Where are we in the AI journey?

You can't turn on the TV or look at news online without coming across stories about Artificial Intelligence (AI) and how it's going to change everything about our world.  I've been doing thinking along the lines of what impact it might have on our business at Roundstone Solutions, which is IT infrastructure.  Here are some of my thoughts:

To begin, AI is not new.  AI has been around since the 1950s (you can look it up).  For many years, AI has helped us in various parts of our lives, and for the most part it has been very beneficial.  But that's not the part of AI that we're talking about now.  What's new about AI is what science fiction novels and movies predicted might happen; when computers take over for humans.  This version of AI, called Generative AI, is where computers ostensibly "learn" from all of the material that exists on the Internet and elsewhere, and assimilate this information into machine actionable actions.  

The most visible of Generative AI (dubbed "GenAI") is the use of ChatGPT and similar large language models (LLM).  Many of us have used these LLMs to help with writing, solving simple planning, and similar tasks.  We're now moving into the next phase of GenAI, which is dubbed "Agentic AI".  Think of this as using computers to replace agents, like customer service agents.  When I personally experience it, I try to go from the AI agent to a real person, but maybe I'm in the minority.

I can think of many great applications of AI going forward, particularly in the medical field.  There will be many, many applications of AI will that will have huge impacts on our life.  To use an analogy, we're probably only in the bottom of the first inning of a 9 inning game.

My thinking of the impact of AI on Roundstone Solutions tends to be surrounding the IT infrastructure, as stated previously.  Let's discuss how that might play out.

The large hyperscaler companies (AWS, Azure, GCP, X, Meta, etc.) have been investing in billions and billions of dollars in new IT infrastructure to support the high compute demands that AI has.  Special chips, called Graphic Processing Units (GPUs) are apparently needed to power LLMs.  The hyperscalers are all out in the market buying up GPU chips from NVIDIA and AMD, buying up land for new datacenters, and purchasing power plants (including mothballed nuclear plants) in order to have the power to run all of the infrastructure.  But that's not our business...we don't work with hyperscalers.  We work with Enterprise/Commercial/SLED organizations with traditional IT infrastructures.

Some of our existing Clients are looking for ways that AI might help their business.  None of them have gone out and purchased IT infrastructure specific to AI yet, which I think is smart.  Let the others outspend each other as they look to gain an advantage.  After all, it's not as if this stuff isn't going to get less expensive going forward.

As we advise all of our Clients, you should only acquire IT infrastructure that has a clear business reason, where you can see the return from your investment (the value).  Unless you can see that, don't spend the money.  

I remember the "dot-com" bust very well.  At the time, people spoke about the "New Economy" as if traditional math and business leaders were dinosaurs and bound for extinction.  We know how that ended...the hype was not worth the result.  The companies that were next up benefited by the experiences they saw other companies had, as well as being able to purchase infrastructure much less expensively.

The net of what I'm trying to say is that for now, don't spend on buying AI infrastructure; things will get clearer going forward, and less expensive.  That's my two cents.