Provider). This latter point accounts for the restrictions on the end customer changing their plans regularly.

A reason I can see for having a contract based arrangement between the LFC and ISP for internet delivery would be to give certainty of total demand (the LFC overall supply 'highway' has to be big enough at peak demand for all ISPs combined) and rate of supply (the width of the 'slip road' supplying each ISP must be big enough). If the size and stream rate of bits of information was not contracted, then neither the ISP nor the LFC would be able to size their switching/routing equipment correctly.

Another point here is that the 'amount of data' an end line user wants and the 'rate at which they want to collect that data' are these days usually linked, because much of the data is for 'video on demand' (VOD). VOD demands a high bitrate to deliver high definition images.

You can think of Chorus and the other LFC's (Local Fibre Companies) as providing the network infrastructure to transport the data packets. The LFC's ensure that their pipes and switching/routing infrastructure are capable of handling the throughput to adhere to any CIR (Committed Information Rate)/PIR (Peak Information Rate) agreements they have, but they aren't overly worried about the total volume of data transported.

From page 2

https://www.nbnco.com.au/content/dam...per-170731.pdf

"There are a lot of statistics involved which form the assumptions of what the quality level of access will be at peak times. If RSPs don’t dimension their own network with enough capacity, if they don’t purchase enough CVC flow through at peak time, or if nbn has not dimensioned its network with enough capacity, service will degrade at peak time."

NBN in Australia seems to be concerned to see that their network has enough capacity.

The ISP's have peering relationships both locally and internationally, and it is these entities that keep track of the total data throughput and charge the ISP accordingly.

Or maybe not charge the ISP accordingly? This could be useful if Netflicks released a popular new movie. Peering would mean that only one of the peered ISPs would have to download the movie to offer it to all peer connected ISPs' customers. That way, one download could be 'sold' to several hundred people, (saving the ISPs traffic charges), but Netflicks could keep all the individual subscriber payments to watch the movie. Clever.

SNOOPY

[Zaphod]

A reason I can see for having a contract based arrangement between the LFC and ISP for internet delivery would be to give certainty of total demand (the LFC overall supply 'highway' has to be big enough at peak demand for all ISPs combined) and rate of supply (the width of the 'slip road' supplying each ISP must be big enough). If the size and stream rate of bits of information was not contracted, then neither the ISP nor the LFC would be able to size their switching/routing equipment correctly.

Another point here is that the 'amount of data' an end line user wants and the 'rate at which they want to collect that data' are these days usually linked, because much of the data is for 'video on demand' (VOD). VOD demands a high bitrate to deliver high definition images.

Yes, it's very important to have some certainty around capacity requirements and those contracts play an important role in that. The RSP's and LFC's have also have a huge pool of historical data to conduct trend analysis upon, so predicting demand and designing their networks accordingly, becomes easier in most respects.

"There are a lot of statistics involved which form the assumptions of what the quality level of access will be at peak times. If RSPs don’t dimension their own network with enough capacity, if they don’t purchase enough CVC flow through at peak time, or if nbn has not dimensioned its network with enough capacity, service will degrade at peak time."

NBN in Australia seems to be concerned to see that their network has enough capacity.

Yes, that's right. The context of my earlier statement above is related to the total volume of data transferred in GB or TB in terms of the LFC. That total is not particularly important to the LFC, but knowing what the peak flow rates are and having the ability to adequately service them are very important. Likewise, the RSP must also ensure their network is capable of sustaining their peak flow rate, while simultaneously monitoring the total volume of data transferred to minimise the costs incurred from peering.

not[/i] charge the ISP accordingly? This could be useful if Netflicks released a popular new movie. Peering would mean that only one of the peered ISPs would have to download the movie to offer it to all peer connected ISPs' customers. That way, one download could be 'sold' to several hundred people, (saving the ISPs traffic charges), but Netflicks could keep all the individual subscriber payments to watch the movie. Clever.
SNOOPY

There are many different methods employed by the ISP, CDN and service providers to achieve this, some of which (such as zero rating traffic for specific services) has come under legal scrutiny. Most SVOD services employ a geographically localised CDN that is directly peered to each of the ISPs, or sometimes just a specific favoured ISP. Sometimes this CDN may actually be deployed within the ISP core network itself. You can see there are plenty of ways to earn a dollar from each of the parties involved, while also charging the customer for data whose costs has in some cases has already been paid for by another party.

There is enormous complexity behind scenes in relation to the core networks that supply internet services, as well as the business models employed by each of the parties involved.

[Snoopy]

The future 2026 scenario as outlined will require 50Mbps for high use households but less than half that for smaller households with fewer people.

Breaking through the bounds of my own limited imagination of a more internet based future, 'the pros' are predicting that up until 2026, a streaming rate of 50Mbps from your internet provider will be more than enough. Only 2% of households are forecast to need more bandwidth than this. These examples are peak loads. We cannot assume that high use households would pay more to operate at peak load for only a short time of the day. They may select a lower speed and cheaper plan that nevertheless satisfies their needs 95% of the time.

The Average Revenue Per User (ARPU) broadband trend for NBN and Chorus (as published by the respective companies) is as follows:

	NBN	Chorus
FY2014	$A37
FY2015	$A40
FY2016	$A43
FY2017	$A43
FY2018	$A44
FY2019	$A44	$NZ47.50
FY2020	$A45	$NZ48.12

There are a few gaps in the 'Chorus' column. That is because although Chorus have been keen to mention in passing how the ARPU of their broadband base is growing, they have been very shy about quoting actual figures. What we do know is that the fibre reaches more deeply into NZ broadband network, with FTTP being the accepted goal whereas in Australia FTTN and even lesser fibre broadband reach has been accepted as a 'cost cutting way' to achieve 'fibre grade broadband' nationwide. It is therefore a little surprising to see that once you adjust for the respective currencies, the NZ system is both cheaper for the consumer and better performing.

My concern about the 'fibre broadband network model' is this. Both NBN and Chorus have enormous debts, because both networks have been built, with government support, ahead of the demand curve. Such a business model was always going to be loss making for many years, while demand caught up with supply. This delay between getting in the revenue and laying out the capital was a good reason for the respective governments on both sides of the Tasman to become involved. However, while the capital spend has largely been completed in both cases, the demand side of the service is still evolving. If both NBN and Chorus are relying on ever increasing ARPU to pay off their massive debts, yet technological developments have - to some extent- caught up with functional demand, what is the incentive for customers to pay more and more for higher bitstream rates? And if ARPU does not continue to increase, what does that mean for the longer term capital position of these network companies (hint: IMO it isn't looking good)?

No one would argue that the NZ roll out of broadband has produced the superior network. Even the Australians grudgingly admit this. But I would argue that the superior NZ network makes the option of paying more for something faster less useful for NZ customers from here on in. That means, in growth in ARPU terms, and taking a network owners perspective, the likes of Chorus NZ is in a worse position than NBN in Australia going forwards.

SNOOPY