Is this third wave of AI going to turn into a rerun of the Dutch Tulip Mania of 1634-37 and the Dot.com Dot Bubble of 1997-2001?

We're all too young to remember tulip mania, which saw the price of some single tulip bulbs reach what would be about $1 million today. But industry veterans will never forget the end of the latter catastrophe, which tore $7 trillion in valuation out of the Nasdaq and plunged Silicon Valley into darkness for many years.

We have to ask because at the moment, AI-driven optimism is framing the entire world if IT and data center development. A story link this week refers to McKinsey calling for a fourfold increase in worldwide data center consumption by 2030, for which 180GW of new (presumably) sustainable power will be needed. This represents 6% of the world's entire current electricity grid and an investment of $2.16 trillion in just the data centers themselves and another $720 billion in the new power plants.

Can such an ambitious construction project be accomplished physically, even if the investments are there? Federal, regional, and local regulatory hurdles, local NIMBY movements, and general political tomfoolery and incompetence can delay single small projects for years and years; can we expect the world as a whole to become compliant and efficient enough to accomplish such a task?

Large organizations – smart ones, at least – are now looking at use cases for AI. ChatGPT, which launched this new era in 2023, is being considered for tasks that require quick research and glib writing. Its large-language model (LLM) is being examined for other general uses, and sibling small-language models (SLMs) are being pursued for specific uses in honing in on particular problems, strengthening data security, and bringing AI's consumption burdens and costs down.

The abilities of AIs to, say, examine thousands of high-res medical images and correlate findings with doctors' diagnoses quickly, or detect very long-term failure patterns in jet engines, have been in evidence for several years. Any organization with very large amounts of incoming data – that is to say, any organization – can now think of ways to pull new insights out of seeming chaos.

The idea of expert systems and knowledge systems, which were the heart and soul of the second AI wave in the 1990s, is gaining renewed life. The original AI conceptions of the 1950s to mimic and recreate the human brain, seem to be hanging on in the work of robotics and concerns about the “singularity” in which an AI, allegedly “smarter” than us, decides we are not worth preserving.

So it seems that much of, perhaps all of, the global IT industry has pushed all of its chips to the center of the table to bet on AI. Sure, cloud computing, IoT sensor development, continuous improvements in long-distance networking, and thoughtful development with blockchain as part of Web3 thinking continue to receive funding and secure room on prestigious conferences and meetings.

But AI is the word today. Does it represent a bubble, or is it truly an enormous general opportunity that is beneficently permeating our thoughts for the good of the industry and society?