Will the AI bubble burst in 2026? We ask the experts
The question of whether we are in an AI bubble and, if so, when exactly it might burst has dominated conversation in the City in recent months. Who better to turn to for answers than some of the UK’s finest stockpickers and strategists?
The year 2025 was not a quiet one for markets. It started with Donald Trump’s historic return to the White House when he was still facing charges for attempting to overturn the previous election result. Less than four months later, he upended the global trading system with his Liberation Day tariff blitz, the legality of which is still being determined by the Supreme Court.
Last year’s rally in gold – the safe haven asset once written off by some consensus commentators as anachronistic and redundant – continued apace in 2025, rising 65 per cent year to date at the time of writing. And all the while, the world’s largest central banks appear – for the most part – to be successfully reining in the inflation that has dogged developed economies in recent history, all while achieving that fabled soft landing.
But without doubt, economic historians will look back on 2025 and judge it to have been defined by a different financial megatrend entirely: the untrammelled rise of artificial intelligence and the veritable frenzy of investor speculation that it has set off.
As the year has progressed, valuations of AI companies have become uncomfortably stretched, sucking investor capital into fewer and fewer stocks. The combined market capitalisation of the 10 largest companies in the S&P 500 now makes up more than 40 per cent of the entire index.
Forever blowing AI bubbles?
And that’s not to mention the emergence of an even frothier breed of AI upstarts, from a new-look Oracle that has doubled down on data centre expansion, to Coreweave, to the Alex Karp-run Palantir where speculation has been as hysterical as it has been gravity-defying. At one stage Palantir’s stock was changing hands at a valuation of 600x that of its forward earnings. By contrast, the average FTSE 250 company trudges along at 13.3x.
For much of 2025, the funding for firms’ wide-eyed attempts to remain at the revolution’s avant-garde largely came from equity markets or firms’ often-gargantuan cash flows. That is to say, tech firms – listed or private – for the most part turned to shareholders or their own profit to fund vast data centre construction projects or enormous purchases of GPUs, the semiconductor chips that make artificial intelligence possible. But as the year wore on, that pattern changed. Equity investment was topped up – and eventually overtaken – by eye-watering credit deals. Records were set for both the largest ever corporate bond issuance and biggest private credit deal within a month of each other.
A pattern of frenetic and astronomic valuations, a vast proliferation of debt and enormous punts on an as-yet unproven technology (and an even less well-established financial track record) would, it is safe to say, fall within most people’s definition of a financial bubble. And yet, there remains a not insignificant chance that the reward at the end of this fourth industrial revolution could be so enormous that despite all of these bear signals, many market participants still think the rally has further to run.
All of which is to say, the question – or questions – on almost every investor’s lips as we head into next year is a simple one: are we in an AI bubble? And if we are, will it burst in 2026?
We turned to some of the Square Mile’s brightest financial minds to find out.
Stephen Yiu, chief investment officer, Blue Whale Growth Fund
Will the AI bubble burst in 2026? No (but depends where you look)
Rationale: Not all bubbles are the same. Unlike tulip bulb prices, which collapsed once they exceeded the value of a collector’s item, or the prices of telecoms companies, which offered value but only far into the future and were hampered by infrastructure frictions, AI is already delivering tangible results. Consider the speed of adoption of ChatGPT, which reached 800m users in just three years – compared to the internet, which took 13.
Follow where the AI spending is going, not AI per se. As in any technology revolution, the AI journey is not a straight line; money will be wasted, with winners and losers, but this does not make AI unreal. Avoid the dot-com era bubbly valuations among the “AI spenders,” but back the “AI infrastructure” companies that are highly profitable and attractively valued – more so than ever.
Michael Browne, global investment strategist, Franklin Templeton Institute
Will the AI bubble burst in 2026? No
Rationale: 2026 will be the year AI proves its worth. This is when early adopters demonstrate real productivity gains and measurable profit impacts, separating hype from reality.
Organisations that moved first are already reaping significant competitive advantages in operational efficiency, innovation velocity, and market position. Those who followed second are now scrambling to close the widening gap, investing heavily to catch up.
Third movers face an insurmountable challenge – the distance is too great, the momentum too strong. The winners are emerging now, and 2026 will crystallise their market dominance. This isn’t a bubble bursting; it’s validation arriving.
Tim Dingemans, investment strategist, Omnis Investments
Will the AI bubble burst in 2026? Yes
Rationale: To be or not be a Cassandra? The incredible rally in AI shares raises concerns. When you consider the extreme concentration of earnings in a handful of stocks, the vast capital expenditure outstripping cash flow (most bubbles fail from a lack of cash flow), and the likelihood hardware will be out of date before fully depreciating from balance sheets, diversifying out of the Magnificent 7 becomes an attractive and sensible prospect.
The bottom line is it’s impossible to know exactly when the AI bubble might burst, but to use an American football term, we are “in the end game”. So yes, I shout from the Walls of Troy, “beware of tech companies bearing AI gifts!”.
Daniel Morris, chief market strategist, BNP Paribas Asset Management
Will the AI bubble burst in 2026? No
Rationale: We do not believe there is currently an AI bubble, hence nothing to burst. Consider earnings and multiples for the NASDAQ 100.
The concern for the earnings side is that they will not rise enough to generate a return on the investments being made. We see numerous ways that tech companies could monetise the models and in any event, capital expenditure is not forecast to continue increasing at the same rate indefinitely.
As for valuations, while high, forward multiples can fall if the index appreciates at a slower rate than earnings, which would still generate a decent return for investors.
Ian Lance, co-manager, Redwheel and Temple Bar Investment Trust
Will the AI bubble burst in 2026? Yes
Rationale: Identifying a bubble is relatively straightforward whereas calling the end of one is harder as it involves predicting changes in the attitudes of market participants.
There are, however, clear signs that investors are starting to question whether the hyper scalers will ever make a return on the huge amounts of capital expenditure they are committed to.
Oracle’s credit default swaps blowing out to similar levels they reached in the global financial crisis is just one example of a shift in investors attitudes and this could be a precursor to the bursting of the AI bubble.
Will Nutting, founder, Nutstuff
Will the AI bubble burst in 2026? Yes
Rationale: The key issue for Tech and AI stocks heading into 2026 is not growth, but simple concentration. Technology now represents roughly a third of the S&P 500 nearer 40 per cent once AI-adjacent platforms are included turning the index into a single, capital-intensive bet.
It worked while capex was accelerating, margins were expanding and cheap money masked the cost of scale. But AI is beginning to resemble a cathedral project rather than a software rollout: magnificent, transformative, and ruinously expensive to build, with payback measured in decades rather than quarters.
When half the market is priced as if the scaffolding will be dismantled quickly, even a pause in construction forces a reckoning. In 2026, tech does not need to break to disappoint it merely needs to stop defying gravity.