As attention has slowly but inexorably shifted from the scale of the AI opportunity to what might go wrong, three major concerns have emerged among investors.

The first is that there is simply too much competition as is typical of literally every new technology. Inevitably, a number of startups - and even major companies - will fall by the wayside as their approach is surpassed by others. Some investors will win, many will lose. Secondly, the increasingly interconnected web of self-dealing among major AI companies has many analysts and investors questioning whether this is a sign that the AI market has peaked and that the businesses involved are now shoring each other up in hopes of keeping the momentum going, which is usually a vain attempt. And thirdly, there are concerns that the entire industry is so dependent on Taiwan for the manufacturing of critical components that a blockade or invasion by China could crash the industry. JL

Alex Daniel reports in Quartz:

For the past two years the dominant question in AI has been how big the opportunity is. More recently, attention has turned to what could go wrong. 1)Fears that the sector is overheating are increasingly front of mind. Of particular concern is a complex web of $1.4 trillion in deals around OpenAI with the likes of Nvidia, Oracle, Microsoft, and others. 2) Taiwan, where much of the world’s most powerful AI hardware is manufactured, and where tensions with China remain high. 3) Scores of companies are racing to bring near-identical products to market. “There is too much competition. Everything that works gets copied in a month. Whenever a new idea gets funded and looks promising, three months later there are 20 other teams chasing the same thing.”

For much of the past two years the dominant question in AI has been how big the opportunity is. More recently, attention has turned to what could go wrong. Through the second half of 2025, warnings about an AI bubble grew harder to ignore, with sharp market swings, eye-watering valuations, and circular investments sparking comparisons to the dot-com bust. Yet the sums pouring into the sector keep growing.

That leaves investors in an uneasy position. AI is too important and fast-moving to ignore – but if the boom falters, they will be the ones to take the hit. Heading into 2026, we asked the people writing the checks what worries them most.

What could derail the boom

Guru Chahal, a partner at Lightspeed Venture Partners, which manages about $35 billion, said the prospect of another global shock hitting the economy is enough to “keep me up at night,” after five years already shaped by Covid, Russia’s invasion of Ukraine, and President Donald Trump’s tariffs.

You wouldn’t know it from the firm’s recent activity. Lightspeed, an early backer of Anthropic, Elon Musk’s xAI and French AI challenger Mistral, raised $9 billion in December – the biggest fund in its history – to back fresh bets on the technology.

Still, the industry’s reliance on advanced chips leaves it exposed to shocks far beyond Silicon Valley. One of the most sensitive pressure points is Taiwan, where much of the world’s most powerful AI hardware is manufactured, and where tensions with China remain high.

“If something disrupts datacenter builds or GPU availability – a Taiwan crisis, another pandemic, trade restrictions – the entire AI buildout grinds to a halt,” Chahal told Quartz. “We’re making investments in AI companies, both apps and infrastructure, predicated on continuous expansion. One major supply shock and you don’t have enough compute to grow.”

‘Too much competition’

Meanwhile in San Francisco, scores of companies are racing to bring near-identical products to market. “There is too much competition. Everything that works gets copied in a month,” said Inaki Berenguer, managing partner at LifeX Ventures, which focuses on AI bets around climate and health. “Whenever a new idea gets funded and looks promising, three months later there are 20 other teams chasing the same thing.”

Part of the problem is that startups are building products on top of the same AI models. Many of them can therefore launch, get pilot projects off the ground, or win small contracts quickly. That can make it harder to tell which firms are genuinely strong and which are simply benefiting from the surge of interest in the tech.

“Most of these companies don’t have any real moat beyond speed of execution,” Berenguer said. “And that is scary as an investor.”

“AI investing seems easy, but it’s not,” added Payton Dobbs, a partner at Hoxton Ventures. “Signs of momentum are everywhere which can be a distraction.”

And with the likes of OpenAI and Google rolling out new products and updates at breakneck speed, features that once helped startups stand out – like AI-assisted shopping – can quickly become standard.

Lexi Novitske, general partner of Africa-focused growth fund Norrsken22, said: “We worry about backing something that looks special today but ends up as a commodity layer, so we push hard on commercial viability and whether being ‘localized’ is a real wedge to scale.”

If the bubble bursts

Fears that the sector is overheating are increasingly front of mind. Of particular concern to analysts is a complex web of about $1.4 trillion in deals around OpenAI with the likes of Nvidia, Oracle, Microsoft, and others. The ChatGPT maker is forecast to have revenues of less than one-thousandth of the planned investment this year.

Even Alphabet chief executive Sundar Pichai warned recently that the tech industry can “overshoot” in investment cycles and that “no company is going to be immune” if the spending boom collapses.

In that scenario timing could be all important. Companies that raised money at high valuations might struggle to secure more funding on similar terms, while those looking to exit may not be able to. That can turn a strong bet on its head for investors.

Mikael Johnsson, co-founder and managing partner of Oxx, said a collapse in public company valuations could eventually have a “ripple effect into the value and exit-ability of private AI companies.”

“We do worry about a broader market correction,” added Novitske, “so we spread our bets and stay disciplined on business fundamentals.”

Bubble concerns were brought into sharp relief again last month when Oracle’s stock fell by as much as 16% after an earnings report showed its AI spending far outpacing returns, wiping about $70 billion off its market value.

Others argue that while parallels with the dot-com bubble are understandable, the current cycle is harder to judge. Dobbs said: “Similar to that period there is a rising tide lifting all boats, but what is different now is that more companies are generating revenue and big tech is being held accountable in the public markets.”

Still, that will be little comfort to most investors. Money keeps pouring into AI and the technology keeps advancing, leaving little room to step aside. But as bets grow larger, so do the consequences if the cycle turns.