A Tale of Cloud Duopoly & Government Regulations in 2020

13 May

On Thursday (5/9) we held the Cloud2020 event at the Switch SuperNAP. The goal of the event was to discuss and ponder the state of IT infrastructure/cloud in the year 2020. The structure of the event was meant to ensure that there was room for discussion and debate.  In this blog I’m specifically calling out the debate created during the “Programmable Infrastructure” panel  with Pete Johnson of Profit Bricks, Randy Bias of cloudscaling, Jonathan Murray of Warner Music Group, George Reese of Entratius and moderated by Larry Carvalho.

Great Panel

It was during the Programmable Infrastructure panel that the crowd of Clouderati finally warmed up to the idea that debate was appropriate and expected.  Most  of the more boisterous debate was in regards to a comment from Randy Bias. Randy was responding to a question posed to the panel about the future of cloud service providers. Randy stated that he believed we would end up with a duopoly of Amazon and Google. He strongly believes that their existing scale, ecosystems and mindshare mean they will continue to be the dominant cloud forces in the year 2020. To be fair, Randy wasn’t the only one who voiced the belief in a cloud duopoly. However, there were a good percentage of the participants (yours truly included) who commented otherwise.

Why is the Duopoly Bad & Unlikely (I hope) to become real

I’ve written before on GigaOm that a monopoly in the cloud marketplace would be very bad for business and especially innovation.  I also state in several places, including Joe Weinman’s book Cloudonomics that the reality of IT infrastructure variation means we’re unlikely at this early stage of cloud infrastructure development to be able to call out a winner.  A recent blog by Pete Johnson (also from the above panel), has many good points on this issue as well. However, in this particular case I want to focus on one theme and that is the potentially negative impacts of a duopoly.

I think most of us can agree that IT is becoming an ever greater part of our daily lives and as such more critical to our ability to operate, manage our cost of living and generate opportunities to innovate.  Well, if history is to be believed (not all of it is), a duopoly or monopoly and or regulated market in any critical segment of the marketplace is potentially fraught with peril for the buyer.  As buyers we have significant experience buying from industries and monopolies that are regulated. We buy products from the banking industry, we fly on airplanes, eat commodities like beef and oranges, we use electricity, and water and so on. However, not all regulated markets have the same barrier to entry that power, water, or railroads do. Imagine trying to run your own power lines or railroad tracks today, cost prohibitive isn’t the word.  So while a railroad is regulated as is a bank or an airline, they are not the same thing. You can buy a building, obtain a banking license and open a bank. You can buy a few jets, obtain licensing and routes and start an airline, it’s not so easy to do the same in railroads or power. All regulated industries, not all the same. My concern that cloud might become a duopoly stems from the real potential for a weakening focus on innovation (as competition is removed) and the guaranteed “success” (profit) from being regulated in a duopoly/monopoly market.

Regulated to protect the buyer vs. for guaranteed service survival

In most areas of the country there is no real competition for railroad service to move your cargo. However, you can easily find any one of 20 different banks in a decent sized city. Yes, both are regulated, but not the same situation.  In the case of the bank they are regulated (theoretically) to keep from misusing your money and to provide you with some sense of security that your money will be returned when you need it. Not so in the railroad industry. While they too have regulations around safety, etc., their regulations are often established in a way that guarantees a healthy profit (like the power and water delivery markets).  If you don’t believe me, I highly recommend you read David Cay Johnson’s book “The Fine Print”, you’ll be angry, but you’ll understand (U.S. market specific).

The Regulated Duopoly and the end of Innovation

I strongly believe that if as suggested the year 2020 brings us a duopoly of cloud providers, we run the real risk of the government recognizing this service as critical infrastructure (I.e., railroad/power delivery).  With a duopoly we already carry the possibility that innovation and continued pricing pressure will be a thing of the past, but even worse we could have the lack of innovation and fixed pricing regulated into place by the government.  One of the points made by Randy Bias was that the barrier to entry for real competition against Amazon or Google would be the existing scale. Imagine if that barrier to entry is further heightened by regulatory controls.

Guaranteed profit and market position do not equal Innovation

While I tend to disagree with the presumption of the cloud duopoly being a done deal for many reasons, I do fear it’s potential. The good news here is that as stated by several others already we are still in the first inning of the ball game that is cloud infrastructure. I’m very optimistic that some of the innovative smaller players in the market today will grow to become forces for change and further innovation going forward. As only through an open and competitive marketplace can the buyer truly hope to get the best service for their hard earned dollar.


Posted in Uncategorized

  1. Greg Ness

    May 13, 2013 at 5:18 pm

    When you compare the railroads to IaaS you are comparing transport to destination. The cloud transport marketplace (think cloud migration) is very immature, even relative to emerging IaaS marketplace. So perhaps a better analogy would be that there will be a few cloud migration players (which would include VMware and others) and perhaps a robust IaaS marketplace driven by specialty services, low cost, security, platforms, etc. The “duopoly” would have to be very enterprise-centric as well… including giving enterprises the ability to orchestrate across facilities/clouds as needed, including between IaaS options. Some might also suggest that enterprises will keep certain apps and databases under tighter operating oversight and that specialized vertical platforms of all kinds will emerge. All of these factors bodes for a more diverse, fluid ecosystem of leaders, followers and niche players than say a hardware/semi business where manufacturing can scale perhaps more easily than specialty services.

    A thoughtful blog as usual Mark.

    • mthiele

      May 13, 2013 at 5:27 pm

      Thanks for the great comments Greg

      The comparison to the railroads was an attempt to compare product/service types but rather businesses that are critical and may even appear to be fighting for your dollars, when in reality their profits are virtually legislated. I think your other comments are fairly accurate (as accurate as anyone who’s looking 7 years out). However, my blog is merely responding to the assumption made by several panelists/attendees that there would be two dominant cloud players (what ever that means), with maybe a few very small players having niche rolls and therein lay my concern. I would much prefer to see a wider set of truly competitive companies who through real competitive situations stay ahead of markets and continue to provide customers options and innovation.

  2. @parkercloud

    May 13, 2013 at 6:04 pm

    datacenter scale as a benefit only works to a certain size and as servers, storage, routers, etc. become faster / resource dense smaller datacenters scale as well. Faster cheaper bandwidth will support smaller faster connected data centers and eco systems. Smaller datacenters equals closer datacenters equals lower client latency. More datacenters to support an infrastructure reduces redundancy cost.. Locality increases mindshare, If nothing else nature abhors a vacuum. Agree with railroad as a needed service but DCs in the future will be built faster less capital expense in comparison to starting a railroad.

    • mthiele

      May 13, 2013 at 6:51 pm

      Rick, thanks for the comment.

      I’ve written at length about the potential for lower tier and maybe smaller data centers as a potential future with greater adoption of cloud. The problems with getting to that point are myriad and complex and dependent in many cases on incorrect assumptions regarding what’s possible with a data center as far as it’s cost/performance vs. size and speed to deploy or modularity.
      – The vast majority of infrastructure is still not designed as pure cloud (legacy apps/legacy infra designs etc). There will likely still be a high percentage of environments that are not easily distributed or made more portable by 2020.
      – Big data will create ever greater gravity for putting people and applications near the data. The greater the size of the data the more perceived value and the more people/apps are drawn to it, also the better it needs to be protected (not conducive to lower tier or smaller data centers).
      – There is an assumption that you need to sacrifice availability and capability to get the right cost and or efficiency (At Switch I believe we’ve proved that theory incorrect).
      – The benefit of the ecosystem is the proximity to people and services. Having your environments distributed puts you at risk regarding network diversity/options. It also limits your ability to attract partners. Lastly, a key benefit of the ecosystem is that it creates net new options/ideas that would otherwise have been missed.
      – It is easier to establish higher availability, security and physical/environmental protections with the economies of scale associated with very large facilities.

      There are other benefits/risks, but I personally think the above is enough.


  3. Matthew Povey

    May 14, 2013 at 1:28 pm

    Regulation is usually a double edged sword for industries that come under it. In return for having to do things in a certain way, the regulated are generally protected from new entrants and / or external competition. These protections can be explicit or implicit but they have the same effect of reducing competition overall, as you describe well. It usually begins with the best of intentions – to engineer away some form of real or perceived market failure. Over time, regulation then goes in one of two directions.

    It can become sclerotic, as with the US railroad industry which found itself both protected from (in agricultural goods) and disadvantaged against (in general goods), road-based freight. Alternatively, in rapidly changing industries such as finance, we find the response of regulators has been to layer on ever more byzantine rules that serve (as in Basel III) to protect and advantage incumbents against new-entrants.

    So what can we expect for the cloud in 2020? Thankfully, the IT infrastructure industry has largely avoided direct regulation. Rather, the regulation we work under tends to accrue at the business and application layer. The reason being that in IT it has been generally understood that only the business can know what the impact of regulation is. When IT has tried to act as the interpreter of regulation (as opposed to designing systems based on a business interpretation), the results have been bad. In IT we will build governance that ensures cloud suppliers are providing the service levels required of applications our businesses build or we build for them. Provided the cloud suppliers are transparent about what they provide, that delivers a clear line delineating where regulation should sit.

    So would a duopoly of cloud suppliers change that dynamic? I don’t really see why it should. Provided the manner in which resilience is provided is understood by customers, there is no reason the regulatory burden should not continue to fall on the businesses that use cloud infrastructure, leaving plenty of space for innovation at the cloud supplier layer. In fact, if we can achieve that situation, there is no reason to think that a duopoly would emerge and if it did, that it would necessarily be a bad thing. After all, resilience requirements arising from regulation (e.g. DC distance separation) haven’t stopped the growth of AMZ against existing infrastructure suppliers. Regulation on cloud suppliers would go the way of the railroads, driving sclerosis and new growth (e.g. road and air freight) rather than the byzantine and protective complexity of financial regulation.

    Ultimately, this discussion is a subset of the wider question of how much regulation should exist and where it should sit. In general, I think the less regulation we can have apply to IT infrastructure itself, the better. The role of IT organizations in translating the regulation driven needs of businesses and applications into infrastructure services will be key to ensuring that is what we get. If we do our job well, the regulators will keep their eyes where they should be. If not, we’ll find them turning their attention where it is not needed. In other words, it is on us!

    Note that even in the EU, privacy regulations apply to businesses, not cloud suppliers. AMZ etc. respond to those regulations on the basis of what their customers demand of them (e.g. a ‘local’ presence is required – which raises the question of what role protectionism plays in all this).

    • mthiele

      May 14, 2013 at 1:40 pm


      Excellent (not that you need my approval) and well thought out response. Personally, I hope that you’re correct and really only posited the idea of regulations potentially being applied in the event a cloud duopoly occurs as “something to consider”. However, my primary thinking on the risk of regulations (outside the risk to pricing and innovation of a duopoly) was that with only two providers an incredible amount of the countries critical systems would be supported by just two companies. Imagine the risk if they were to make drastic changes in pricing or security models. Even the thought of a major problem occurring could make a legislator run for the regulatory hammer. As it relates to current EU issues, it’s not yet the same problem because there is very little in the cloud (1), and what does exist “could be” put in any one of two dozen different clouds (2), thereby removing the risk I’ve associated with the duopoly.

      Thanks again for the great comment.

  4. @parkercloud

    May 14, 2013 at 6:45 pm

    A duopoly (the usual suspects Amazon and Google) might assume that most IT will be primarily Public Cloud PaaS and IaaS services, There is a possibility that in 5-10 years the majority of IT will be a federation of Public SaaS services (and a possible minimazation of Amazon and Google) and in the case of a duopoly that would require Amazon and Google to develop or acquire software companies. In 2020 SaaS might be the new kings

    • mthiele

      May 15, 2013 at 6:15 am

      Rick, I agree there will be significant private cloud, but the discussion only centered on “who” would be “owning” the public cloud space. As for SaaS I agree there will be a much larger portion of apps delivered that way. I think the interesting question is “will those SaaS apps be delivered by either Google or Amazon infrastructure.

  5. AWS is the McDonald’s of the cloud. Who’s the Burger King? — Tech News and Analysis

    May 17, 2013 at 8:15 am

    […] topic arose at the clouderati-filled Cloud 2020 meetup in Las Vegas last week (where someone suggested we’ll have a cloud duopoly of Amazon Web […]

  6. AWS is the McDonald’s of the cloud. Who’s the Burger King? | Earthgrid

    May 21, 2013 at 3:29 pm

    […] topic arose at the clouderati-filled Cloud 2020 meetup in Las Vegas last week (where someone suggested we’ll have a cloud duopoly of Amazon Web […]

  7. Future of I.T. Infrastructure Discussed at Cloud 2020 Summit

    May 23, 2013 at 8:06 am

    […] Visit http://cloud2020summit.com/ for more information. Mark Thiele’s blog on the summit can be viewed at http://switchlv.com/switchscribe//?p=262. […]

  8. AWS is the McDonald’s of the cloud. Who’s the Burger King? | Dan Gorman's Technology News

    August 15, 2013 at 8:12 am

    […] topic arose at the clouderati-filled Cloud 2020 meetup in Las Vegas last week (where someone suggested we’ll have a cloud duopoly of Amazon Web Services […]