Anthony Lobretto, Senior Vice President at 11:11 Systems, Network as a Service.
getty
For two decades, enterprise IT had a clean division between the two industries, living in harmony. Telecom carriers were the highway that moved your traffic to your cloud. Cloud providers run your workloads at the other end. The two industries operated in parallel lanes with different products, service guarantees and operational cultures.
AI is collapsing that boundary. Hyperscalers need to get closer to the customer to serve latency-sensitive AI workloads, and telcos need to move up the stack to stay relevant as the network becomes software-defined. It's now getting harder to tell who the carrier is and who the cloud is, and most enterprise buyers haven't adjusted their procurement, architecture or operational playbooks.
The clearest example comes from AWS, which is one of 11:11 Systems' partners, and its SiteLink feature that allows enterprises to route traffic between their own data centers across the AWS global backbone, bypassing AWS's compute regions entirely. A cloud provider is now selling long-haul connectivity between customer sites, directly competing with the traditional carriers. AWS Cloud WAN extends the same idea to inter-region and branch connectivity. A cloud provider is monetizing its global fiber as an alternative to traditional carriers.
The concept flows in the other direction, too. Megaport, a software-defined carrier and another partner of ours, acquired compute platform Latitude.sh for up to $300 million, pairing its network with on-demand GPU infrastructure. A network company bought a compute company to chase AI workloads. Carriers are expanding into adjacent markets as quickly as cloud providers are building into theirs.
Sitting in the middle is a category of managed service providers—such as ours—built around the complexity, integrating carriers, cloud workloads and hyperscaler backbones, streamlining the chaos into unified contracts and end-to-end SLAs.
Underneath it all, hyperscalers are pouring capital into subsea cables and terrestrial fiber on a scale historically reserved for telcos. That fiber will be monetized, and the most logical buyer is the enterprise that previously procured the same capacity from a carrier.
For technology leaders, the practical implications matter, and most procurement and operations playbooks haven't caught up.
Carrier circuits are typically priced at a fixed rate based on committed bandwidth. Cloud-delivered connectivity is consumption-based, billed by usage. Neither is universally cheaper.
I recently worked with a large automotive brand expanding its autonomous mapping program into emerging markets. On day one in a new country, the workload is modest, and the fleet is small. Two years in, it may need to burst to 100 Gbps to backhaul data during peak hours.
Committing to 100 Gbps redundant, fixed-priced circuits on day one is an enormous upfront cost when utilization is low. Starting with consumption-based connectivity lowers the barrier to entry and turns fixed-price circuits into a high-class problem as throughput scales.
The mistake I see most often is procurement teams comparing a fixed monthly carrier price to a consumption price at peak utilization, which makes the cloud option look more expensive than it is.
A Tier-1 carrier network service includes end-to-end SLA commitments for latency, jitter and availability, backed by financial credits. Most cloud-provider network services guarantee availability but don't commit to the granular performance that carriers do. For workloads where performance affects user experience or revenue, such as real-time inference, financial transactions and voice, this is a board-level risk rather than a technical detail.
What I consistently see is customers who want both worlds: the dynamic scalability and multi-path redundancy of a hyperscaler network combined with the 100% SLA and fixed pricing of a traditional carrier. That conundrum doesn't resolve cleanly today, and bridging it is where MSPs come in.
Traditional carrier provisioning is ticket-driven and multi-week. Cloud-delivered connectivity is portal- and API-driven. Turning deployment times from weeks to minutes changes what's possible for the business, but your network team needs different skills, with software automation alongside traditional networking.
I've seen network teams resist this shift, only to suddenly need connectivity when a business unit outpaces IT. Budget for the talent shift, or budget for a partner who has already made it.
Cloud-delivered networking deepens hyperscaler dependency at exactly the moment many enterprises are trying to reduce it. A hybrid approach using carriers for some legs and a cloud network for others costs slightly more but can preserve pricing leverage and prevent single-vendor outages from cascading.
The value of a partner that can integrate across carriers, cloud networks and hyperscalers while translating among their contract models, service guarantees and operational cultures is rising sharply.
The question is no longer "telco or cloud." It's which combination of fixed and consumption, manual and automated, end-to-end and best effort fits the workload in front of you. The organizations that adapt best over the next five years will be the ones designing around workload requirements instead of legacy definitions of cloud versus carrier.
Forbes Technology Council is an invitation-only community for world-class CIOs, CTOs and technology executives. Do I qualify?
此内容由惯性聚合(RSS阅读器)自动聚合整理，仅供阅读参考。 原文来自 — 版权归原作者所有。