AI Capex in 2026 looks like a glorified pet rock.

When you work in the technology business for as long as I have, you notice a few trends. For instance, when I worked at KeHE Foods, my vice president received an iPad for Christmas. By March, he was asking why our user interfaces could not look like Apple products. He also demanded an overhaul of all websites so the company could look good on seven-inch and ten-inch iPads. This bit of career nostalgia looks like the Artificial Intelligence integrations we see today. Today, I want to discuss the ways that agile professionals can deliver value when executives are pushing change on an organization.

Like many web developers during the rise of smartphones and tablet computing. I understood that I needed to do something to accommodate this new technology. The trouble was that web development tools were not that accommodating. The mobile computing revolution was taking over, and web developers like me were desperate to keep up. Often, I would find Cascading Style Sheets and stack them on top of each other, hoping that, based on the screen size, they would render correctly. It was a hack that worked until someone found an embarrassing edge case. Usually, it was the Vice President of Application Development who found it on his shiny new iPad.

Fortunately, open-source projects like Bootstrap emerged, and developers found online tools to help legacy systems adapt to mobile computing. When I attempted my SaaS startup, I promoted this type of functionality. It was a hassle to build multiple platforms and web browsers, but the result was web pages that worked if there was an internet connection.

Soon, access to App Stores became a matter of life or death. Now you needed an app. Sadly, it had to be written one way for Apple Products and another for Android. It also required tools outside of Microsoft. Thus, one of the ways Microsoft fell behind during the mobile computing revolution was its lack of tools to help developers chase the gold rush. By the time they purchased Xamarin and got into mobile development, it was already too late.

I look back on this time almost fifteen years ago with an unusual mixture of cynicism and admiration. The market pressure from executives with shiny new toys demanded new features and directions, and smart, overburdened people in the trenches attempted to cobble together solutions to make it happen. For those outside the technology world, the ingenuity was impressive. That's why we can do anything we can do on a PC on our mobile phone.

Today we are in a similar situation. Instead of mobile devices, the hot new trend is Artificial Intelligence. Microsoft was an early investor in OpenAI and, over time, has provided more than $100 billion in cash and Azure infrastructure to help make the company a success. As someone involved in the early days of the Artificial Intelligence boom, I became aware that AI would require significant time, expertise, and energy. Companies are spending billions to build data centers and models that will help them develop the next killer app the business world cannot live without. It is also a convenient excuse to lay off employees, as they can argue that, to fund capital expenses, they must fire people.

It is a high-stakes bet, and if they want to reap the rewards of eight billion people using their products, they must scale to meet the demand. Unfortunately, this massive build-out is causing problems. The complexity of these systems makes them difficult to manage and scale. Next, many organizations have data and structural problems in their business that make it nigh impossible to implement Artificial Intelligence systems. It does not help that most people have deeply conflicted notions about Artificial Intelligence. They like it as a helpful tool but fear being replaced by automated systems and agents. Finally, investors and CFOs are not seeing a return on investment.

A typical subscription to Microsoft Office is about $282 a seat per year. If you have 5,000 employees, your company needs roughly $1.4 million to keep your office running. It does not count renting server space from cloud services like Azure or Amazon Web Services. It also does not cover electricity or the actual computer equipment in the office. Now, Microsoft is offering a $18 add-on for Copilot and AI services a month. The almost 50% price increase has serious implications for businesses and businesspeople, and they are beginning to push back.

So far, that return has not happened. According to the International Monetary Fund, Artificial Intelligence is reaching about 20% to 30% in almost all organizations. It is a great percentage for hitting in baseball but lousy in software utilization. The challenges of data and organizational structure keep cropping up, and many decisions require human judgment and cannot be automated. Thus, Artificial intelligence is expensive and does not meet business expectations.

This is where we find ourselves in the summer of 2026. We possess technology potentially as transformative as electrification, yet it remains prohibitively expensive and notoriously difficult to implement. The rush to get data centers online is sparking local mutinies over energy grid strain in communities worldwide. Ultimately, the staggering upfront cost of the tech still vastly outpaces the actual productivity gains.

As agile professionals or project managers, we need to be aware of this environment. Executives are afraid of missing out, but decisions like that tend to get people fired.  We must insist that executives audit their current licenses, halt blanket rollouts, and demand baseline metrics; otherwise, we are cutting checks for glorified pet rocks.  We must be the voice of reason in the room who counsels caution. Agile people need to use radical candor to explain that some teams might benefit from Artificial Intelligence tools, but agents across the enterprise might be more expensive than human employees. Finally, we must insist that people make decisions and be accountable, rather than that large language models do.

Only then can we have sensible conversations about Artificial Intelligence and business. Once again, it is we poor folks in the technology trenches who must do this work. It inspires both cynicism and admiration.

Until next time.