It seems you can’t open a browser these days without being hit by a tsunami of AI hype. Nvidia’s market cap has rocketed past that of entire nations, and every company from your local bakery to global banking giants is scrambling to tell you about their “AI strategy”. The narrative is simple: artificial intelligence is changing everything, and you’re a fool if you’re not all in. But what if the loudest voices are missing the most important part of the story?
When a respected economist like Jason Furman, who advised President Obama and now teaches at Harvard, says he’s worried, it pays to listen. In a recent Bloomberg interview, he dropped a truth bomb that cuts right through the noise. “I’m more worried about the financial valuation bubble than I am a technological bubble,” he stated. This isn’t just academic nit-picking. It’s the single most important distinction in a market awash with capital and hype. Are we building the future, or are we just building a fantastically expensive mirage?
The Two Bubbles: Tech vs. Treasure
Let’s be clear about what Furman is saying. He’s not arguing that AI is vapourware. The technology, from OpenAI’s models to a dozen others, is genuinely impressive. The risk isn’t that AI itself is a complete dud. The real danger lies in the staggering valuation concerns and the sheer amount of money being thrown at it, with little regard for actual, measurable returns.
Think of it this way. The dot-com boom of the late 1990s wasn’t wrong about the internet. The internet did, in fact, change the world. The bubble was in the financial valuations of companies like Pets.com, which raised millions on the promise of a digital future but had no sustainable business model. The investors who piled in got burned, even though the underlying technology was revolutionary.
This is the core of the potential AI investment bubble. We are seeing what Furman describes as “hundreds of billions of dollars a year being spent on data centres, energy and the like.” This is a colossal bet on infrastructure. But are we placing this bet on the right horses, or are we just funding a very, very expensive hardware party?
An Economy Firing on Just One Cylinder
Furman’s analysis gets particularly sharp when he looks at the broader economy. “We do not have a US economy that is firing on all cylinders,” he warns. “We have a US economy that is firing on one cylinder right now.” That one cylinder, of course, is AI-related demand. Companies are spending a fortune on chips and cloud services, which boosts GDP and makes tech giants look invincible.
But this isn’t organic growth. It’s a closed loop of market speculation. Tech companies, flush with cash and high stock prices, are spending billions on AI infrastructure from other tech companies. It creates the illusion of a booming economy, but scratch the surface, and you find that the productivity benefits aren’t trickling down into other sectors like manufacturing, healthcare, or logistics.
The money is generating a lot of heat and light within the tech ecosystem, but it’s not yet powering the rest of the economic engine. This creates a significant economic risk. If the promised productivity gains don’t materialise soon, those sky-high valuations will start to look less like genius foresight and more like collective delusion.
The Slow, Complicated Reality of Adoption
So, why isn’t AI boosting productivity across the board? Anyone who has actually tried to implement a new enterprise system in a large organisation knows the answer. As Furman puts it, “People in the wild are just slow and kind of complicated.”
Real-world adoption is messy. A business can’t just plug in GPT-5 and expect its efficiency to double overnight. It requires changing workflows, retraining staff, integrating with legacy systems, and navigating a minefield of data privacy and security issues. This is a multi-year slog, not an overnight revolution. “They figure out one use case this year and the next use case the next year,” Furman notes, highlighting the gradual, piecemeal nature of true technological integration.
This slow pace of adoption is a direct threat to the current tech economics. The market is pricing in a revolutionary impact today, while the reality on the ground is evolutionary at best. When Wall Street’s timeline collides with Main Street’s reality, things tend to get ugly.
Does History Offer a Guide?
The dot-com parallel is instructive. A huge amount of capital was spent laying fibre optic cables across the globe, leading to a glut that bankrupted companies like Global Crossing and WorldCom. Yet, ten years later, that cheap, abundant bandwidth became the foundation for YouTube, Netflix, and the entire streaming economy. The vision was right, but the timing—and the initial investments—were disastrously wrong.
We could be seeing the same pattern with AI. The current frenzy is driving a massive build-out of data centres and GPU capacity. Many of the companies pouring money into this may not see a return on their investment. But the infrastructure they leave behind could pave the way for the real AI winners—companies that might not even exist yet but will leverage this cheap computing power to build something truly transformative.
The question for any investor today is: are you betting on the company laying the cables, or can you spot the future Netflix hiding in the wings? The former is a game of high-stakes musical chairs, while the latter requires patience and a genuine understanding of where real value will be created.
As a journalist covering this space for years, the narrative Furman is pushing, as highlighted in Gizmodo’s analysis, feels like a necessary dose of cold water. The hype cycle is intoxicating, but the fundamentals of business and economics don’t just disappear because a new technology arrives. Profitability, productivity, and real-world utility still matter.
The current AI investment bubble is less about the promise of the technology and more about the patience of capital. The market is betting on a sprint, but true economic transformation is a marathon. What happens when the investors realise they’ve trained for the wrong race?
What do you think? Is the current AI spending a necessary investment in our future infrastructure, or is it a classic case of market speculation leading us towards a painful correction?
