Let’s have a frank conversation. Every time I open my feed, it’s a relentless barrage of AI evangelism. We’re told generative AI will cure disease, solve climate change, and probably even do my taxes without complaining. And with every grand proclamation comes an even grander valuation. We’re deep in the territory of what I call “aspirational accounting,” where a company’s worth is measured not in profits, but in the sheer audacity of its vision. It all feels a bit familiar, doesn’t it? A bit too much like that heady, champagne-soaked party in 1999, right before the music stopped. The question isn’t if AI is transformative—it is. The question is whether the current gold rush has decoupled from reality, creating an AI market bubble that’s about to pop.
So, What Is a Market Bubble, Anyway?
Before we dive into the AI frenzy, let’s get our terms straight. An economic bubble isn’t just a period of high prices. It’s a specific kind of madness. It happens when the price of an asset—be it Dutch tulips in the 1600s, dotcom stocks in the late 90s, or American houses in 2007—soars far beyond its fundamental value. This surge is fuelled by what former Fed Chair Alan Greenspan famously called “irrational exuberance.”
Investors pile in, not because they’ve meticulously analysed a company’s balance sheet, but because they see prices going up and are terrified of missing out. It becomes a self-fulfilling prophecy, for a while. Everyone feels like a genius. The problem with bubbles, however, is that they are inherently unstable. They are built on belief, and when that belief shatters, the fall is swift, brutal, and unforgiving. The hangover from the dotcom crisis wiped out an estimated $5 trillion in market value, and we all know how the 2008 housing bubble ended. These are not just lines on a graph; they represent real financial pain.
Today’s AI Gold Rush: Bigger, Faster, More Unicorns
And so we turn to today’s main character: Artificial Intelligence. The numbers are, frankly, staggering. A recent CNBC report highlights that global spending on AI is projected to hit $375 billion in 2024 and rocket past $500 billion by 2026. The private market is churning out “unicorns”—private companies valued at over $1 billion—at a dizzying rate, with nearly 500 of them now roaming the AI landscape. It’s a boom of epic proportions.
The poster child for this boom is, of course, OpenAI. The company is reportedly in talks for deals that could push its valuation towards an eye-watering $1 trillion, all while its own revenue projections sit at a comparatively modest $13 billion. Think about that for a second. That’s a valuation-to-sales ratio that would make even the most optimistic dotcom CEO blush. This isn’t just optimism; it’s a belief system. And it’s a belief system that has sent Nvidia’s stock into the stratosphere, anointing it as the indispensable arms dealer in this AI war. But when valuations are based more on narrative than on numbers, you have to start asking some uncomfortable questions.
The Glaring Investment Risks Lurking Beneath the Surface
There is a fundamental tension at the heart of the AI market. On one hand, you have genuinely world-changing technology. On the other, you have financial behaviour that looks suspiciously like a classic speculative bubble. The sky-high tech valuations are the most obvious warning sign, but the dangers run deeper.
Are We Just Trading Money in a Circle?
One of the most concerning patterns, as pointed out by seasoned investors like GMO’s Ben Inker, is the rise of “circular revenue deals.” Inker told CNBC he’s “certainly seeing lots of evidence of bubble-like behavior,” and this is a prime example. So what does this mean?
Imagine this: a tech giant like Microsoft invests a huge sum into an AI startup like OpenAI. The startup, in turn, needs massive computing power to train its models. Where does it get it? From Microsoft’s Azure cloud platform, of course. So, a significant chunk of that initial investment flows right back to Microsoft as revenue. On paper, both companies look great. The startup has cash and a “strategic partner,” and the tech giant can report booming cloud sales driven by the AI revolution.
But is new economic value actually being created, or is it just clever accounting? It’s like a café owner giving a baker £100 for a ‘strategic partnership’, and the baker immediately spending that £100 on coffee from the café. The money just went in a circle, but both businesses can claim £100 in new revenue. This isn’t a sustainable model for an entire industry. It’s a financial carousel, and at some point, someone is going to want to get off. These are the kinds of financial patterns that should set alarm bells ringing for anyone looking at investment risks in this sector.
Echoes of Bubbles Past
If this all feels like a movie you’ve seen before, that’s because it is. The parallels to the dotcom bubble are impossible to ignore. Back then, the mantra was “get big fast.” Companies like Pets.com spent fortunes on Super Bowl ads while haemorrhaging cash, convinced that capturing “eyeballs” would eventually lead to profits. It didn’t. Today’s mantra seems to be, “build the most powerful model, and the business case will follow.” The focus is on technical capability, not commercial viability.
Jared Bernstein, Chair of the White House Council of Economic Advisers, has pointed out that the share of the economy being ploughed into AI investment is already nearly a third greater than the peak of the dotcom boom. When the government’s top economists are drawing these comparisons, you should probably pay attention. The core lesson from 1999-2000 is that a revolutionary technology and a rational market are two very different things. A sector correction wasn’t just possible then; it was inevitable. The same logic applies today. We should also not forget the 2008 financial crisis, which serves as a stark reminder of how complex financial instruments and a herd mentality can bring the global economy to its knees.
The Hyperscaler Hedge: Are the Tech Giants Playing a Different Game?
Now, here’s where the narrative gets more complicated. The people pouring the most money into this frenzy aren’t naïve day traders. They are the biggest, most powerful companies on the planet: Microsoft, Amazon, Meta, and Google’s parent, Alphabet. They are investing tens of billions in data centres, custom chips, and AI startups. Are they all just caught up in the hype?
Not exactly. From their perspective, this isn’t just a speculative bet; it’s a strategic necessity. They are in a brutal war for technological supremacy, and AI is the key battleground. For them, investing in AI isn’t just about the potential for a short-term return. It’s about ensuring that the next generation of computing—the entire AI ecosystem—is built on their platforms. They are selling the picks and shovels in a digital gold rush they helped create.
As Anneka Treon of Van Lanschot Kempen put it, “We continue to see opportunity because ultimately, AI bubble or not, it boils down to real dollars being spent on real capex.” This capital expenditure on servers and infrastructure is very real. It’s the foundational layer of the AI economy, and the hyperscalers own it. This creates a powerful moat. Even if many of the AI application startups fail, the giants who provided the underlying cloud services will have already been paid. It’s a brilliant, if somewhat cynical, strategy.
The Problem of Long-Term Bets and Short-Term Hype
This brings us to a critical disconnect. Building genuinely transformative AI is a long-term, incredibly expensive endeavour. It requires patient capital. BlackRock CEO Larry Fink argues these massive investments are vital for America’s economic leadership. He’s not wrong. But the public markets and venture capitalists often operate on much shorter timelines. They are chasing quick wins and dramatic growth, fuelling the speculative frenzy.
What happens when the short-term hype collides with the long-term reality of R&D costs? What happens when investors realise that profitability for many of these AI companies is years, if not a decade, away? That’s usually when the panic starts. Howard Marks, a man who has successfully navigated more market cycles than most, offers a more measured take. He sees the current tech valuations as “high but not crazy,” suggesting we haven’t yet reached full-blown bubble psychology. Perhaps. But we are certainly walking along the edge of the cliff.
Navigating the AI Haze: A Pop, a Fizzle, or a Soft Landing?
So, are we in an AI market bubble? The honest answer is that it’s a tale of two markets. The technology itself is profoundly real and will reshape our world. But the financial market built on top of it is showing all the classic signs of speculative mania. We have sky-high valuations disconnected from revenue, circular investment patterns, and a frantic fear of missing out.
A significant sector correction feels not just possible, but probable. It may not be the cataclysmic pop of the dotcom bust, but a painful deflation of the most overhyped names seems inevitable. The companies with real business models, real customers, and real profits will survive and thrive. Many of the 498 “unicorns” whose only asset is a good story and a powerful model might not be so lucky.
The key for any investor, big or small, is to separate the technology from the hype. Don’t invest in a narrative; invest in a business. Look for companies solving real problems for real customers. The infrastructure players—the Nvidias, Microsofts, and AMDs—are likely a safer bet, as they are the ones selling the essential tools for the whole industry. But even their valuations warrant a healthy dose of scepticism.
The AI revolution is happening, but revolutions are messy, chaotic, and often brutal for those who bet on the wrong horse. The landscape is littered with enormous potential and equally enormous investment risks.
So, what do you think? Is this 1998 all over again, with the real crash still to come? Or is this different, a necessary and sustainable investment in the future? The next 18 months will be telling.
