AT&T has cut a collaboration deal with Amazon. Is it a step forward for both companies, or is AT&T inviting the wolf into the fold? We’ll try to outline the case for both, then pick the one we think is on the mark.

It’s pretty clear what Amazon gets out of this. They get a Tier One telco deal for hybrid hosting of what I hear will be components of OSS/BSS. They get AWS experts familiar with that aspect of telco-land, which is the place where it’s easiest for cloud services to penetrate. They get what I hear is a heck of a deal on their Leo satellite services to augment AT&T mobile, and they get fiber interconnection of AWS data centers at a bargain rate.

AT&T gets the other side of these deals, meaning then get a head start in OSS/BSS modernization, they get to shed some capex for hosting all of that, and they get revenue from the DCI services (though not retail-level) and a good deal on satellite voice, text, and broadband. They may also learn useful things about extending IoT real-time applications, digital twins, and AI.

On the surface, this looks like a pretty balanced deal, in short. Let’s dig deeper to see if it is.

For Amazon, the basic revenue gains from the deal are a plus, and the data center interconnect is a greater one. The question will be whether they can exploit the on-ramp to the telco world. Telecom is locked in a deadly cut-costs-till-you-bleed-out saga, which means that every step they take will be squeezing every nickle out of spending, all the more as low apples continue to be picked and the benefits thus become harder to get without maximizing ROI. On one hand, they avoid trying to do better in the enterprise side of things, where every other cloud provider has a better shot. On the other, the focus on telecom would mean backing into the enterprise space through the intermediary of telecom services. Anyone who wants to get an entree into anything via telecom is hitching their wagon to a glacier, not a star. By the time telcos figure out the game, it may be won by others who took a more direct route.

For AT&T, things are a bit more complicated. They gain by having AWS guide them through an OSS/BSS modernization that, if done by telcos alone, would likely modernize into a slightly changed version of what’s already in use. This all happens with minimal outlay and effort, other than on AT&T’s part. They also get some DCI revenue and Leo revenue, but in a sense they may be disintermediating themselves. The more stuff that lives in the cloud, and the more of enterprise networking shifts from a VPN service to a part of cloud services. With Leo, they may be establishing a pattern that cedes service differentiation to a third party; other than cost and coverage, how do you differentiate connectivity?

The thing about the cloud’s front-end mission is that it means that more and more business traffic starts off on the Internet, then lives within a cloud provider network. You can easily see that if an enterprise were to use cloud front-end elements for worker/branch sites, you’d shift away from MPLS VPNs to a single cloud on-ramp to business data centers. This is obviously a loss of revenue to telcos, and that means that they end up losing revenue even if they don’t make deals with cloud providers that could accelerate cloud adoption. And if real-time services are a big future opportunity for telcos to engage in differentiated service sales, might they be giving more of that away? But if those future opportunities don’t develop, they’re giving it all away, and cloud providers may be a key to getting real-time missions going.

The thing to watch for here is an indication that private 5G will play in the deal, which has not so far been mentioned. As a direct opportunity, Amazon already knows that private 5G is a sideshow. However, private 5G could well be a pathway toward evolving current premises-bound IoT applications to larger facilities, and eventually into the WAN. You can’t get to a massive real-time future in one jump, and private 5G is plausibly on the glide path to an evolution in the right direction.

Which of the two is best-equipped to address that? The problem is not so much who’s best, but who’s capable. Staying with “best” for a minute, Amazon has the edge here. They have private 5G and they have application and hosting experience. They already sell cloud services to enterprises, and Outpost, an element of their OSS/BSS hybrid with AT&T, is a candidate tool for broadening real-time application reach beyond the premises. AT&T would have to wait until the opportunity evolves to the point where some sort of WAN service could figure in.

Another issue AT&T would have to face is Leo. It is not out of the question that you could expand real-time applications out of a building via low-earth satellite. Generally, there’s “real-time” and “really-real-time”, meaning process applications that are incredibly latency sensitive. Most of them are, today, confined to a single facility. Go further out, and the movement of stuff across the larger distance takes more time by far than the connection latency. But could AT&T or Amazon push that approach, and be successful? They could try.

While they’re doing that, Microsoft and Google might be working as hard, and from a better position. Google has the most AI credibility of any cloud provider, and the fact that AI is still a name to conjure with is demonstrated by the fact that the release I cited above mentions it.

Both Amazon and AT&T could benefit mightily from an expansion of real-time applications. It would be, by a massive margin, the best possible symbiosis that could emerge from this deal. It might be the only one that really justifies it, in fact. Absent such a future benefit, I think the balance of risks and benefits is at best a narrow win, and if you factor in the risk that the deal will kick of competitive reactions by players who are more likely to be aggressive, anything less than this is a loss all around.