Scarcity-based business models may elicit government intervention

Government intervention in the broadband market conflicts with free markets. It arises from fears of market failure as a result of:

• Ineffective duopoly i.e. oligopoly and too little competition.
• Underinvestment by the incumbent, the cable company, and challengers.
• Scarcity-based business model at both the incumbent and (if any) the local cable company.

About scarcity vs. abundance as the foundation of the business:

                                                                   

                         Scarcity            Abundance

Revenue driver           price               volume

Bits are like            wood, water         electrons

Bandwidth                throttle            maximize

Business focus           connectivity        VAS

Business model           usage-/speed-based  one proposition

Market                   saturated           growing

Innovation               none                maximize

Vertical integration     yes                 no

Infra-based competition  yes                 no

Structural separation    no                  national infra

Marketing, propositions  maximize            minimize

Customer focus           increase spending   increase satisfaction

Business model           subs, transactions  ads, transactions

Customer data            full                growing

It appears that telcos, cablecos and ISPs traditionally belong to the Scarcity group, whereas Amazon and Google are in the Abundance group. Netflix, Facebook and Apple seem to be moving from left to right. Microsoft and Nokia may be moving in the opposite direction.

If the governement believes there is market failure, or that Abundance is a better foundation for economic growth than Scarcity, then it has several options to intervene and contribute to broadband performance and coverage:

• Limit the effort to bringing broadband to rural areas where xDSL doens't work and cable is absent. But then, it could be left to the market to provide LTE services, or even satellite-based broadband.
• Facilitate all market participants by reducing red tape, providing loans, becoming an anchor tenant, etc.
• Participate at market terms to create a third infrastructure (FTTx). But do not expect existing players to simply quit competing and become service providers on your FTTx network.
• Regulate in such a way that competition truly increases. For example:

1. forbid the incumbent to consolidate the market, or even do any takeovers at all.
2. force structural separation upon the incumbent (see New Zealand). But raising wholesale prices is clearly a way of helping just the incumbent, and foregoing the interests of its wholesale customers.
3. nationalise (see Australia, NBN).

Dumb-pipe operator or dumb pipe-operator?

Two remarkable releases from the Dutch mobile industry:

• Vodafone NL CEO Rob Shuter on the company blog: "The multiband auction last December brought an unexpectedly large windfall for the Dutch treasury of EUR 3.8 billion. For us, however, this meant a solid bill shock of 1.38 billion! We were obviously fully aware when auction prices for mobile frequencies were established, but all parties agree that the prices were much higher than expected."
• KPN, Vodafone NL and T-Mobile NL are suing the Dutch government over the 2012 multiband auction. Vodafone claim that reserving spectrum for a newcomer created artificial scarcity.

It is rather ironic to have mobile phone companies complain about bill shock and artificial scarcity, because these are at the heart of the industry. In fact, they have sort of invented those terms.

As to bill shocks: mobile operators charge ridiculous prices for international roaming (which next year will end) and for some reason there is still no pan-European pricing (T-Mobile Austria charges €17 while T-Mobile Germany charges €96 for the same smartphone tariff allowances). Further, the companies are using the fact that most people didn't see high prices coming (we did), which created a buzz in the press about overpaying. But in reality, other auctions brought in more on a per MHz per capita per year basis. And the result wasn't impressive in relation to the 2011 UMTS auction. And last but not least: buying a 17 year mobile license creates an enormous entry barrier and comes at the expense of EUR 230 per sub or just over 1 euro per month per sub for Vodafone.

As to artificial scarcity: this is exactly what telecoms operators create, slicing & dicing available bandwidth (i.e. throttling) in order to create a 'business model'. Mind you, this whole throttling of a 24 Mb/s ADSL service to tiers of e.g. 1, 2, 4, 8, 12 and 24 Mb/s (or a 1 Gb/s service to 50, 100, 200, 500 and 1,000 Gb/s) is extremely expensive from a capex/opex point of view (equipment, billing, marketing) and is hardly warranted by the associated costs (a 24 Mb/s sub is probably not much more expensive than a 1 Mb/s sub).

Bayonette (Norway) abandons the artificial scarcity model for broadband

We have written about artificial scarcity in the broadband market before and are happy to see an operator addressing the issue quite so literally - Norway's Bayonette:

"Networks make a lot of money by selling stepped internet connections. We think differently, and don't want to milk customers by offering different service levels that are only marginally cheaper to produce."

What needs to be added here is that a lot can be saved on costs, if an operator abolishes throttling ('stepped connections'). The number of propositions goes down dramatically (to one), with obvious implications for sales & marketing, as well as billing.

Bayonette's offerings is much like Google Fiber's: 1 Gb/s for 500 NOK/mo and there is a free service (3/1 Mb/s).

FTTH vs. HFC is about opex, capex and timing

The FTTH/HFC controversy continues. Our views are always in flux and here is an update.

1. The government should stay out. And if they wish to interfere, there's only one way to justify this: nationalize the infrastructure, and separate the network from services.
2. We assume that an all-fiber network requires 'high' capex, but saves 'considerably' on opex. Evidence is growing:
• KPN states that FTTH opex is at least 30% less than copper opex.
• FTTH capex is continuously falling, as best practices grow.
• At Ziggo, opex and capex are continuously rising. Capex is doubling over the last 3 years.
• UPC states that EBITDA (OCF in Liberty Global speak) margins will structurally head lower.
• Small cable companies in the Netherlands, that have no shareholders demanding decent quarterly returns, all do FTTH. Some even skip Docsis 3 and go straight from Docsis 2 to FTTH.
• Netflix is taking a toll. It simply must.
• Better compression and other efficiency gains are nice, but even at a 50% improvement, they only buy 1 year of delay, since data traffic grows at roughly a 50% rate.
• Stratix put out a report that calculates what the roadmap, as laid out by TNO, would cost in terms of capex. Stratix claims the TNO gigabit (!) roadmap would be more expensive than overbuilding with fiber.
3. "Cable (HFC) will serve the market well beyond 2020". This statement from TNO must be read in 2 different ways:
• To say this, is to state the obvious. Any network can last. But one network will require more work (HFC) than others (FTTH) in terms of opex.
• Most importantly, it is a responce to an earlier TNO report, covering the 2010-2020 period. It suggested to some that cable companies would cease to exist on December 31, 2020. Which is of course total nonsense.
4. What the opex/capex implications of infrastructure choices are, ...
• ... is relevant to the cable operator. "Do we upgrade our old car, or do we buy a new one?", so to speak. It's a matter of timing.
• ... is a priori irrelevant to the end user. It may translate into a slower network and/or higher tariffs, and then the end user hopefully has an alternative to go to.
• ... is also irrelevant to the government (see 1.).
• ... is not irrelevant to shareholders, which is why public cable operators claim that they have future-proof networks.

UPDATE December 3, 2013

• Supply side: is about fiber (or 5G), future-proofing, skipping interim technologies (VDSL, Docsis 3, ...) versus legacy
• Demand side: expected traffic growth and business model (scarcity or abundance)
• Customer's role is limited to becoming a subscriber or quit.
• Shareholder's role is limited to buy or sell shares.
• Government role is to regulate, facilitate and when the market fails: intervene.
• Management's role is to make a choice between scarcity/minimising capex and abundance/maximise capex; or strike a balance between customer interest (maximise capex) and shareholder interest (minimise capex).

Doing FTTC is ignoring a family of elephants in the room

The reasons to do FTTC/VDSL/vectoring instead of FTTH are obvious: they are about capex (cheaper) and demand (supposedly there isn't any). But when BT, DT and the Coaliton in Australia gamble on FTTC, it creates a family of elephants in the room. If you want to do FTTC, that's fine, as long as you acknowledge that it has a few issues.

1. Cost
It may save on capex, but FTTH saves on opex. And if FTTC proves insufficient, it risks being a 'regret investment'. It is going for the short term, instead of the long term. Put differently: the greatest benefit of FTTH is its opex savings of around 20%.
Further, vectoring is still quite experimental. It is completely unclear what the real-world performance will be and which share of the lines it can be applied to.

2. Demand
Doing a bottom-up assessment of the number of devices per household and the usage per device may render a picture of the demand side, but it completely ignores innovation and unexpected use cases. 4K is moving faster than previously expected and gaming may be a major growth driver.
To be fair, ongoing compression is relevant and also the fact that we are moving from a downloading to a streaming world. And streams are offered at a certain fixed rate. Making the access network 'faster' doesn't help. And the access network isn't the only bottleneck in the system.

3. Quality of service
Any bandwidth will suffice for anybody, as long as we don't care about QoS. As demand rises and networks aren't upgraded at a similar pace, QoS will go down.

4. Geography
If one could isolate the 30% (or whatever it is) that would be willing to pay for a gigabit, it would be easy to roll out FTTH there, and FTTC everywhere else. Only this doesn't work for several reasons: people move every 7 years or so; building nationwide FTTH in any country will probably take 10-20 years; and the 30% that needs a gigabit changes every month, so in reality it is probable closer to 50% (or whatever it is). Compare Cisco's statement: "the top 1% is actually the top 5%".

5. Business model
Traditional telcos are rooted in scarcity. They will upgrade not ahead of demand, but somewhat behind demand (in such a way that complaints are controllable). It is their game to minimise capex in order to maximise shareholder remuneration. And capex peaks are to be avoided, when shareholders must be reported to on a quarterly basis.
Google on the other hand is rooted in advertising. Governments, like Google don't want bandwidth to be a scarce resource. Google because it wants to maximise usage in order to create advertising inventory, governments because of the economic, social and environmenal benefits of migrating online.
Both business models can be applied, but looking at it this way explains why others besides incumbents are getting involved.

6. Specs
FTTH is not about download speeds alone, but also about the uplink and latency. And if it's not GPON but point-to-point, it is also dedicated instead of shared.

7. FTTH has a lot of supporters
Governments such as the current one in Australia, Google, KPN (sees FTTC as an interim technology), challengers such as Free and HKBN, several cable companies, etc. prefer FTTH over FTTC for any or all of these reasons.

Let's just say that traffic growth will remain 100%

Rudolf pointed me to this great Ars Technica article. I included it in the right hand column (under 'Favorite Articles') and here is a link to a Google Doc for easier downloading (Ars Technica allows the .pdf file to be downloaded to paying subs only, but hey: Content is not king anymore).

First, here are the main points:

• There is no 'exaflood' (Bret Swanson), since traffic growth is at a manageable (for backbones, that is) 100% sort of level, and actually falling. Put quite differently: it is here already, since worldwide traffic totals 3-5 exabytes/month. In itself, this undermines arguments aimed against net neutrality. Capacity can rather easily and cheaply be upgraded.
• The last mile is the real bottleneck. In this respect the US is falling behind (even if there are different points of view) because of a lack of competition (no line sharing). South Korea has almost as much traffic as the US, or 6-7 x more per capita.
• Media companies (such as the Hulu venture) face exploding bandwidth costs. They will flee to P2P delivery systems.
• Andrew Odlyzko suggests ISPs should stimulate usage rather than limit it. In David Isenberg's words: "In other words, the problem is completely mis-framed. Comcast and Verizon -- and even Net Neutrality Advocates -- are talking how to manage scarcity. We should be talking about how to achieve abundance."

Second, let's make another effort at pinpointing sources of traffic growth and reasons to build FTTH networks. I don't know if growth will keep falling, stay around the 100% level or actually accelerate (to the 300-500% level Cisco's John Chambers predicts), but there is a lot yet to come:

• Video. Include 3-D, HD, holography. Not 'just' YouTube, but also VoD, streams, Hulu and BBC's iPlayer, place-shifting (SlingBox, PC-to-TV boxes such as Daily Media, etc.).
• User-generated content.
• Gaming and virtual reality. Massive multi-player online gaming, which depends on very low latency. Second Life moving to a level of realism we know from Pixar.
• Cloud computing, teleworking, telepresence, e-health, e-learning, monitoring.
• P2P (even though the effect will be mitigated somewhat by P4P). Includes legal filesharing. Traffic on munifiber explodes when on-net filesharing among subscribers is enabled.
• In other words, there is truth in 'Field of Dreams' after all: 'Build it, and he will come'. Don't forget about new applications!
• There are some cross connections. Many of the above require symmetric connections, i.e. FTTH. Also, 3-D, HD and holography will be coming to gaming, telepresence, monitoring etc.
• Not to mention regular growth contributors and enablers such as increasing penetration (Internet, broadband, PC, credit card), emerging markets coming on board, etc.
• And from a different perspective: don't forget all the social, economic and environmental benefits of true broadband (i.e. FTTH) networks.

2025 AI outlook (source: perplexity)

Source

AI Transforming Industries

By 2025, enterprise AI adoption is expected to surge, with predictions indicating that over 70% of businesses will implement AI technologies. This widespread adoption will be driven by advancements in multimodal learning and contextual awareness, enabling AI to deliver more nuanced and industry-specific solutions. In the education sector, AI innovations will focus on personalized learning paths, AI tutors, and smart content creation, revolutionizing the way students engage with educational material.

Key trends in enterprise AI and industry-specific advancements for 2025 include:

• Integration of AI into a wider range of industries beyond traditional tech sectors.
• Automation of business processes through AI and robotics, streamlining operations across various sectors.
• Implementation of AI-powered enterprise search systems to enhance knowledge accessibility within organizations.
• Transformation of customer experience (CX) through AI-driven personalization and predictive analytics.
• AI-enabled advancements in healthcare, finance, and manufacturing, with a focus on improving efficiency and decision-making processes.

AI Agents and Automation

By 2025, AI agents are expected to become more sophisticated and autonomous, revolutionizing how we interact with technology and automate complex tasks. These AI-powered agents will evolve from simple assistants to proactive problem-solvers, capable of handling a wide range of digital tasks with minimal human intervention. Multi-agent networks are predicted to emerge, allowing AI systems to collaborate and tackle more complex problems collectively. This advancement will lead to the development of vertical agents specialized in specific industries or tasks, similar to how SaaS platforms revolutionized software delivery.

Key developments in AI agents and automation by 2025 are projected to include:

• AI agents managing aspects of users' digital lives, such as scheduling, email management, and information retrieval.
• Increased autonomy in decision-making processes, with agents capable of executing tasks based on high-level instructions.
• Integration of AI agents in enterprise environments to streamline workflows and boost productivity.
• Enhanced natural language processing capabilities, allowing for more intuitive human-AI interactions.
• Emergence of AI-powered personal assistants capable of handling complex, multi-step tasks across various platforms and applications.

Advancements in AI Models

By 2025, AI models are expected to become more sophisticated, with a focus on multimodal capabilities and enhanced reliability. Multimodal AI, which can process and analyze various types of data simultaneously (such as text, images, and audio), will deliver more context and nuanced understanding in applications across industries. These advancements will lead to the creation of more realistic and dependable AI systems, thanks to innovations in security guardrails and evaluation frameworks.

Key developments in AI models by 2025 are projected to include:

• Improved large language models (LLMs) with enhanced reasoning capabilities and domain-specific knowledge.
• Greater emphasis on AI safety and ethics, with models designed to prioritize human values and long-term impact.
• Integration of AI models in critical decision-making processes, requiring robust evaluation and testing methodologies.
• Continued research into artificial general intelligence (AGI), though full realization is not expected by 2025 despite claims from some tech leaders.

Contrasting Expert Forecasts

Amy Webb and Gary Marcus, two prominent figures in AI research and forecasting, offer contrasting perspectives on AI's trajectory in 2025. Webb predicts a convergence of key technologies, including AI, biotech, and advanced sensors, leading to what she terms "living intelligence". She anticipates a shift from prompt-based AI to more conceptual interactions, and warns of potential data scarcity as high-quality training data becomes exhausted.

In contrast, Marcus maintains a more skeptical stance. He predicts that artificial general intelligence (AGI) will not materialize in 2025, despite claims from tech leaders like Elon Musk. Marcus also anticipates that AI "agents" will struggle with reliability outside of narrow use cases, and that profits from AI models will remain modest. Both experts emphasize the need for careful consideration of AI's societal impacts and the importance of addressing issues like bias and power concentration in the tech industry.

The case against technology-neutral spectrum

Rewheel showed that 'independent challengers' add a lot to competition in mobile markets in Europe. The E5 (DT, FT, TEF, TI, VOD) have a combined market share of 80% across 27 nations. There are indie challengers in 14 of those. In the other 13, minimum prices can be 140% higher.

Competition in telecoms is limited as a result of an extremely high entry barriers. In fixed markets, it is legacy infrastructure held by the incumbents. In mobile markets the scarcity of licenses is added. Hence regulation.

Take note that Vodafone is an incumbent on the mobile market, just like Bouygues and SFR. Tele2 also is an incumbent, at least in markets where it joined when 2G (and 3G) licenses were tendered. Now Tele2 is entering the Dutch market on the back of winning a 4G license (not counting its presence as an MVNO). So, Tele2 NL will most likely act as an indie challenger. Hutchison 3G was a challenger in the 3G era, but may not act like one when 4G licenses are issued and a new entrant appears.

Now, there's always scope for an incumbent/E5 operator to behave like a challenger. One could call this atypical, in light of the Rewheel study. For instance, at the latest strategy update, DT told T-Mobile NL to behave like an 'unconventional attacker'. Hence, T-Mobile NL may display atypical behaviour as well.

Back to the entry barrier. It has been lowered considerably by allowing the timing of issuing new licenses to coincide with rise of new technology (4G/LTE). So, Tele2 NL enters the market as an MNO and at the same time can use all the advantages of LTE.

The question now is: is it wise to make spectrum technology neutral? The regulator decided to do so, which at first sight seems to make a lot of sense. However, by the time the next auction will take place (2029 - not counting the 1900/2100 auction probably taking place in 2016), new technology will be around (5G) and all 4 MNOs will have introduced it. So, the new entrant coming to the market in 2030 will not have a technology advantage like the current wave of 4G entrants have.

War of the words continues between fiber and copper

Recently, the Dutch cable sector (i.e. Ziggo, UPC and NLkabel - NOT the smaller cable companies and CIF) decided to step up its lobbying efforts against the fiber lobby (led by Reggefiber/KPN). It all seems to come down to the use of the words 'future-proof' (toekomstvast or toekomstbestendig). Now, unfortunately, Hartwig Tauber of the FTTH Council Europe, decided to enter into this silly discussion, claiming that FTTH is the only future-proof network:

“Fibre to the Home is de enige toekomstbestendige oplossing. Consumenten profiteren nu al van glasvezelverbindingen die 1 Gbps en meer bieden voor het up- én downloaden. Bovendien is Fibre to the Home in staat een niet te kloppen servicekwaliteit te bieden. In plaats van consumenten te misleiden met maximumsnelheden kunnen Fibre to the Home-netwerken aan ieder huishouden een gegarandeerde snelheid leveren. Operators die gegarandeerde snelheden van 100 Mbps en meer aanbieden, bevestigen dat klanten tevreden zijn met die snelle verbindingen.
Er zijn echter nog steeds spelers op de markt die stellen dat hun koperoplossing toekomstbestendig is, omdat consumenten geen hogere snelheden nodig hebben. Het logische gevolg hieruit is dat deze partijen hun klanten voorschrijven wat toekomstbestendigheid inhoudt en de ontwikkeling van innovatieve diensten belemmeren, waaronder diensten met grote socio-economische voordelen.”

Of course, this is total nonsense. No single network is future-proof. Every network needs work: maintenance, repairs, upgrades, etc. And it all starts with the question: is this about the passive or the active components? Fiber appears to be the way forward, but you can never tell; and 100 Mb/s equipment seemed to be enough, until Reggefiber decided to upgrade to 1 Gb/s - with implications for backhaul (i.e. fiber) as well.

And then there is the use of the words future-proof: linguisitically (semantics), it is incorrect to talk about 'more' or 'less' future-proof, in the same way as it is nonsensical to say that something is 'more perfect' or 'less perfect'. Something is either future-proof, or it isn't. And no single technology is future-proof.

We have been proponents of FTTH for many years, but the discussion has to be fair. NLkabel and the FTTH Council are bending the truth (to say the least). And this isn't helping. Of course, fiber is better than HFC: more capacity, easier to upgrade, symmetric, open. But FTTH is on the roadmap of HFC as much as it is on the roadmap of copper networks.

Unfortunately, Hartwig's statement fails on more points:

• He forgets to mention openness.
• He talks about bandwidth guarantees, but networks are all shared at some point in their architecture.
• He doesn't distinguish between GPON ('shared') and WDM-PON and p2p ethernet fiber (dedicated) networks.

We have written on the subject twice recently. In summary, there are three points to be made, two of which are mostly overlooked:

1. FTTH beats HFC in many respects (see above). Cable operators are billing their networks as NGNs, which is total nonsense. One needs to distinguish between the passive and the active networks. Every network can be upgraded. FTTH is on the roadmap of all.
2. There is a sharp distinction between the advantages for users (QoS, real estate value), for operators (capacity, opex, new business models) and for governments (socio-economic and environmental benefits).
• Fact is, that many cable customers still are happy customers.
• Fiber operators are trying to make the network a worry for their customers ('choose the network that is most future-proof'), but the customer shouldn't care. The network is the operator's worry. Fiber operators also claim that cable networks cannot keep up, but again, that is the cableco's worry. Or that they can't afford it - which is probably true, but let John Malone worry about that. If their upgrades start falling behind demand growth, the market will do its work and subscribers will migrate to FTTH. There's one problem though: FTTH penetration. In areas where FTTH will not be deployed for many years, HFC only competes with DSL and hence has no incentive to move to all-fiber, which could create a digital divide.
• The government has no voice, because it decided to leave telecoms to the market (apart from rural areas, where the market fails), and so: either shut up or create a national infrastructure, either the Australia (nationalisation) or the Singapore way (tender).
3. The way the market works, is that capacity is added as demand grows. Business models are based on scarcity. Moving to an all-fiber gigabit network is a big leap and only makes sense when other forms of upgrades make no sense from a return point iof view, or when new business models, based on abundance, are explored.