Vendor View

Why Telcos Won't Rule the Cloud

There is a common belief that the future of IaaS clouds will be dominated by telcos. That belief is wrong for many reasons.

• By Boris Renski
• 12/01/2011

There is a common belief that the future of IaaS clouds will be dominated by telcos. To justify this idea, people point to a recent spree of acquisitions by telcos: Verizon acquiring Terramark, CenturyLink acquiring Savvis, etc.

Until I gave this subject a bit more thought, I went with the crowd in believing that telcos would dominate the infrastructure cloud market. After all, there were many logical arguments supporting this. Telcos own the network which powers all infrastructure clouds today. Telcos have been an asset-intensive business, supporting enormous infrastructure investments that amortize over many years. Telcos have brand recognition and the trust of enterprise customers. And, yes, most telcos already offer a variety of value-added services around Internet infrastructure, ranging from MPLS to data center hosting.

While all of these arguments may be convincing on the surface, closer examination points in a different direction. The physical ownership of network connectivity is no more relevant to operating an infrastructure cloud than ownership of a freeway to the manufacturing of cars. Pacific Gas and Electric is just as well suited as telcos to invest in the cloud. PG&E owns the lines that power the servers and the telcos offer connectivity to the data center.

Do the future market powerhouses in cloud infrastructure have to be giant companies that already have the ability to make large infrastructure investments? Not necessarily. RackSpace, now second in IaaS services, started its cloud storage service with two racks of servers that were gradually scaled out with market uptake.

The cloud is an industry shift associated with new ways of delivering IT resources. This shift resulted in massive value creation, and this value now stands to be divided by all kinds of companies (and several industries). Cloud dominance refers to being able to capture the biggest chunk of this value and consequently, have the influence across new markets that emerge as a part of the cloud shift.

We can draw parallels between this evolution with other significant industry shifts, like the emergence of computers in general. What are the companies we think of when we think about computers? Apple, Microsoft, Intel, Oracle, etc.

In my view, asking if telcos will dominate the cloud is equivalent to asking if we'll immediately think of Verizon and AT&T when we think of the cloud. That isn't happening.

Today, when we think of the cloud (and I am focusing exclusively on infrastructure ), we immediately think of Amazon Web Services (AWS). People frequently put Amazon, VMWare, OpenStack and telcos all in the same bucket. The logic is as follows: Amazon is THE cloud today and it offers IT infrastructure on demand. Who else is well positioned to offer IT infrastructure? Telcos are, because they are a) big and powerful; b) can offer on-net data center capacity; c) have a channel into the enterprise. So all telcos have to do is add the "on-demand" part and everyone else in the cloud market is toast, right? Wrong.

Making predictions about telcos in the cloud by drawing parallels with Amazon is altogether incorrect. Amazon's cloud dominance is temporary. Its offering is a vertical integration between IT resource provisioning and data center management. This was perfect to initially disrupt the market, but it is not where the majority of cloud value will be captured in the future. As the cloud matures, people will stop comparing telcos to Amazon and VMWare and three distinct industries will compete to capture the value created by the cloud shift: cloud software, IT infrastructure hotels (what AWS is today), and managed cloud data center services. And, how and where do telcos fit?

IaaS clouds are ultimately a product of innovation in middleware and software industries and it is these industries that stand to benefit from the associated value creation the most. The bits and bytes written by the developers at Amazon that made it possible to provision virtual machines and storage-on-demand is what ultimately set the wheels of the cloud revolution in motion. From then on, hundreds of companies from application infrastructure giants like VMWare to hosting service providers like RackSpace with its OpenStack initiative, aligned themselves to capture a chunk of the cloud software pie.

Being successful in IaaS today is about helping developers and applications interact with hardware in a new way.

Will telcos compete in that space? I can bet that they won't. Some argue that the telcos' established enterprise sales channel will supersede their lack of capabilities in software. If this were true, then why did telcos not emerge as key players in the content delivery network market when it was up for grabs 10 years ago? There were opportunities for telcos to enter software-intensive markets, but there is no history of innovation by telcos in this space. The enterprise sales channel is ultimately about relationships with customers and all relationships carry a particular context. Just because I have a relationship with my gardener doesn't mean I would trust him to do electrical work in my house. Similarly, just because enterprises buy bandwidth and data center capacity from a telco, doesn't mean that enterprise would prefer to run off a cloud stack built by a telco over VMWare, Microsoft, or OpenStack.

Telcos are about asset management and financial engineering; their competence is in making long-term investments in commodity assets (such 3G antennas or fiber optics) and ongoing maintenance of these assets. At best, they have some competence in physical maintenance of hardware in the data center, but zero competence in the space where value creation (cloud) is taking place today.

How about data center hotel space, i.e. vertically integrated offerings like Amazon? In my view, as the cloud market matures, the value behind such vertical integration will gradually erode. IaaS offerings in their present state will cease to exist and the market will be split between companies monetizing the software by offering various flavors of the cloud orchestration stack and companies monetizing the services by offering managed hosting of these respective cloud stacks. The natural aversion to vendor lock-in will be a big driver in this. We see this happening already with efforts like OpenStack and companies like RackSpace, Internap, and Korea Telecom opting to embrace open standards for cloud orchestration vs. Amazon's proprietary model.

And what about the managed cloud hosting service? I think this is where telcos really have a shot at claiming part of the cloud business. Moreover, it is exactly where telcos are aiming with their recent acquisition of large data center operators like Terremark, Savvis, etc.

Whenever some innovation disrupts a market, an upward shift in the entire value chain occurs. At the top of the chain is where most of that value is captured. At the same time, the bottom of the chain becomes yet even more commoditized. Infrastructure clouds became possible because of cloud orchestration software. That's where there was innovation and it is where the value will be captured. This, in turn, commoditized data center management services to a greater extent. In the early days, managed hosting was about continuously and manually rewiring the network, reinstalling OSes on physical nodes and reconfiguring access rights. Cloud-like infrastructure automated most of these tasks. Hence, the business of data center management has gotten quite a bit closer to maintaining the communications network and fixing/installing wireless cells--a commodity service that telcos are well positioned to offer.

Does this mean that there is a new market for telco services? Yes. Is it fair to say that telcos will dominate the cloud? Not at all. Telcos will dominate the scraps of the cloud; the commodity bottom end of it with razor thin margins. And will we ever think of AT&T, when we think of cloud? Not likely.