The froth has begun to settle, and the view from the Square Mile is precarious. After a 24-month mania that has sent shares in artificial intelligence firms into the stratosphere, the first serious cracks are appearing. The City of London is bracing for a tech rout that analysts now describe as inevitable. The question is no longer whether the bubble will burst, but how deep the correction will be.
The alarm bells have been ringing for months, but they have now become a deafening klaxon. On Tuesday, a cascade of sell orders hit the London Stock Exchange, wiping £12bn from the FTSE 350 tech index in a single session. The trigger: a whispered report from a Shanghai research house suggesting that the latest GPUs from Nvidia are underperforming in real-world applications. The rumour was unconfirmed but the damage was done. In the age of algorithmic trading, panic travels at the speed of light.
To understand why this correction is so dangerous, one must examine the nature of the AI gold rush. For two years, investors have poured money into any company that could append “AI” to its name. We have seen a proliferation of startups with no business model, no revenue and no clear path to profitability. They have been propped up by a belief that artificial intelligence is a magic wand that will transform every industry overnight. This is the same fallacy that drove the dot-com bubble and the crypto craze. The hype cycle always repeats because human greed never learns.
The London market is particularly exposed. Unlike Wall Street, which hosts the big tech titans like Microsoft and Google, the City has become a haven for speculative AI minnows. These firms often have thin revenues and heavy burn rates. They trade on multiples that would make a tech analyst weep. One prominent example is CognitiveDynamics, a company that claims to have built a “general-purpose reasoning engine”. Its market cap has soared to £4bn despite producing a product that industry insiders say is barely more advanced than ChatGPT-3. The company has no patents and its chief technology officer is a 28-year-old philosophy graduate with no formal AI training. This is not innovation. This is alchemy.
The Bank of England has begun to take notice. Governor Andrew Bailey warned in a closed-door meeting last week that “unrealistic valuations pose a systemic risk to the financial system”. He is right to be worried. The trouble with bubbles is that they do not simply deflate. They pop, and the shrapnel cuts across the economy. If the AI rout deepens, pension funds and retail investors who piled in late will face substantial losses. The FTSE 100, which includes legacy banks and energy firms, could be swept up in the sell-off. Contagion is the shadow that follows every crash.
But let us pause and consider the deeper irony. The very technology that has sparked this mania may also be the instrument of its undoing. Market-making algorithms, trained on years of irrational exuberance, are now learning to detect patterns of fear. They execute sales faster than any human can blink. We built machines to maximise efficiency, and now they are turning on us. This is a classic Black Mirror scenario: our own creations have become the source of self-destruction.
Where do we go from here? The answer is not to abandon AI but to re-calibrate our expectations. We must separate the signal from the noise. There are genuine breakthroughs in quantum machine learning, ethical AI frameworks and personalised medicine. But these advances are incremental, not revolutionary. They will not deliver world-changing results in the next quarter. The City must learn to value patience over hype. Otherwise, we will repeat this cycle every decade until no one has any savings left.
For now, the advice to retail investors is stark: reduce exposure to AI equities, diversify into defensive sectors, and ignore the siren call of get-rich-quick schemes. The bubble may take months to fully pop, but when it does, the sound will reverberate through Threadneedle Street. We have been warned. The question is whether we have the wisdom to listen.
