What People Mean When They Say ‘AI Bubble’
My financial advisor brought it up during my annual review in January. I didn’t. He did. We were halfway through discussing my portfolio when he paused and asked, “So, what do you think about this AI bubble?”
That’s when I knew I needed to write about the AI Bubble so clients will understand.
I see articles about it almost every day now. When people find out what I do, the question comes up within five minutes. Sometimes it’s curious, sometimes worried, but it’s always the same question: “What about the bubble?”
Here’s what I tell them: the bubble is about the vast amounts of money being thrown into AI development and infrastructure. At some point, the people investing want to get a return on that money. That creates pressure. That pressure creates distortion. And that distortion is what people mean when they say “bubble.”
But that’s the short version. The real answer takes longer, and it matters if you’re trying to figure out whether to invest in AI for your business. So I’m writing this series to help you make smart decisions regardless of what the markets do.
Why This Feels Familiar
If you were in business in the late 1990s, you remember what happened. Every company suddenly needed “.com” in its name. Venture capital flowed like water. Pets.com spent millions on a Super Bowl ad. Valuations stopped making sense.
Then it all crashed.
The NASDAQ Composite, heavily weighted towards tech stocks, collapsed by more than 75% between March 2000 and October 2002, wiping out over $5 trillion in market value. Companies like Pets.com, 360networks, and eToys.com became cautionary tales of businesses with extreme valuations but weak fundamentals.
But here’s what people forget: the internet didn’t stop working. Email kept running. Amazon survived and became one of the largest companies in the world. So did eBay and Cisco. The technology was real. The valuations were not.
We’ve seen this pattern before. The 2017-2018 crypto bubble followed a similar path. Bitcoin surged from roughly $1,000 in early 2017 to nearly $20,000 by December, then crashed hard in 2018. A wave of low-quality coins launched through ICOs, driven by FOMO and speculation. Some crashed to zero. Others survived and the underlying blockchain technology kept developing.
The pattern repeats: massive investment, unrealistic expectations, correction, then the actual technology settles into doing useful work at more reasonable prices.
The Simple Version (What I Tell People)
Global data center investments jumped 51% year-over-year in 2024 to about $455 billion, with the top 10 hyperscalers responsible for more than half of that spending. That’s largely due to AI infrastructure build-out. Revenue for GPU-equipped servers nearly tripled in Q4 2024, with Nvidia controlling over 90% of AI GPU server shipments.
Morgan Stanley estimates around $3 trillion of cumulative data center spending from 2025 through 2029. McKinsey’s central scenario puts it at about $5.2 trillion through 2030.
That’s real money. The people who invested those billions expect returns. Some will get them. Many won’t. The math has to work eventually, and when it doesn’t, corrections happen.
That gap between investment timeline and reality timeline is what creates the bubble pressure. It doesn’t tell you whether AI works. It tells you the money is getting impatient.
The Three Different Conversations Happening
When people say “AI bubble,” they’re usually having one of three conversations. Understanding which one matters for your situation.
What Investors Mean
They’re looking at valuations that assume perfect execution and massive scale. They see billions going into infrastructure before revenue catches up. They’re calculating whether the returns will justify the investment.
Think about Pets.com again. It had a $300 million valuation at its peak. Today, Chewy exists and actually works because it solved real logistics problems and built sustainable unit economics. The idea wasn’t wrong. The valuation and execution were wrong.
You’re not in the investor’s position. You’re not betting on whether an AI company will return 10x to its venture backers. You’re deciding whether a tool solves a problem you have.
What Vendors Mean (When They Worry)
AI vendors are feeling pressure to show ROI quickly. Competition is driving prices down. There will be consolidation, and not every vendor will survive it.
This matters if you’re picking vendors. You want to know your vendor will be around in three years. You want to avoid building your operations around a tool that disappears after the company runs out of runway.
But this is a vendor selection problem, not an AI problem. You solve it the same way you’d solve it in any market: look for companies with real revenue, real customers, and real staying power.
What Business Owners Mean (When They Ask You)
When a business owner asks me about the bubble, they’re usually asking one of three things:
“Should I wait?” “Will I look stupid if I invest now?” “Is this going to be around in three years?”
The real question underneath all of these is simpler: “How do I not waste money?”
That’s a good question. It’s just not answered by predicting whether there’s a bubble.
What Happened After Previous Tech Bubbles
The dot-com crash was terrible for investors. Lots of people lost lots of money. But email didn’t stop working. E-commerce didn’t disappear. Online banking kept processing transactions.
The companies that survived had real revenue and solved real problems. Amazon was losing money in 2000, but it was also filling real orders and building real distribution capabilities. When the bubble popped, those capabilities didn’t vanish. They became more valuable because the competition collapsed.
Infrastructure got cheaper after the crash. All those fiber optic cables that telecom companies laid during the boom didn’t disappear. They got bought for pennies on the dollar and made bandwidth cheaper for everyone. That helped the next generation of companies build better products faster.
Bad actors and hype merchants got flushed out. That was painful but ultimately healthy. It became easier to tell who was solving real problems and who was just riding the wave.
The actual technology became more accessible and practical. The survivors weren’t trying to prove the internet worked anymore. They were just using it to run better businesses.
Bubbles punish speculation. They don’t kill utility.
What the Bubble Talk Gets Right
The concerns are legitimate. Investment levels are genuinely unprecedented. When GPU server revenue surpasses traditional CPU servers and makes up nearly two-thirds of the server market in a single quarter, something unusual is happening.
Some promises are outpacing reality. Not every AI marketing claim holds up under scrutiny. There’s a lot of AI-washing happening, similar to the “air coins” that flooded the crypto market during the ICO boom.
A correction of some kind is probable. When this much money moves this fast, corrections usually follow. Lots of current vendors won’t survive. If you’re evaluating vendors, that’s something to factor in.
What the Bubble Talk Gets Wrong
Your problems are real regardless of market sentiment. If you’re spending too much time on data entry, that’s a problem today. If your customer service team is overwhelmed, that’s a problem today. Market corrections don’t change that.
Technology capability and financial valuation are separate things. AI can process natural language and recognize patterns and automate repetitive tasks. Those capabilities exist whether AI company stocks go up or down. You’re buying access to capabilities, not shares in a company.
Good solutions survive corrections. They often get better and cheaper after bubbles pop because competition decreases and infrastructure costs drop. The businesses that waited for the “all clear” signal after the dot-com crash missed years of competitive advantage.
You’re buying tools, not making venture bets. When you implement AI to solve a specific problem, you’re making an operational decision with a payback period. That’s different from speculating on whether the AI sector will grow by 40% annually.
The Question That Actually Matters
Not “Is there a bubble?” but “Does this solve a real problem for me?”
Your timeline is operational, not speculative. You need to know: Will this pay for itself in 12 months? Will it make my team more productive? Will it help me serve customers better?
If it makes sense today based on those questions, market conditions don’t change that. If it doesn’t make sense today, the fact that AI is hot doesn’t fix it.
What’s Coming in This Series
This is blog post one of six. Here’s where we’re going:
Post 2: Why the Bubble Talk Might Be Right (And Why That’s Okay) We’ll look at why the concerns might be valid and why that doesn’t have to derail your plans.
Post 3: The Stuff That Survives Bubbles What endures when hype deflates, and how to focus on solutions that will still be useful in three years.
Post 4: How to Invest in AI Like a Bubble Is Coming Tomorrow A practical framework for making smart investments that work regardless of market conditions.
Post 5: The Midwest Advantage in an AI Correction Why regional businesses are actually well-positioned if there’s a correction.
Post 6: Your AI Strategy Checklist: Bubble-Proof Questions A practical tool for evaluating any AI investment decision.
The Bottom Line
I’ll keep getting asked about the bubble. My answer will stay the same: focus on value, not on predicting markets.
The businesses that thrived after the dot-com crash weren’t the ones who predicted the timing of the crash. They were the ones who knew what problems they were solving and kept solving them through the noise.
That’s what we’re going to help you do.
Want to cut through the noise? We offer free consultations to help evaluate AI investments based on your actual business needs, not market hype. Let’s talk about what makes sense for your company. Schedule a consultation
