Industry Voices — Chua: CSPs, hyperscale clouds and edge - beware the camel's nose

Once hyperscalers have inserted themselves into telco networks, it might be hard to get them out. (Ashish Sangai on Unsplash)

A few short years ago, we were gearing up for another spectacular showdown — the battle for the edge. Both from a mobile network operator (MNO) or wireline network operator standpoint, there was a sense that we in telecom had ceded the spoils of 3G, 4G and wireline broadband networks to the cloud providers and the OTTs. In the telco consulting and analyst space, you'd be hard-pressed not to find PowerPoint decks replete with slides showing the ascent of cloud providers (and associated valuations and profit margins) compared with the generally tepid growth of key business indicators at communications service providers (CSPs) worldwide. The telecoms industry had lost the battle for the cloud, but with 5G rollouts in process, it was certainly not going to lose the battle for the edge.

The edge looked like a rich oasis telcos could go after. It was projected to be a significant portion of cloud computing in five years. Gartner says that by 2025, 75% of enterprise-generated data will be processed at the edge. Whether you believe that number is closer to 30%, 50% or Gartner's 75%, it's still a tremendous opportunity.

Telecom operators who had previously failed in building successful public cloud offerings vowed not to mess up this time. They planned to build MEC (multi-access edge computing) platforms that would be the next generation of AWS, GCP, Azure. But now, fast forward a couple of years, instead of building out sizable MEC platforms at cell-towers, or at mobile switching centers or at regional edge data centers, most carriers are partnering with the hyperscale cloud providers.

Sponsored by Anritsu Company

Free eBook: Efficient Testing Will Contribute to Long-Term Success in the 5G Marketplace

As companies worldwide race to launch 5G successfully and avoid post-deployment setbacks, we see a massive rise in the demand for 5G test and measurement equipment. We will discuss how efficient testing can contribute to long term success in 5G marketplace.

Cloud partnerships galore

AT&T and Microsoft Azure have partnered on Azure Carrier Edge Zones. Right behind AT&T, Azure has Rogers, Telefonica, Vodafone Business, SK Telecom, Telstra, Etisalat, and NTT Communications lined up on similar partnerships. And AT&T, back in July 2019, had announced that they would use Microsoft's cloud service for non-network workloads.

AT&T has also partnered with AWS on its Greengrass IoT platform in the past, and with Google on Google's Mobile Edge Cloud (GMEC) efforts, powered by Google Anthos, a next-gen orchestration system.

Meanwhile, Verizon has partnered with AWS on their AWS Wavelength edge cloud platform, opening up their 5G infrastructure to hosting AWS Wavelength at crucial locations on their network. Of course, Verizon has its own "edge" service in the form of its CDN for content hosting but soon to be upgraded to host applications.

Keeping your enemies close

There's plenty of movement across other CSPs worldwide in a similar vein. In AvidThink's conversations with numerous CSPs, the strategy appears to be about keeping your "frenemy" close. In addition, every telco aims to take a poly-cloud approach, hoping that by befriending multiple cloud providers, they can play one against another, negating the negotiating power of a single hyperscale cloud provider.

Furthermore, their thinking goes a little like this:

  • Building and running profitable clouds are hard. Hyperscale cloud providers are good at it, and it's their core competence. CSPs are good at building and running networks. So, cloud providers should run clouds, and CSPs should run networks.
  • Monetizable edge use cases are few and far between. CSPs are still engaged in ongoing discovery and experimentation. Therefore, CSPs need to enable developers to develop and test new candidate edge applications quickly. However, CSPs don't have access to the developer ecosystem, but the hyperscalers do. And so, CSPs should work closely with hyperscalers to leverage their relationship and access to application developers.
  • Some CSPs believe that if they work closely with the hyperscalers, they may learn more about hyperscalers' edge cloud stacks and the secrets of operating those stacks efficiently. Likewise, CSPs hope to get an early view into new edge applications and use that information to build monetizable use cases.

Indeed, the above all sound like valid reasons for CSPs to justify partnerships with all of the top cloud providers.

Beware the camel's nose

It certainly looks rosy and promising right now, with the chants of kumbaya filling the air. Hyperscalers are courting carriers for partnerships and permission to access carriers' networks without definitive business models. However, we at AvidThink suggest CSPs be cautious because:

  • Edge monetization and business models are not clear right now, and when splitting the spoils gets contentious, partnerships break up quickly. If the monetization is happening on the application developer side of the value chain, CSPs will lose out.
  • That real estate at the mobile switching centers, aggregation centers, central offices, and eventually street-side cabinets or cell tower bases, is highly valuable. That's one of the unique assets that CSPs have that hyperscalers want access to. A single rack-unit (1RU) in those locations is much more expensive than 1RU in a centralized cloud data center. CSPs need to fully understand the opportunity cost of that 1RU before letting their cloud partner place their server racks in prime real estate that they own or lease. And hopefully, they understand the benefits that they are getting in return.
  • CSPs still need to be cautious even if hyperscalers are only looking for a fiber cross-connect from their local public edge clouds into carrier networks. Network access to a carrier's aggregation center, mobile switching centers and central offices needs to be carefully gated. That low-latency access is another unique asset and giving it away without understanding its full value may be a hard-to-reverse decision. Certainly, low-latency connections to hyperscaler clouds benefit a CSP's mobile and wired subscribers and can be a form of competitive advantage. Nevertheless, CSPs need to consider how much value is being created on both sides and whether eventual value distribution is fair.
  • Despite their heft, hyperscalers are much more agile than most CSPs. If there's any learning as a result of early proof of concepts, our bet is on the hyperscalers being able to react and capture value before CSPs even realize what's going on.
  • Hyperscalers are already acquiring assets and talent better to understand the deployment and operation of mobile technologies. Consider Microsoft's recent acquisition of Affirmed Networks and MetaSwitch, all while reassuring their network equipment partners and carriers that they come in peace and that it's synergistic to carrier relationship.

Once CSPs let hyperscalers into their networks, the undo function will be quite expensive and perhaps impossible. It's worth considering second-tier cloud providers, or interexchange carriers and colocation providers as safer options for partnerships. Regardless, if there's value in the edge, CSPs need to quickly figure out where it is and secure it before they accidentally give it away by letting that pesky camel's nose into the tent.

Roy Chua is founder and principal at AvidThink, an independent research and advisory service formed in 2018 out of SDxCentral's research group. Prior to co-founding SDxCentral and running its research and product teams, Chua was a management consultant working with both Fortune 500 and startup technology companies on go-to-market and product consulting. As an early proponent of the software-defined infrastructure movement, Chua is a frequent speaker at technology events in the telco and cloud space and a regular contributor to major leading online publications. A graduate of UC Berkeley's electrical engineering and computer science program and MIT's Sloan School of Business, Chua has 20+ years of experience in telco and enterprise cloud computing, networking and security, including founding several Silicon Valley startups. He can be reached at [email protected]; follow him at @avidthink and @wireroy

Industry Voices are opinion columns written by outside contributors—often industry experts or analysts—who are invited to the conversation by FierceTelecom staff. They do not represent the opinions of FierceTelecom.

Suggested Articles

Now that Nvidia has pivoted into the networking and mobile chip market with its acquisitions, large networking incumbents should be very nervous.

Windstream Holdings announced on Monday that it's now free from the shackles of Chapter 11 bankruptcy.

VMware's VeloCloud SD-WAN has found a home in Equinix's data centers, which means VMware can now offer it as a virtual network function (VNF.)