My blog on telco challenges, reasons, and possible solutions included a comment on the advantages of having a relatively low internal rate of return. This, I contend, lets telcos invest in projects that would be financially unattractive to OTTs, including cloud providers. It generated a dozen CFO comments, both from telcos and enterprises, and so I’m following up with a more detailed look at IRR impacts, based on these comments.
Let’s look first at the IRR advantage telcos have. Social media companies and cloud providers have a return on invested capital (the company-wide average of IRRs) in the 30’s percent range. Telcos have ROICs in the high single digits, so we could say they’re a third of the cloud and AI providers. A project with an ROI of 20% would fall far short of the target for the former group, and would be a significant improvement for a latter. Let’s test that through the comments.
CFOs tell me that you need to look at tech project spending through the lens of corporation budgeting and public company financial reporting. First and foremost, they say, IRR is regularly compared with what some call the “hurdle rate”, which is the rate of return that’s the de facto minimum as seen by boards of directors, senior management, and Wall Street analysts. A many of the CFOs tell me that the hurdle rate is perhaps a better name for the benchmark against which projects are assessed, but others say that the CFO would normally set an ROI target somewhat higher than the hurdle rate, since the latter represents the point where you transition from “questionable value” to “negative value”. Hurdle rates, being totally financial in makeup, also don’t consider any risk premium, which typically rises as the payback period for a project or the useful life of the assets increases. So far, the “theory” of the IRR advantage holds, but with identified complications.
CFOs in verticals where the economic output is goods for sale also note that the rate of inventory turnover compared to the length of the supply chain is also a factor that impacts project decisions. If you can sell something before you need to pay for having acquired it, your cash-flow benefits are considerable and you actually don’t have much capital tied up. Projects that can accelerate this sell-before-you-pay will reap a financial benefit beyond the normal ROI calculations.
It is true, CFOs tell me, that companies with a low IRR/hurdle rate can justify projects whose ROI is lower, but whether that’s a leverageable benefit isn’t a given. For example, the finances of two companies in the same vertical but with widely different hurdle rates would normally be assessed by investors and creditors in a way that preferences the one with the higher hurdle rate, because it shows that company is using capital better. Where things are more complicated is where we’re looking at companies in different verticals, like telcos and OTTs of various types, and where the risk premium for a potential project is high.
Let’s look at the central issue in opportunity creation in real-time applications, which is where the applications are hosted relative to the location of the processes involved. Today, almost all real-time applications are self-hosted by enterprises at points close to the processes. This makes the process/application control loop “short” and reduces the risk that a connection problem will stall the process being controlled. Could you achieve better economy of scale with a reliable edge computing service and goods edge-connection QoS? Likely so, but whether this would be a good project for the company with the application depends on the relationship between the savings in infrastructure economy of scale compared to the cost of the service and connection.
From the perspective of the provider of edge hosting services, or of connections to support these services, the risk premium is high for two reasons, which bears on the question of whether telcos could actually exploit their ROI advantage. The obvious reason is that there’s no current convincing proof that these services can make a business case for buyers, and if they can it could take some time. The less-obvious reason is that the public cloud providers are already preparing for an edge opportunity to emerge, and if it does they could well be entrenched competitors with application skills, in a space new to telcos, who also lack the essential application skills likely to be needed.
Every major cloud provider offers a premises-hosted middleware toolkit whose goal is to commit an enterprise to an application structure that favors their cloud services. In many cases, this toolkit is used to link on-site edge computing in process control, with the front-end piece of cloud applications or front-ends. It’s reasonable to say that these create a bridge between today’s self-hosted model of edge computing and any future model, likely more metro-ized, since process control integration is most logical when facilities are close enough to engage in shared activity.
Low-latency service needs are not likely to be enough to save a telco role, or even be valid. Integration of two or more adjacent-but-not-connected facilities would mean goods/parts/material would have to be transported among the facilities, and this is a macro-time process. You don’t need millisecond latency to control something that would likely take minutes or even hours to complete when undertaken. Where physical stuff has to be moved, integrated process control value is limited. Thus, only mobile applications would likely benefit from low-latency services, and the number of these is limited.
It’s possible that, given time, mobile real-time needs would grow. I’ve blogged in the past about the opportunity link between augmented reality glasses and real-time services, and I believe that link exists. The questions are how long it would take for it to mature, and whether telcos could supply the skills needed to play in it, or would just be trapped in a pure connectivity mission that would, like consumer broadband, inevitably commoditize.
AI is a special problem right now, according to both telco CFO-staff and enterprises. There are three risk factors to consider there. First, can the technology actually make a business case, given that to do so it likely has to be substantially self-hosted in order to resolve governance concerns? Second, can it be trusted not to create a problem that even human oversight can’t properly deal with? Finally, is it moving so fast that any investment in AI will be rendered obsolete far faster than an investment in traditional hosting? Over 80% of enterprises say that right now AI is assigned a higher risk premium. Telco comments on that are too sparse to be useful.
This all bears on the question of telcos’ IRR advantage. Can telcos win at hosting? Is there any real value to edge-as-a-service without it? The answer is that IRR advantage alone isn’t going to be compelling. There’s no reason to exploit a financial advantage to exploit a non-opportunity. This means that telcos need to develop application-level relationships, which would demand they frame a strong edge computing model with effective APIs, and do it right now. Otherwise, the cloud providers will own the application model and make it very difficult for telcos to gain any traction with real-time applications.
