Is the AI Bubble About to Burst?

Kieren Sharma
Oct 20
5 min read

Updated: Oct 24

The technology world is currently dominated by one word: AI. From revolutionary large language models (LLMs) to the slightly less necessary AI-powered toothbrush, everyone is talking about AI-powered everything. But beneath the dazzling veneer of innovation lies a deep economic anxiety, symbolised by valuations that defy gravity. Currently, just one company, Nvidia, sits comfortably over a $4 trillion market capitalisation, dwarfing the entire London Stock Exchange, which is valued at about $3.4 trillion. This striking disparity raises the core question: Is AI being massively overvalued, and are we witnessing an economic bubble about to burst?

Listen Now

Defining the Bubble: When Hype Outpaces Value

To determine if we are in an AI bubble, we must first understand what a financial bubble is! An economic bubble is defined as a period when asset prices surge far beyond their intrinsic value—that is, how much a company should actually be worth based on its sales and physical assets.

One key metric used to gauge this overvaluation is the Price-to-Earnings (P/E) ratio, which compares a company's current share price to its earnings per share. A high P/E ratio suggests investors are willing to pay significantly more now, betting on enormous future revenue. For instance, Nvidia’s current P/E ratio is between 28 and 29 times its annual revenue.

The Engines of Exuberance

Bubbles are largely driven by exuberant investment behaviour, often sparked by a compelling real-world innovation or breakthrough. This behaviour is then fuelled by FOMO (fear of missing out) where investors worry about missing the train while everyone else profits.

The danger arrives when speculation overtakes rational valuation, detaching the boom from actual economic value and leading to an overinflated, and inherently fragile, state.

History teaches us that bubbles always pop sooner or later because the values are not grounded in anything real. When they burst, they trigger a sharp market crash, potentially causing businesses to fail, wiping out net worth, and in the worst cases, leading to a recession. Notable examples include the 1920s stock bubble leading to the Great Depression, and the 1990s dot-com bubble.

Echoes of the Dot-Com Crash

Today’s AI boom closely resembles the dot-com bubble of the 1990s. That era saw wild, rapid investment in internet startups, driving tech stocks to record highs before the bubble burst in 2000, wiping out tens of thousands of jobs and causing the NASDAQ to plunge nearly 80% from its peak.

The similarities are striking:

Revolutionary Technology, Irrational Hype: Just as the internet was game-changing, AI is a revolutionary technology promising AGI (Artificial General Intelligence) and massive productivity benefits. Yet, the hype has pushed current valuations to irrational heights.
Startup Madness: Tiny AI startups, sometimes consisting of just three people and some AI powered idea, are securing millions in investment—a situation Open AI’s CEO Sam Altman called “insane" and “not rational behaviour".
Extreme Concentration: Stock market gains are heavily concentrated. Seven companies (Alphabet, Amazon, Apple, Meta, Microsoft, Nvidia, and Tesla—the “Magnificent Seven") now account for 36% of the S&P 500’s total value, a 50-year record high concentration. Furthermore, AI-related stocks have accounted for 75% of S&P 500 returns since late 2022.
Revenue Complications: The bubble is further fuelled by Nvidia's monopoly over optimised GPU chips and roundtripping like behaviour among major players (like Oracle buying graphics cards from Nvidia, and Nvidia investing in OpenAI, who partners with Oracle). This complex exchange of billions makes true revenue incredibly hard to track.

The Bear Case: Arguments for the Burst

A compelling case exists suggesting the AI bubble is on shaky ground:

Missing ROI: There is currently little financial evidence to justify the spending. A recent MIT report found that 95% of businesses that had integrated generative AI into their operations have not yet seen any return on investment.
The $2 Trillion Problem: US tech firms must generate at least $2 trillion of additional annual AI revenue by 2030 to justify current capital spending—a massive amount that feels farfetched.
Resource Bottlenecks: Scaling AI infrastructure requires immense resources. Electricity supplies are becoming a significant bottleneck, with utility companies squashing plans for new data centres. Already, 5-gigawatt data centres are being built, demanding energy equivalent to the power consumption of London.
GPU Deterioration: These massive investments in data centres require constant upkeep. GPUs must be replaced every two to five years, adding continuous costs to infrastructure spending.
Expert Alarm Bells: Warnings are coming from the highest financial authorities. The Bank of England warned this month that the risk of a sharp market correction driven by an AI stock slump could pose a “material risk" to the financial system. Jamie Dimon, CEO of JPMorgan, stated he is “far more worried than others," suggesting the burst could be six months or two years away. Even Sam Altman has warned that people will “over-invest and lose money" in this current phase.

The Bull Case: Is AI Different This Time?

Despite the dire warnings, some experts argue that the AI boom is just the beginning of a “super-cycle"—a multi-decade period of massive growth—not a typical bubble.

Stronger Fundamentals: Unlike many dot-com startups, many leading AI companies (Nvidia, Microsoft, Google) are making substantial profits outside of AI, which they can reinvest to fund their AI development. This foundation means AI is grounded in stronger fundamentals than past bubbles.
Exponentially Evolving Technology: The underlying technology of AI is “simultaneously increasing" in capability. While the infrastructure of the internet stayed largely the same, generative AI models are rapidly getting smarter, potentially allowing companies to “ride the wave" of fundamental technological improvement.
Untapped Potential: Lisa Su, CEO of AMD, argues that skeptics are “thinking too small" and that AI's potential will spark a decade-long super-cycle transforming industries globally. We argued that society has not fully tapped into the potential of having a “mini Einstein in our pocket," partly due to the current lack of AI literacy.
Humanoid Robots on the Horizon: A massive source of untapped return on investment could be humanoid robots. Companies like Tesla and Figure are developing advanced robots that will apply the current AI “brains" to physical labor, potentially entering the workplace within the next five years and delivering huge physical labor efficiency.

Conclusion: The Long-Term View

The jury remains out on whether the impending crash will be a swift, devastating bubble burst or merely a necessary correction in a longer-term super-cycle. If the bubble does burst, history suggests bad things will happen. A major crash could trigger financial instability, cause retirement funds (like those invested in the S&P 500) to lose value, and potentially increase inequality as the wealthy buy up devalued assets.

However, the consensus is that AI technology itself will ultimately pay off and be a net positive for society in the long run, just as the internet has been. The question is whether the current hype and investment levels are sustainable until that long-term potential is realised. Perhaps a slowdown in development and spending would not be entirely negative. It would give society time to address major unsustainable issues arising from the rapid scaling of AI, including:

The immense climate impact and reliance on fossil fuels for energy.
Unresolved issues around data privacy and copyright.
The fundamental lack of understanding regarding how these large models work and ensuring they are aligned with human values.

While the immediate future holds volatility, the underlying technology promises transformation—it's just a matter of which companies, and how many investors, will survive the inevitable shakeout.

If you enjoyed reading, don’t forget to subscribe to our newsletter for more, share it with a friend or family member, and let us know your thoughts—whether it’s feedback, future topics, or guest ideas, we’d love to hear from you!