Telecoms is dying. Yippee!

TMT. Telecoms. Media. Technology. It’s such a familiar moniker. There are endless people with job titles of “TMT analyst”, copious professors of telecommunications law, and roving hordes of telecoms regulators.

Which is all somewhat odd, as telecoms is dying. It looks decidedly pallid, and is about to become thoroughly putrid. The industry that acquires the name “telecoms” is slowly going out of business. And that’s just great!

To avoid accusations of hyperbole, I need to clarify what do I mean by “telecoms”. We still need physical infrastructure, and “holes and poles” have value. It is just that you will have to look in the utilities section of the newspaper to find it. It is the active layer is in the process of being absorbed by the computational cloud borg, as if some kind of sci-fi horror is taking over.

By “telecoms” I mean the historical “Telco 1.0” business. This sells circuits (or circuit-like products) characterised primarily by bandwidth, but may come sub-divided into minutes or messages. This is still the great bulk of today’s revenue: voice, SMS (as a voice by-product) and nearly all data.

Kapoof! The sweet profit in these is gonna get Blackburied, leaving just the costs as a bitter aftertaste in the mouth of burnt pension funds.

I have some track record in making such prognoses:

  • In the very early 1990s, now lost to the mists of time, I had a 45Mb hard drive, and immediately saw the music industry’s profits would collapse from data compression plus rapidly falling costs of storage allied to east-to-copy digital formats like the CD.
  • I started this telecoms futurism game in 2002 with a widely-ignored strategy paper inside Sprint on the “OTT threat” that correctly forecast doom for voice and messaging revenues.
  • I consulted to Motorola back in 2004 on this, then Nokia in 2005, and made a prescient analysis of what to do about the OTT phenomenon. They didn’t take heed, and they are finished with. Don’t believe me? Read the report here.
  • Go to the Wayback Machine for my archive of Telepocalypse blog posts from 2003–2006 for much good sense on how apps like Skype would rise (and fall due to lack of social media features).
  • In 2006 I predicted ‘net neutrality’ was technical nonsense and I was proven right in 2015.
  • When at Telco 2.0, I effectively forecast the rise of SDN and SD-WAN, even if I didn’t have name for them, with the slice and dice approach.
  • My predictions from 2009 when I was strategy director at BT are also looking pretty robust, when I said that the world would coalesce around a small number of regional or global cloud comms and commerce platforms.
  • In 2012 when I co-founded the Hypervoice consortium I said voice would be the next big thing well before that was fashionable and Amazon’s Alexa stole the show.
  • I’ve been way ahead of the curve on 5G/SDN (with RINA) and network performance science (with the quality attenuation framework and ∆Q calculus), which are slowly gaining ground in the R&D community.

Let’s take a look at the many factors that indicate it’s coming time to put “telecoms” into palliative care, and to start shopping around for undertakers.

The first is technical. We have seen a long history of tussles between computer industry and telecoms. Think of OSI, ATM, SONET, and 3G versus Ethernet, TCP/IP, and WiFi. In this process “nobody ever lost their job betting on the computing side”. This is now coming full cycle. Telecoms is just a form of computing that calculates the identity function at a distance, and its only observable effect is to impair the copying of information (when compared to the ideal of being instant and perfect).

That resource allocation is coming under increased software control. Technologies like SDN and SD-WAN allocate resources and choose who gets what impairment. The can combine multiple routes and bearers, and defer demand by microseconds to months to match available supply. What we are really building is not a “telecoms” network any longer, but an “ultracomputer” or “hypercloud”. This has a distributed OS for the cloud/edge compute and inter-process communications.

Unfortunately for investors, the telecoms business is at the losing end of this change. There are many mispriced and misallocated resources. The telco SDN “trading platform” offerings are feeble compared to the profit-sucking “high-frequency traders” of the SD-WAN business. So just as the Telepocalypse of the last decade was voice and messaging, that of the next decade is most enterprise access revenue going down the drain.

The progress of science and technology is also not in favour of the industry. Up until now there have been handsome rewards for generating more bandwidth, with technologies like DWDM, beam-forming, or CDMA. These transmission advances typically appeal to a physics mindset, with some math-dependent advances (like OFDM).

We are now seeing a shift: more bandwidth cannot solve your resource control problems. The speed of light isn’t changing, and the state of stochastic packet networks is evolving too fast for present control systems to manage. The process of throwing bandwidth (i.e. extremely idle assets) at all “congestion” problems is going into reverse, as you merely arm end users with bazookas to hose their backhaul and attack their neighbours.

This is a fundamental shift in the nature of the value delivery constraint, from resource creation to resource allocation. These new technical issues can be fixed with a new science drawn from supercomputing, safety-critical and distributed systems. However, it is not a traditional telecoms skill set, and vendors are struggling to understand why they are hitting scaling barriers.

It involves a new mathematics, too, which is rather like the jump from classical to quantum physics. The vendors don’t have the humility to learn yet as they haven’t suffered badly enough from the poor scaling and unpredictable performance engineering. Nobody wants to admit they haven’t got a clue what to do, and their PhD isn’t helping matters.

So, what else is fucked in telecoms? Ooh, let’s see. I’ll take the product set next. All service industries are facing a transition from fee for service to fee for value. This means selling fitness-for-purpose and articulating value (and making promises) in the customer’s terms, not your internal metrics and measures.

The good news is that this does exist in telecoms for the service wrapper, with phone support levels and tiered times to repair tailored to consumer, small biz and enterprise needs. The bad news is that’s about all there is.

What is needed are varying and segmented levels of data service resilience and performance, that can be tied (loosely or tightly) to delivery of some kind of application outcome or experience. The difficulty is that if you can’t describe the minimum quality to expect, or say what it’s good for, or assure the outcome (however weak), then you can’t charge for the benefit. If the alternative is to persistently and wildly over-deliver performance (hence bloating cost) then you will go out of business.

So the telecoms business faces a shift from selling “quantities with quality” to “quantities of quality”. The catch is that every existing OSS/BSS system is locked into an obsolete bandwidth model. There is no box in the product catalogue to capture the quality and performance parameters. Nor is there a system to assure them or bill them. This, to my mind, suggests a fundamental collapse and take-over by cloud providers is coming.

Speaking of whom, let’s spend a minute reviewing the commercial context of telecoms. We are seeing rapid growth of the scale, scope and value of giant cloud platforms. You can think of Netflix (media), Amazon (general compute & retail), or Microsoft/IBM (verticals).

Payment for communications is going to increasingly come from the cloud providers and their customers, via wholesale mechanisms. The user transaction originated with one of the applications in these clouds, and the money flows out from there. The network’s core destination is not a physical location, or another network, but a compute and commerce and comms ecosystem from one of these vendors.

This represents a paradigm shift in business model to multi-sided market platforms. (I told you so, again, although many seem confused by the definition of a multi-sided market.) This is simply inevitable as the softwareisation of telecoms means it is all down to who owns and controls the “invisible engines”.

This blows up the financial model on which investments in telecoms are presently made. These are based on typical infrastructure play, with a “landlord and tenant” model using long-term “rental agreements”. These contracts fail to capture the dynamic nature of the market (think of apartment rentals vs Airbnb).

As such, the profit moves to places where supply and demand are matched up with greater speed and accuracy, and priced to clear the market. This Uberisation of telecoms is already happening, and is going to gather speed as more start-ups seize the opportunity.

The tragedy about to unfold is that telecoms business asset values price in neither the downside risks (you’re now the Uber driver), nor the upside opportunities (you’re the restaurant whose high-margin alcohol sales go up as posh people don’t need to drive home or slum it in a taxi).

Furthermore, there are large network arbitrage opportunities in all markets, both in terms of service pricing and regulatory regimes. As a result, there are likely to be many mispriced assets, resulting in a huge buying/shorting opportunity as the centres of power shift. Nice work to get if you have the tools to see and exploit these. (Hint: I have a map of where the buried treasure is.)

If that wasn’t bad enough, the regulatory system appears determined to pretend that this all isn’t happening. We are seeing in a variety of cases where common carriage (and the circuit mentality) is being misapplied to a distributed computing system. The legal scholars involved in this mistake don’t have the epistemological tools to understand they don’t understand. They certainly don’t have the scientific knowledge needed.

The result is that regulators are tasked with regulating an industry that is disappearing and being subsumed into another. This brings into question the legitimacy of their scope of actions, as the enabling laws cease to have relevance to the world. It’s not just telcos and their vendors that needs to adopt new science, skills, metrics, and processes. Those in governance need a fundamental “upgrade” if they are to identify economic bottlenecks in hybrid ecosystems of telco, cloud, comms and commerce.

So if not “telecoms”, what comes next? We clearly need some industry that puts the “distributed” into “distributed computing”! This is a new “transputation industry” with data centres, network controllers as well as transmission facilities. I’ve discovered over the years that advice is only taken seriously when it is reassuringly expensive. I’m not going to write down all the answers for free here.

Whilst it might be too early to hold a wake for telecoms, it’s definitely time to stock up on scented candles in the sales. Not a dirge, but a fanfare. For what replaces telecoms (and cloud) is a distributed computing industry that is far more responsive to customer demand, responsible for the quality of its products, and efficient in its operation.

Telecoms is dying. Let’s celebrate its demise with gratitude and anticipation of good things to come, not sadness nor regret for its ending!

Additional related reading