I had the dubious pleasure of living through the dot.com bubble, from the nascent early web in 1995 through the crash in 2000.
It’s no wonder, therefore, that today’s generative AI (genAI) bubble is giving me a serious case of déjà vu.
Been there, done that, got the T-shirts to prove it.
Now I’m older and wiser. So listen up, young ‘uns, and let me pass along some hard-won wisdom from the last millennium.
Crash? What crash?
Just as with the dot.com era, there are plenty of prognosticators rambling on about how genAI is going to change everything, how it’s disruptive and revolutionary, blah blah blah. Just like the ‘New Economy’ the pundits were touting in 1998.
There are three obvious warning signs, however, that a crash is coming:
Diseconomies of scale
The big LLM vendors are all operating at a loss. The infrastructure necessary to get this technology to work is extraordinarily expensive – and as the LLMs get bigger and badder, the costs are only going to go higher.
Remember how Amazon.com lost money for years until it was finally big enough to turn a profit? Bezos and company took advantage of economies of scale.
GenAI is the opposite. The competition among the big genAI vendors is ratcheting up costs faster than profits will ever grow.
The gap between promise and profitability
Today, enterprises are scrambling for a genAI strategy. The AI hypesters have got them all convinced they absolutely need genAI.
No one seems to know, however, what genAI is for. Sure, there are plenty of use cases from copilot assistants to chatbots to document summarization.
What’s missing from this list? Anything that actually makes the enterprise any money.
GenAI is full of promise. Profitability? Not so much.
Unsustainable sustainability
It seems like only yesterday every enterprise and software vendor was bending over backward to implement sustainability initiatives. Even the cloud providers were loudly proclaiming that they would be carbon neutral in only (mumble) number of years!
Then along came genAI with its massive consumption of electricity and all thoughts of sustainable IT went out the double-paned window. Hell, genAI consumes even more electricity than Bitcoin – and that’s saying something.
How many more monster hurricanes do we need before we come back to our senses?
What the crash will be like
Some of the big LLM vendors will simply fall over. It doesn’t matter how many billions of dollars of crazy VC money you have to burn – if your business strategy includes burning it at an ever-increasing rate, eventually you’ll run out.
Even today, many genAI tools can connect to multiple LLMs, enabling enterprises to hedge their bets in case some of them should fail. Such hedging will mitigate some of the risk – but for many organizations, LLM failure will drive a nail into the coffin of their genAI plans.
The independent software vendor community will also have its crisis moment.
It seems that every vendor is falling over itself to build genAI into its offerings. If the dot.com bubble teaches us anything, it’s that a few of these vendors will survive the crash, but many won’t.
Will the one you’re counting on (or working for) be one of the survivors? Do you feel lucky?
Another important dot.com lesson: tomorrow’s survivors aren’t necessarily today’s leaders. Remember AltaVista and Lycos? They were leaders in the web search marketplace, until a latecomer named Google showed up and ate them for lunch.
If you look around at the leaders in various genAI categories (OpenAI? Nvidia? Microsoft?), be warned: just because they’re leading during the bubble doesn’t mean they’ll survive the crash.
Smart moves now to protect yourself
Follow the business value
GenAI has a serious case of the ‘shiny things’ syndrome: some technology comes along that is so shiny, people throw money at it first and then ask what it’s good for afterwards.
Don’t fall for this hype. Always – and I mean always – start with the business problem. What pain point does your organization struggle with that genAI is uniquely qualified to address?
Once you answer that question, look for where the business value lies. It may not be where you expect. For example, LLMs may not make nearly as much business sense as small language models (SLMs) or perhaps some other variant on the genAI theme.
Beware the money
VC money helps companies grow – but too much money can lead to unreasonable growth expectations that force them to take ill-advised risks. Such expectations can make companies brittle, increasing the risk of succumbing to competition or outright collapse.
Let the VCs make the big bets – but if you’re smart, you’ll place your bets elsewhere.
More baskets, more eggs
GenAI is not the only basket for you to put your hard-won eggs in.
It’s not even the only AI basket. There are plenty of other types of AI that provide a diversity of business value. There’s no reason to place all your bets on genAI.
Another reason to spread your AI investment around: if the genAI crash leads to the inevitable budget cuts, the more exposed you are, the deeper the cuts will be. Do you want the genAI crash to cut your IT budget by 10% or 50%?
Sometimes the best genAI strategy is no genAI strategy at all
If everybody else is zigging, sometimes your best move is to zag.
If most of your competition is slapping AI onto whatever they’re doing, then putting your time and money into something else – anything as long as it’s not genAI – will actually become an important differentiator for you.
Not only will the egg splatter the other guy’s face and not yours, but you’ll have put your effort into something more worthwhile – and profitable – in the long run.
The Intellyx Take
It might seem like I’m a genAI skeptic through and through, but actually I’m not. I bring a healthy dose of skepticism to the hype around the technology, more so than the technology itself.
In other words, I question the sizzle not the steak.
The dot.com bubble may have been a time of crazy excesses, but look at the web now. It’s fine. It’s bigger than ever and fully integrated into our way of life.
For every pets.com or kozmo.com (remember kozmo.com? No? Precisely), there is an Amazon.com or Google – dot.coms that made good.
The same will happen to genAI to be sure. Once the bubble pops and all the bubble dust settles, the true, long-term value of genAI will become clear – and the companies that make the right moves today will be the success stories of the future.
Copyright © Intellyx BV. Intellyx is an industry analysis and advisory firm focused on enterprise digital transformation. Covering every angle of enterprise IT from mainframes to artificial intelligence, our broad focus across technologies allows business executives and IT professionals to connect the dots among disruptive trends. As of the time of writing, Microsoft is a former Intellyx customer. None of the other organizations mentioned in this article is an Intellyx customer. No AI was used to write this article. Image credit: Johnny Ryan, Craiyon, and Jason Bloomberg.