Telcos should learn the lessons from AWS in pursuing edge compute

Telcos planning on offering edge compute services should draw wider lessons from public cloud computing. For that, we need to go back to 2006 when Amazon launched S3 and EC2. At the time and for several years after, Amazon's cloud services were seen as more of a curiosity than a serious alternative for enterprises to entrust their mainstream compute workloads and data: it was seen as unreliable, insecure and more expensive than owned infrastructure (particularly once enterprises had extracted benefits of adopting virtualisation for their private clouds). And what's more, the services came from a online retailer: hardly a credible IT service provider.

The leading adopters of public cloud were start-ups and developers in enterprises who embraced the flexibility that public cloud offered them: with a credit card, they could be up and running for a fraction the cost and none of the hassle or delays associated with securing "physical" IT resources: making a business case, securing budget, satisfying compliance, selecting over-specified hardware and software from approved vendors, ordering, shipping, installing, on-boarding support, testing, maintenance. The unsanctioned use of public cloud for "grey" computing was tolerated by enterprises as a short-term expediency: if these temporary "skunkworks" proved useful, they could be moved back under the formal IT umbrella for production. But it didn't work out that way. Enterprises adopted SaaS and then embraced "hybrid" cloud. Multi-cloud is the new norm and public IaaS a $50bn market. Public cloud has become cheaper, although arguably it is still more expensive than owned-compute infrastructure "if" high levels of utilisation can be achieved for the life of the hardware and all costs (including opportunity costs) are not factored in. This is a key point that cannot be overemphasized for telcos planning their edge compute businesses: success depends on lowering adoption barriers.

I have been speaking to strategists and innovators at telecoms operators about their plans for edge compute (XaaS to enterprises and third-party application providers). Although most see potential opportunity for operators in offering edge compute services, none claim to have a precise idea where this opportunity lies: the enterprise edge? the mobile edge? the IoT edge? the city/sovereign cloud? A few are doubtful that there is any opportunity at all: they argue that since none of their customers have asked them for edge compute, there is no potential demand. Others point to operators' track record in public cloud and state that although there could be potential demand for edge compute, telcos are "genetically" incapable of capitalising on this.

In simple terms, operators looking to build edge compute business should:

• Make it easy, fast and simple for developers and application providers to start using their telco edge services.
• Assume that telco edge will be used in conjunction with on-premise edge and core (private and public) cloud. Plan and design for this from the outset.
• For operators pursuing a strategy focused on enterprise edge, develop "hybrid edge" propositions, including in partnership with on-premise edge compute vendors.
• For operators pursuing "city cloud" strategies, focus on identifying and developing approaches that minimise up-front investment from local authorities while allowing them to maintain control.
• For operators targeting IoT edge, acknowledge the importance of hyperscale players (Azure, AWS, Google) as a channel. Co-opetition applies here.

Telcos hoping to "bridge the gap" between on-premise/on-device edge and hyperscale need to focus on building bridges, not building walls around empty gardens.