Thursday, April 8, 2021

Where are we at with the NBN?

 It has been nine months, since we passed the target build completion date for the NBN. In the half-yearly report, NBN generated a $0.4B positive EBITDA (effectively cashflow from operations), largely due to declining Subscriber Costs (down $0.6B), and increasing revenue (up $0.4B), while CAPEX is also close to half the previous year, declining from $2.5B to $1.4B for the half-year.

NBN also commenced a post-build investment of $4.5B, announced in the Sept Corporate Plan, with headlines of enabling 75% of the fixed line network to reach gigabit speeds (the highest offered). This will enable all the FTTP, HFC and FTTC (and maybe FTTB) to reach the fastest speeds, and a proportion of the FTTN to be upgraded to enable an FTTP overbuild. There is $700M for regional business fibre hubs, and $300M for co-investment with rural and regional communities.

On the demand side in 2020/21, there has been a significant slowdown of brownfield's new signups (See Figure 1; 2021 Avg 206k/Qtr, 2020 Avg 393k/Qtr, 2019 336k/Qtr, 2018 355k/Qtr):

  • Q1 20 - 325,000 additions
  • Q2 20 - 250,000 additions
  • Q3 20 - 130,000 additions
  • Q4 20 (budget) - 120,000 additions. Around 1% of 10M ready to connect premises.

Figure 1 - NBN Consumer adoption (2016 - 21); Source NBN Weekly Summary. Data: Google Sheet.


Figure 1a. - NBN Brownfield Additions 2021 - significantly slowing. Source/Data: per Figure 1.

NBN retailers (RSPs) are being consulted about NBN's proposed pricing, looking forward for two years, and continue to complain about CVC variable costs, which put pressure on RSP margins, if consumers are not increasing their plan speeds and prices. Both Senate and large RSPs are keenly interested in the potential for NBN to reduce reliance and exposure to rising CVC costs due to increased consumer data demand.

NBN revenue is nearing a $5B annual rate, with more boost from business usage (inc new fibre zones), and demand for consumer speed increases. NBN #FocusOnFast is the 2021 plan to lift subscribers from 50Mbps to 100Mbps and above, through a six month discount of $20-30 on the most expensive plans; reducing gigabit retail prices from $149 to $119, and 250Mbps from $129 to $99, and 100Mbps from $99 to $89 per month. These plans are mostly for unlimited data, so can incur greater CVC costs, though NBN bundles more CVC with the fastest plans to encourage higher plan usage.

Current NBN penetration, given slowing additions, are at the following rates (see Figure 2):

  • Brownfields - 71% and slowly rising (2020 - 64%)
  • Greenfields - 66% and steady (2020 - 64%)
  • Wireless - 57%, jumped from 2019, but now slowly rising (2020 - 53%, 2019 - 43%)
  • Satellite - 26%, very slowly rising (2020 - 23%, 2019 - 21%)


Figure 2 - NBN adoption by Technology (2016 - 21); Data NBN Weekly Summary: Google Sheet

So NBN is reaching a revenue headwind period, where revenue growth will become more challenging, (and expensive) and come from:

- new additions; new homes in Greenfield sites, or 1% organic growth in Brownfield sites, plus some potential for growth in Satellite and Wireless usage

- lifting consumers to higher speed plans (increasing ARPU); with discounts to entice usage of faster speeds (#FocusOnFast); which will also be impacted by levels of NBN satisfaction

- engaging more businesses on to the NBN (increasing overall ARPU); but which will require investment in a Business Sales force.

A lesser opportunity for growth is through CVC, where rising consumer data usage incurs increasing CVC, which can either squeeze RSP margins, if prices are inflexible, or create price inflation, which could reduce demand and/or decrease satisfaction.

 Figure 3 - Sales and NBN Users - Corporate Plans (2016 - 24). Data: Observable

Other issues:

- NBN is refinancing its $20B debt with the Commonwealth, with private debt by mid-2024, and has also raised $10B private debt at low rates, enabling the post-build investment, and retiring of some Commonwealth debt.

- NBN faces increasing competition from both 5G in slower FTTN suburbs, and more rural wireless sites, and LEOs in the Satellite footprint, but the $140 per month LEO pricing will be more attractive to higher end consumers or businesses. 

- ABAC: The Minister, Paul Fletcher has appointed an Advisory Council to provide guidance around broadband generally and ABAC released its first report, Riding the Digital Wave. The Council will focus on specific industries eg Agriculture and Health, plus Digital Inclusion, Digital skills and small/medium businesses. (p.3)

- In global competitiveness, Australia still compares poorly  (Data: Figshare) against our top ten trading partners in average national speeds (per Speedtest Global), though Feb 2021 saw a 20% increase in Australian national average download speed (from 59Mbps to 71Mbps; which looks like a year's growth in two months), perhaps from more gigabit services coming online. The speed jump would equate to about 1% takeup of gigabit plus 1% superfast (250Mbps) services; significantly higher (8X) than the 25,000 over 100Mbps services the ACCC reported at Dec 2020. ACCC's March report (expected late May) will provide takeup information for over 100Mbps services.

Comparing Broadband Speed with Trading Partners

Speedtest has made its data available globally (in anonymised 600m^2 blocks), so it is possible to see distributions of speed across the country. This enables our cities to compare with our Top Ten Trading Partners eg Shanghai, Bangkok and Los Angeles. Melbourne, for instance, has far slower distribution of broadband services compared to these overseas cities.

Figure 3. Comparing Speedtest results of Australian and Top Ten Partner cities, Q3 2020 (Data; Figshare | Pic: Twitter thread).

Further work - Broadband performance data (Speedtest) vs NBN Technology

An activity to compare NBN technology maps (see (showing regions of FTTP, HFC, FTTC, FTTB, FTTN etc) and average Speedtest results is underway. This will produce maps of Australia cities, showing distribution of speeds, and eventually average speed by NBN technology. There is also a plan to compare broadband speeds with SEIFA index, the socio-economic index of areas, to see if broadband speeds vary by wealth of neighbourhoods.

Figure 4. Speedtest results across Melbourne Q2 2020; colour coded by speed (Data Figshare; Source: Twitter).

The results show speeds vary across the city, with patches of faster broadband, north of Geelong, and in outer west, and in wealthy inner-east suburbs and slower on the fringes of the city, but across the suburbs, your mileage will vary. See Twitter thread for more images and analysis. This data is also loaded to the AURIN geospatial workbench, where Australian researchers can access the data online (Figure 5).

Figure 5: Speedtest data available at AURIN geospatial workbench (Data/Image: Figshare).

Figure 6: Speedtest data Q3 2020 - AURIN - three colour (less than 20Mbps; grey; over 90Mbps; blue; Rest = red) (Data/Image: Figshare).

Thursday, November 5, 2020

Has FTTN paid for itself?

I will analyse Communication Minister Fletcher's recent claim that NBN's FTTN has paid for itself. FTTN is the poor cousin to FTTP, the gold standard of broadband. Where FTTP is capable of terabits per second (10 ^ 12 bps), but twice as expensive as FTTN to install ($4000 vs $2000),  in contrast, FTTN suffers from degradation over distance from the fibre at the node, so 1/3 get very fast (<500m; up to 100Mbps and a bit more), 1/3 fast (500-750m; up to 50Mbps) and 1/3 slow (>750m; less than 50Mbps). NBN's plans in 2020 include enhancing half of NBN's FTTN footprint with fibre on demand, and upgrading FTTC and HFC to gigabit speeds, giving 75% of NBN's fixed line customers access to gigabit speeds.


FTTN is one of the fixed line technologies, NBN (the National Broadband Network) uses to deliver broadband to Australian homes. It is part of Australia's MTM (multi-technology mix) approach that aims to deliver broadband to Australians at lowest cost to the Government and as early as possible, ensuring that nearly all Australians were ready to connect with broadband when the COVID-19 pandemic arrived in 2020. In contrast, Australia's prior government envisioned an NBN with largely FTTP broadband delivery, which is more expensive to install, but capable of very high speeds over potentially future decades. 

FTTN uses copper phone lines to deliver the last stretch of broadband, while FTTP takes fibre right into the house. The longer the copper, the more the interference through the medium and the slower the broadband speeds to the user. NBN Co reported in its Corporate Report 2021 that FTTN users fell into one of three categories; 1. 1.2M who could get 100Mbps; 2. 1.9M who could get 50Mbps, and 3. 1.0M who could only get the baseline 25Mbps. FTTP in contrast currently enables 1000Mbps, with some countries now activating 10,000Mbps access speeds over fibre. 

Australia's current government argues that most households are satisfied with their current access speeds, but are providing a way forward, as households demand higher speeds. NBN for instance launched cheaper 250Mbps and 1000Mbps services in May 2020, which attracted a few thousand new subscribers to these high speeds, and may have caused a 25% uptick in Australia's average broadband download speed.

Upgrades Corp Plan 2021-24

Following NBN Co's completion of the volume rollout in 2020 (99% complete, with 100,000 more difficult premises to finish), its Sept 2020 three year Corporate Plan announced investment to lift Australian broadband performance. To that point, NBN's goal in its Statement of Expectations was to provide all premises with at least 25Mbps by 2020, and 90% of fixed line premises with 50Mbps (as soon as possible). The Corporate Plan announced several investment strands, including:

- business zones where businesses could get fibre services connected for $nil upfront on committing to a service

- upgrade to HFC and FTTC technologies to gigabit service levels

- upgrade of 50% of FTTN footprint to enable fibre on demand, when customer demands a fast service, which its FTTN can not provide, for $0 install cost to customer.

Minster Fletcher

In announcing the NBN Co Corporate Plan and committed investment, Minster Fletcher stated:

"By 2023, the FTTN network will have generated $9 billion in revenues, cost $7.3 billion to build and $1 billion to operate. FTTN will have effectively paid for itself - and will continue to provide a very good service to millions of premises for many years, generating cashflows well into the future". CommsDay Speech, 27.10.2020. 

Let's look at the "paid for itself" claim. 

Assessing the claim

The claim consists of three parts;

- FTTN has generated $9 billion in revenue - I agree

- FTTN has cost $7.3 billion to build - my estimate $11 billion

- FTTN has cost $1 billion to operate - my estimate $2 billion, but full costs add another $20 billion.


The NBN has generated $10 billion in revenue in the last five years (2016 - 2020); NBN Co Annual Report 2020, p.56. Over the next three years, NBN is budgetted to generate another $15.6 billion (Corporate Plan 2021, p.54). NBN revenue can be prorated by the types of technologies to apportion their revenue by technology type. At 2020, FTTN/B has 3.1 million activated premises from a total of 7.3M active connections - 42% of connections. By 2023, FTTN drops to 36% of connections as another 1.5M premises are activated but very few are FTTN (less than 10%). A first cut estimate of FTTN revenue is 43% of pre 2020 revenue, and 36% of post 2020 revenue; equating to about $9.6 billion. So close to Fletcher's claim.

Build Cost

NBN reports cost per premise (CPP) by technology. Per NBN Corporate Report 2020, NBN had 4.7M homes serviced by FTTN/B as CPP of $2340 (NBN Corporate Plan 2020, p.52). This gives total Build Cost for FTTN of about $11 billion. Added to this should be a proportional share of common CAPEX. The 2020 Annual Result presentation shows Commons CAPEX at only around 10% of total CAPEX, so would only make a minor difference. The NBN Annual Report does not break down Network assets by network technology type, nor identify levels of common capex. But NBN shows on their books a network worth $42.5 billion less $9 billion depreciation, for 11.7M premises ready to connect - an overall average of around $3600 per premise. So, overall the Minister's claim of FTTN costing $7B to build seems low compared to my $11B estimate.

Cost to Operate

In 2020, NBN reported Direct Network Costs of $641M, about 30% of total Operating Costs ($2.0B) NBN Operating Result Presentation 2020, p.11. From 2016 - 2020, NBN reports Operating expenses totalling $9.4B (NBN Annual Report 2020, p.56). Using the 42% share of revenue as above, the FTTN share of Operating Expenses would be around $4B, and likely direct network costs at 30% of $1.2B. For 2021 - 2023, Operating Expense can be calculated as Revenue - Subscriber Payments - EBITDA or ($15.6B - $1.4B - $8.6B = ) $5.6B. The FTTN share at 36% would be close to $2.0B, with Direct Network Costs (30%) around $0.6B. So my estimate of total Direct FTTN Network costs of ($1.2B + $0.6B) around $1.8B, close to double Minister's Fletcher's estimate.

However,  how much of non-direct Operating Expenses should be attributed to FTTN. Should it include employee costs (around 30% of operating expenses), subscriber acquisitions costs (close to $10B). Addition of these costs would increase FTTN related expenses to $2B (employee costs) and roughly $4B for related subscriber costs). There is also depreciation and interest cost incurred by NBN, which in 2016 - 20 are in the order of $15 billion. A share of FTTN would add another $6B to FTTN's bill. Thus total expenses related to FTTN would be closer to $14B (2016 - 20) while the direct costs to operate, I estimate are around $2B (2016 - 23). To add the 2021 - 23 portion of FTTN costs, would add another $2B of employee expenses and 3/5 of depreciation/interest, so close to $3.5B. Grand Total Expenses: $21.5B  Thus the $1B FTTN Cost to Operate is extremely light on in excluding employee costs, subscriber acquisition costs, depreciation and interest, understating total costs by close to $20 billion over 2016 - 23.


Thus I would argue that FTTN has not paid for itself. Even with a narrow view of network operation costs, margin earned by FTTN services has contributed around $7 billion to NBN operating funds, while I estimate build costs at $11 billion. A fully expensed costs of FTTN is closer to $20 billion for 2016 - 23, including employee costs, depreciation and share of subscriber costs.

Thursday, July 2, 2020

To Gigabit or not to Gigabit

As the NBN rolls past its completion date, the Government is strangely silent choosing not to mark the occasion. Perhaps remaining quiet before this weekends by election. A new report out in the last few days is titled “ the gigabit gap “ argues for Australia’s weak international ranking on gigabit aspiration.

  • Gigabit is 40 X the mandatory speed promise for all. (25 Mbps)
  • Gigabit is 10 X the previous top speed available widely. There were business plans ($800/mth) and 250 plans availed on FTTP, which makes up about 20% of the NBN.
  • Gigabit is on the roadmap for 50% of NBN, inc FTTP, HFC and FTTC.
  • NBN has committed that all HFC will have access to 250 Mbps by June 2021. But at release of new high speed plans (end May) only some 7% of HFC could get gigabit speeds. Home Superfast (250 Mbps) was available to 32% of NBN ready to connect, NBN said, inc up to 70% of the HFC footprint.

What’s in a name?

NBN calls fast 100Mbps, Superfast 250 Mbps, and Ultrafast 500-1000Mbps.
But ACCC called Superfast when referring to 25Mbps.
NBN calls a 25Mbps service, Home Basic.

The NBN SOE aims for universal very fast broadband - “ The Government is committed to completing the network and ensuring that all Australians have access to very fast broadband as soon as possible, at affordable prices, and at least cost to taxpayers. The Government expects the network will provide peak wholesale download data rates (and proportionate upload rates) of at least 25 megabits per second to all premises, and at least 50 megabits per second to 90 per cent of fixed line premises as soon as possible.”

I would agree that 250 Mbps is very fast, beyond fast, and so the NBN according to the SOE should be pursuing universal access to this service AS SOON AS POSSIBLE. So for now Australia has universal basic broadband, and we are still waiting for universal very fast broadband.

“The Statutory Infrastructure Provider (SIP) obligations ensure that all Australian premises are able to access superfast broadband services (25 Megabits per second (Mbps) or better).” Dept of Communications

“Our priority is to help deliver high speed broadband to premises across Australia and, as we complete the initial volume build to 11.5 million premises7, we are starting to unleash higher speed tiers on a phased basis,” said Brad Whitcomb. “Launching the three new higher wholesale speed tiers is the next step in our network evolution and we will continue to upgrade the network to offer higher speed services to more customers over time.”, NBN said.

Thursday, January 30, 2020

The future of the NBN (7); Inquiry into the business case of the NBN

In October 2019, the Parliamentary body overseeing the NBN - the Joint Standing Committee on the NBN, launched an Inquiry into the Business Case for the NBN and the experiences of Small Business.  Submissions were due a couple of weeks ago (17 Jan), and my submission (#16) went public this week. Data is at Figshare. I previously responded to a similar Inquiry in 2018 (Data) with a model to calculate the value of the NBN. On 25 Feb, I will talk on this topic at Telsoc. Slides now available at: Slideshare.

Committee Focus

The Committee were interested in "the rollout of the [NBN] and the performance of nbn co", particularly:
a.the economics of the NBN, including key operational and financial performance forecasts in the Corporate Plan 2020-23;
b.current pricing structures, including wholesale pricing, affordability and take-up, particularly as they relate to low-income and rural and regional customers; coverage issues; including reporting of outages planned and unplanned
d.the delivery of the business segment strategy, including to enterprise and government customers, and small to medium businesses;
e.the experiences of small and medium business in relation to the utilisation, accessibility, customer service and affordability of the NBN;
f.compliance with the NBN Statement of Expectations and adequacy of that Statement
g.any other related matters. Source: Parliament of Australia

Australia vs OECD and Top 10 Trading Partners

I presented the material from the future of the NBN posts (1-6), especially NBN speeds vs OECD, but narrowed that down to Australia's Top 10 Trading Partners (Source; OEC - Observatory of Economic Complexity), which are an interesting mix of countries, including:
  • China | United States | Singapore
  • South Korea | Japan | New Zealand
  • Thailand | India | Germany | United Kingdom.
Australia per OECD (2018) was close to the bottom (see Figure 2 below), but post-build NBN, we come back to middle of the pack. An Appendix shows the NBN plans for each of these countries (Table 1). The Broadband Commission produced a nice summary graphic (Figure 1) comparing NBN target speeds and coverage for UK, EU, USA, South Korea, NZ and DE (plus others).
Figure 1. Broadband Commission summary of NBN Plans; Australian and Top 10 Trading Partners highlighted. Broadband Commission: The State of Broadband (2018), p.37.

Figure 2. Appendix 2 of Ferrers (2020) Submission. Australia Broadband speeds vs OECD (2018). Source: OECD 2018. Raw data (OECD xls).

Summary of Australia's Top 10 Trading Partners - NBN Plans - Appendix 1 - Submission

CountryTargetSpeed (Mbps)Year
Australia 90%50 2020
Australia 100%25 2020
China 50%100 2020
China 70% 502020
China some10002020
India 100% 50 2022
India Villages10002020
India Villages >10002022
Japan 98% Fibre Now
New Zealand 75% Fibre 1002019
New Zealand 87% Fibre 1002022
Germany 100% 502018
Germany Regional1000 2025
Singapore ?? oops left it out - in error
South Korea 99% Fibrenow
South Korea 50% >10002022
Thailand Cities 100 2020
Thailand 95% broadband 2020
United Kingdom 15M(55%)Fibre 2025
United Kingdom 100% Fibre 2033
United States 100M(80%)100 2020
United States extensive wireless2020
Table 1: Comparing Australia's and Top 10 Trading Partners - NBNs.
Source UNESCO/ITU Broadband Commission | List of Plans (NZ: Wikipedia)(South Korea, Japan International Broadband Scorecard, Ofcom)

Fast Broadband (at/over 100Mbps) - Australian and Selected Top 10 Trading Partners

What was really interesting, and surprising, was China has rolled out an NBN since 2013, and now has 250M fast broadband connections (77% of fibre users; which are 93% of internet users; Future of NBN 6), compared to US 30M fast broadband users, where fast is at or over 100Mbps. Both countries report coverage of 90% for access to fast broadband services. US reports about 10 subs (of fast broadband) per 100 people, whereas China has close to double, around 20 subs per 100 people.

Figure 3.  Figure 1b. in Ferrers (2020) Submission. Millions of fast Broadband Subscribers. Australia and selected Top 10 Trading Partners. NB: Japan, South Korea did not report in OECD 2018.

I also looked at more recent data from Ookla, which reports average speeds by country of its Speedtest service. Australia does not fare well on this metrics. Australia rates near the bottom of our Top 10 Trading Partners (Figure 4).

Figure 4. Appendix 4 in Ferrers (2020) Submission. Download Average Speeds - Australia and Selected Top 10 Trading Partners. Ookla Global Index 2018, 2019.

What I realised (in Figure 2 - gold bars), was that other countries have a lot of fast (over 100Mbps) connections whereas Australia does not.  In 2018, we reported 0% fast broadband to OECD. In the Ookla graph from mid-2019, we showed close to 0% at or over 100Mbps. This year (around May 2020), NBN will release more affordable gigabit services for around 50% of users (FTTP, FTTC, HFC). I argued in my submission that even small percentages using gigiabit will have a significant impact on Australia's average speeds (which is often complained about). New Zealand currently has about 10% gigabit users, so this level of takeup seems quite possible for Australia too, if the price is affordable. So I am optimistic that Australia can increase its national download performance, if affordable prices can be set for gigiabit services.

ScenarioAverage Speed Ranking against Top 10 Trading Partners
OECD 2018 21mbps 10th
NBN post-build 50mbps 10th
5% gigabit 98mbps 7th
10% gigabit 145mbps 3rd
20% gigabit 240mbps 1st
Table 2: Adding Gigabit substantially raises Australia's national average broadband speed, relative to Top 10 Trading Partners
Source: Appendix 5, Ferrers (2020) Submission.

Overall my recommendations in my submission (data at Figshare) were:
  1. Encourage NBN gigabit services at affordable prices.
  2. Update NBN Co's Statement of Expectations to upgrade the network (preferably bi-partisan), and guide how NBN Co will spend cash NBN will generate on debt repayment, dividends and network improvement.
  3. Focus NBN Co on encouraging use of the NBN (so we are more like China - heavily used, than US - accessible but less used at fastest speeds).
  4. NBN Co publish customer satisfaction results with the network by major customer categories eg business, consumer, FTTP, FTTN, FTTC etc. (bearing in mind that satisfaction comes from more than just speed eg ITU/UNESCO 2025 Broadband Goals), and lastly
  5. NBN Co include in its Corporate Plans a section on how Australia compares to other countries, along key performance metrics, and particularly the Top 10 Trading Partners. Ongoing NBN performance targets should reflect shareholder preference for Australia's performance relative to our major trading partners.
 In contrast, and we should not forget, Australia has come a long way, with NBN and broadband from ten years ago. In 2009, nearly everyone had less than 25mbps, and about 30% had less than 2Mbps. Going back to 2006 (from ABS Internet 8153.0 Usage), some 45% of Australia was on dialup. In 2009, those still on dialup (1M users) were downloading 10Mb per month. The average in 2009 across all users was 4Gb per month. NBN recently (at end 2019) reported NBN users were using 250Gb per month). Extraordinary growth in download consumption.

Figure 5: NBN Speeds 2009, 2018, post-build. Source: ABS, OECD 2018, NBN Co. | Tweet

Update: China reports (p.11,12) 2020 Internet development -
As of December 2019, the number of FTTH/O (home/office) 2019 users had reached 417 million, accounting for 92.9% of all fixed Internet broadband subscribers. The number of subscribers at/over 100Mbps reaches 85% of all fixed broadband users (up from 77% in 2019).

Thursday, November 21, 2019

The future of the NBN (6): UK(2030) vs China(2019)

Two reports recently from China and the UK on their broadband rollout plans, report:
  • a promise "to deliver free full-fibre broadband to all individuals and business by 2030.... [further we] will aim to deliver free full-fibre broadband to at least 15-18M premises within five years" Link, Nov 16, 2019
  • Update: June 2019 report on Internet released (pdf); noting 395.75M Fibre (FTTH - Fibre to the Home, FTTO - Fibre to the Office) connections "accounting for 91.0% of all Internet broadband access users" (p.11) and 77% of broadband subscribers at 100Mbps or over; (p.11), up from 39% at the end of 2017.
The first is the promise of the UK Labour Party for the upcoming December 2019 UK General
 Courtesy: Nick Youngson CC-BY-SA
Election. The second shows the progress China has made in the last ten years. See graphs from the China report below.  You can compare earlier analysis I made of China progress (NBN Future Part 2), from 2012-17.


UK regulator, Ofcom, reports (Sept 2019, p.9) that 95% of premises can access a 30Mbps broadband (superfast) service, while only 54% can access a 300Mbps (ultrafast) service (about 16M premises). In contrast, Ofcom reports current broadband connection (by technology) Q2 2019 as:
  • ADSL 8.4M
  • Cable 5.3M
  • Other (inc FTTx; predominantly Fibre) 13.0M - but full-fibre homes (2018) - 1.8M(Connected Nations, p.1).
Another Ofcom report (International Broadband Scorecard 2018), comparing UK to EU countries and 18 other international countries (2017, Figure 1.8), breaks UK broadband down by technology and speed, noting VDSL is active in 35% of connections, cable 20%, ADSL 44%, full-fibre (1%). Also they report 14% of connections above 100Mbps (including 4% above 300Mbps), 42% under 30Mbps, including 2% under 10Mbps, and 44% in between (Figure 1.7).

The same Ofcom report (International Broadband Scorecard 2018; Figure 1.26) notes that at the end of  2017,  Japan has 98% full-fibre coverage, and Korea over 99% full-fibre coverage (Source: IHS MARKIT), and (Figure 1.27), meaning that close to 100% (98-99%) of households in Japan and Korea have access to the fastest broadband (over 300Mbps). China is reported having 45% of households who can access similar (ultrafast) speeds (US 61%, NZ 70%, DE 34%, AU 0%; Figure 1.27, Fixed Broadband coverage by advertised speed). In Figure 1.29, the percent of broadband connections over 100Mbps (end 2017) is reported, including Korea (76%), Japan, (74%), Sweden (78%), US (28%), China (14%), UK(14%), NZ (11%), AU (3%), among .
Interactive data is also available (full-screen required) providing international comparison.

China Fibre Broadband (2019)

Fig 11 Scale and Proportion of Fibre Broadband Users
- CNNIC China Internet Statistics 6.2019, p.11

Fig 10 Proportion of Subscribers of Broadband at the Speed of 100Mbps or above
- CNNIC China Internet Statistics 6.2019, p.11

China, in its annual Internet Statistics (CNNIC), recently updated for 2019, demonstrates a significant investment into fibre broadband. It also shows that China is close to completing a full fibre network with 395M fibre connection users; some 91% of all broadband users. Of these users, 77% are accessing broadband at 100Mbps or above (substantially more than the 14% reported in the Ofcom report - International Broadband Scorecard, at end 2017, mentioned above). When China (CNNIC) says fibre, they explain in a footnote (p.11) they mean FTTH/O - Fibre to the Home, or Fibre to the Office. An earlier post (Future NBN 1), shows China Fibre rollout 2012-17.

The Labour Party says the UK is falling behind in broadband, with only 8-10% of broadband users accessing full-fibre services, which they compare unfavourably to Japan (97%) and South Korea (98%). These percents turn out to be percent of households with access to, rather than subscribers of, fibre broadband. Japan has 60% households connected to full-fibre, whereas South Korea has 40% or 80% connected to fibre, depending how you classify one type of common Korean broadband connection.

South Korea

My previous examinations (NBN Future 2) of South Korea, showed in 2017 only around 40% on full-fibre, with 25% still on HFC and ADSL. A significant category called LAN, is likely to be FTTB (connecting the basement of residential buildings to fibre, and using LAN to distribute broadband within the building), had around 40% of broadband subscribers. Even if LAN and FTTH were both counted (about 80%), the total is substantially lower than the 98% the Labour Party suggests.
Recapping (Netmania), South Korea's broadband connections (2017) consist of:
  • LAN (presumably FTTB): 8.6M (40%)
  • Fibre to the Home: 7.7M (36%)
  • HFC: 4.0M (19%)
  • DSL: 0.9M (5%)
NB:See further Korea broadband data noted in the UK Ofcom report (International Broadband Scorecard) mentioned above.


The Japan Statistics Organisation - Statistics Bureau of Japan reports on telecommunications including fixed broadband services, for their 53M private households (Link, Ch. 2). The 2019 Statistical Handbook of Japan reports total 41M fixed-broadband subscribers (Table 8.6. Subscribers to Telecomms Services (2018), p.93 - totals to 42M, strangely):
  • Fibre to the Home: 30.3M (72%)
  • Cable Internet (HFC presumably): 6.9M (16%)
  • DSL: 2.1M (5%)
  • ISDN: 2.9M (7%)
NB: See further Japan broadband data noted in the UK Ofcom report (International Broadband Scorecard) mentioned above.While these figures represent significant users of broadband (Japan is placed 3rd in the world behind US 110M, and China 393M; p.94), rather than who can access broadband, the number of full-fibre users is substantially lower than the Labour figure, closer to 60% of total households, than the Labour Party suggested 97%.


What the Labour Party seems to be missing, is that while the UK is far behind Japan and South Korea, the UK is also substantially falling behind China. China's 395M full-fibre connections (Fig 11 above) is a significant proportion (86%) of their 456M total households (Wikipedia). Not only are fibre connections in place, but there are subscribers using the services, and 77% of those subscribers are consuming broadband at 100Mbps or more. In Summary, China's broadband is:
  • Fibre to the Home/Office: 395M  (91%)

So, the UK is looking ahead to a broadband future, where everyone is connected in 2030 to full-fibre (and 60-70% within five years; that is 15-18M households of total 26M; Wikipedia). Japan is 60% full-fibre connected there now (but every house can access fibre, per Ofcom). Korea is 80% full-fibre connected there now (if you count FTTB as full-fibre, only 40% if you don't). But China is close to already fully connected with close to 90% already on full-fibre, and with the result that nearly 80% of subscribers are using broadband at speeds at or above 100Mbps.

In contrast, Australia reports to OECD (end 2018), 0% broadband users at speeds at/above100Mbps (NBN Future 1).

Thursday, August 29, 2019

The Future of the NBN (5): Fast National EU Plans vs Fast Services

In the last post I looked at EU NBN (national) plans for 28 countries. These national plans nicely group into several different target clusters, including:
Compare: Pixabay

  • 30mbps coverage by 2020 (Many including Greece, France, Spain, Poland, Hungary, Ireland and 12 others )
  • 50mbps coverage by 2020 (Germany)
  • 100mbps coverage by 2020 (Sweden, Italy 85%, Netherlands 2023)
  • 1000mbps coverage by 2025 (Luxembourg 100%, Sweden 98%, Belgium 50%, Netherlands vast majority).
I wanted to compare these clusters against current Speedtest world speed rankings. When they didn't compare neatly, I thought about what other broadband indicator is going to give a better indicator of fast average usage, than a national target. That's when I thought to compare average speed (per Speedtest Global Index) with OECD count of number of very fast users (over 100mbps).

In national targets, Australia sits close to Germany, with 90% of fixed line aiming to get 50mbps, and rest 25mbps.

Let's compare.

NBN target (mbps)Cntry Speedtest Ranking  Speed Down Users/100 over 100mbps(OECD 2018)
30mbpsGR 96230
30mbpsFR 151085
30mbpsESP 171079
50mbpsAU 60 400
50mbpsDE 34736
50-100mbpsCN 2490>4
100mbpsIT 46518
1000mbpsLX 16107na
1000mbpsSWE 1411026
1000mbpsNE 209914
Table 1: Speedtest results (2019) - Ranked by NBN Target
The first table compares NBN targets for country with Speedtest actual speeds. The Speedtest rankings do not line up very well with the NBN targets for each country. France and Spain have low targets but high rankings. Italy and Australia have a quite high national target but quite low Speedtest rankings. NB: China has 4% superfast users - data is only found for gigabit users (post 2).
Now we retry the comparison, sorting by percentage of very fast users (over 100mbps). I include CN (China) data from my post below. OECD data (xls) comes from post below too.
NBN target (mbps)Cntry Speedtest Ranking  Speed Down Users/100 over 100mbps(OECD 2018)
50mbpsAU 60 400
30mbpsGR 96230
50-100mbpsCN 2490>4
30mbpsFR 151085
50mbpsDE 34736
100mbpsIT 46518
30mbpsESP 171079
1000mbpsNE 209914
1000mbpsSWE 1411026
1000mbpsLX 16107na
Table 2: Speeedtest results (2019) - Ranked by Users of 100+mbps (OECD 2018).
Source: Speedtest Global Index - July 2019, EU summary of NBN plans (post 4), OECD (post 1), China (post 2).
The second table compares adoption of over 100mbps speeds (OECD 2018) with Speedtest actual speeds.
This analysis suggests that the Speedtest ranking is more closely related to (1) how many users are on fast (over 100mbps) speeds than (2) what the country's NBN target is. Again Italy stands out as a high performer on number of very fast broadband users, but scores low average Speedtest results. This suggests there are very fast and very slow speeds available across the country leading to a lower average. Australia, Greece and Italy are the only countries not exceeding their national broadband targets (excluding the gigabit targets).
Conclusion: To improve your average speeds, you can either lift everyone's speeds, or push superfast access to a small minority. Small percentage levels of 1000mbps can substantially lift average speeds.

Tuesday, August 20, 2019

The Future of the NBN (4): EU collates their National Broadband Plans

As Australia's NBN approaches completion, some 12 years after Kevin Rudd took the NBN to an Australian election, many nations are now nearing completion of their own version. A European document, neatly compares 28 countries within the EU, and their national broadband targets. I summarise this list below.

Data: See the EU document (pdf - Slide 37), html. Full Report (pdf 328pp), and EU Strategy for a Gigabit Society.
EU National Broadband Plan - Targets
Some examples include:
  • Slow: Greece/Italy 100% coverage - 30mbps - 2020 (plus 17 others)
  • Usage: Greece/Spain 50% penetration (ie usage) - 100mbps - 2020 (plus 14 others)
  • Medium: Germany 100% coverage - 50mbps - 2020
  • Fast: Netherlands at 100% coverage - 100mbps - 2023 (plus six others similar eg Italy 85%, Sweden 95% at 2020)
  • Superfast: Sweden 98% coverage - 1Gbps - 2025 (plus Netherlands (vast majority), Belgium 50% and Luxembourg 100%)
  • UK 100% fibre coverage - 2033 (and 15M premises by 2025).

Comparing EU to AU

Looking at these clusters, Australia's NBN seems to sit closest to Germany, with 90% of fixed line network, getting at least 50Mbps, as well as the base 25mbps target for 100% coverage. Australia has neither a usage target (besides Corporate Plan targets of 75% usage) nor gigabit target. The current MTM allows up to 50% of premises to access gigabit once NBN activates upgrades to HFC, FTTC, FTTP.
Structured data (csv) follows: (MacOS 10.14.6 Numbers 6.1)

Country,Cover%,Type,Speed (mbps),Date,Comment
Greece,100,coverage,30,2020,"Inc Cyprus, Spain, Malta, Poland, Lithuania, Latvia, Italy, Czechia, Croatia, Bulgaria, Estonia, France (2022), Hungary (2018), Ireland, Portugal, Romania (inc 80% > 30mbps), Slovakia, Slovenia (last 4%)"
Greece,50,Penetration,100,2020,"Inc Bulgaria homes (Businesses 80%), Croatia, Cyprus, Czechia, Estonia (60%), Hungary, Ireland (inc 17-21mbps up), Latvia, Lithuania, Malta, Poland, Portugal, Romania (45%), Spain, Sweden"
Belgium,50,Penetration,1000,2020,"Inc Luxembourg (100% coverage, 500Mbps up), Netherlands (vast majority by 2023), Sweden (98% coverage by 2025)"
Denmark,100,coverage,100,2020,"Inc Denmark (30mbps up), Sweden (95%), Austria (99%), Finland (2025), Italy (85%), Netherlands (2023), Slovenia (96%)"
UK,100,Full fibre,n/a,2033,Inc 15M premises by 2025
// Source: - slide 37