Enterprises have been telling me for years that IT is changing because more and more of their IT budgets are focused simply on orderly modernization of existing infrastructure, justified by existing business cases. They point out that any major new investment in any area of IT, including networking, would have to be justified by new benefits. Logical, right? But where might those benefits come from? I ran back through three years of comments on this, and found 121 thoughtful remarks that were on point, and here’s how I analyze the results.
The key point is that just short of 70% of the potential targets for new business cases would relate to worker empowerment, which accounts for over 80% of today’s business cases. The majority of the remainder focus on what’s often described as “buying”, meaning the presentation of information to prospects and customers to support the sales process, to support products already sold, etc. The “buying” category has seen the most growth in the last decade. This category is currently seeing a boost because of the high level of economic stress, which has driven businesses to try to improve their prospect-customer evolution. It’s why cloud computing has picked up, for example.
Worker productivity is largely (65%) targeted at indoor workers who regularly use the computer, terminal, or other IT device, including teleworkers. This category of worker can be subdivided into “analytic” and “administrative” worker groups, which make up 28% and 72% of workers, respectively. The former group use company data extensively, where the latter tend to be involved mostly in support processes that deal more with the routine tasks of doing business than with making business decisions. Enterprises say that the analytic group are almost always able to use AI effectively, even if governance rules prevent them from feeding their analytic data to a cloud provider, because they have a unit value of labor two or more times that of the administrative group. A little help here pays off, and this is the worker target category that makes most of the new AI business cases today.
Relaxation of governance issues, opening up new business cases that exploit corporate databases, would help the current cloud-AI model a lot. Not only would it remove a data security barrier, it would help project progress by allowing for expensed services. Enterprises say that the tenuous nature of almost any “new” business case means capital projects based on it are harder to push forward, but those that can be expensed are much easier. Look for action by the AI players here, but the stories about how cloud AI models can hack sites isn’t helping enterprises gain confidence.
Administrative office workers very significantly in whether they can actually justify paying for AI tools, which means they vary significantly in whether any IT project can make a business case by improving their productivity. Some have tasks that readily adapt to IT enhancement, while others may present few or even no such opportunity. This, combined with the lower unit value of labor, means that new IT projects that focus on the majority of the class of workers currently empowered with IT will be difficult to justify.
Where enterprises say that worker empowerment is currently low are with the rough half of our “administrative” category who are indoors but not working at a desk. This would include most retail employees, for example. Some enterprises in the retail sector have told me they’re doing more with IT, particularly focusing on things like phones or handheld devices as a means of coupling these workers to applications. This group seems to be the category that contributes most to the current new-project-new-benefits inventory.
Roughly 40% of workers perform jobs that are largely physical, meaning that they require interaction with things or people, indoors or in part or totally outdoors. This includes jobs like vehicle operators, facility repair (electrical, water, sewage, telecom), and many in healthcare. This group of workers is generally untapped according to enterprises, but is also considered difficult to address because of the real-time nature of their work and the challenges of coupling them with useful information via some sort of device. Enterprises think that the benefit pool this group represents is as large as the one that launched today’s IT, but they admit that it is very difficult to frame a project to reap these benefits. Broad internal skills are a problem most (78%) cite, because vendors tend to push very narrow project goals to generate a quick revenue hit and to reflect their own specialization. The big consulting firms are rated low (38%) in this area too, so it’s up to enterprises to try to build the skill set needed to integrate project elements across multiple tech/product disciplines.
What does all this net out to in terms of project opportunities? Enterprises are, today, mostly interested in new projects that directly interact with non-workers, meaning prospects/customers. This sort of project is seen as both an opportunity to increase sales and to decrease the size of the workforce, but it also addresses the growing shift of consumer focus to online retail. The next area getting attention is the indoor-not-at-desk administrative class; mass-market retail organizations are increasing their focus on this group.
Telework is in a sense related to this online retail area, because it’s easiest to provide using the same sort of cloud tool used for online shopping. Teleworkers as a group tend to be high-unit-value-of-labor types, more likely to have advanced degrees, and so are good targets for empowerment. This is true even though the COVID boom in telework interest has passed and many enterprises say they have a return-to-the-workplace mandate.
Among the part-to-all-outdoors classification, the big focus is on dispatch/delivery workers, in large part because the online-sales boom has created a home/business delivery boom. This is only a small number of the outdoor-real-time group, but it is more readily addressed because the activity is less information-complex and because the fact that a vehicle is involved means that some information coupling can be facilitated through it. However, the real interest in the longer term is autonomous vehicles, replacing workers completely. That, in fact, is something enterprises see as a goal for many of the job classes not yet enhanced via IT.
The challenge for all this is the problem I’ve called “ecosystemic” in the past. Most of these new IT missions, unlike most prior IT missions, are more than simply a matter of collating, analyzing, and delivering information. They introduce what used to be called “methods analysis”, the task of analyzing how a task can be done with minimum effort/time investment based on physical constraints and not just information needs. The problem is the spread of technology elements driven by a spread of requirements. One single player may simply not get a big enough piece of the benefit pie to justify their making a commitment, and the absence of any player could derail the entire project.
If you look at one financial site’s view of Nvidia, you can see a bit of this. The article I cited notes that Nvidia is looking to build chips beyond AI chips to complete a “product”, so that it can drive a combined feature set more likely to fulfill business needs without a partner who at best will want a share of profits and at worst may drag their feet and threaten progress overall. But if Nvidia wants to be, as the article suggests, the “architecture of AI”, that may not take things far enough. Nvidia clearly knows that, and has a number of projects aimed at real-time and industrial automation missions. For now, they seem to be “sandboxes” rather than actual attempts at creating an ecosystem, but that’s likely to change over time, particularly the focus on cloud-hosted AI rather than self-hosted or even local AI.
The final point is that it has to change. It’s not surprising that the tech world has focused on the low apples of benefit-generating projects, but if that were a durable strategy we’d all be using terminals on mainframes today. Sometimes you need to leave your comfort zone, and usually that happens when what used to be comfortable isn’t any longer. Does that sound like tech today to you? It sure does to me.
