Katharine Wooller is a respected commentator in bleeding-edge banking and financial services technologies.
Discover top fintech news and events!
Subscribe to FinTech Weekly's newsletter
Read by executives at JP Morgan, Coinbase, Blackrock, Klarna and more
Much hand wringing and column inches have been devoted in recent months to pondering if the riotous investment in AI supersedes reality and suggesting that AI maybe a bubble about to implode.
Finger-pointing analogies are drawn with other investment cycles that have been more hype than substance: the tulips of XYZ and the dot.com era of the new millennium. Certainly, there have been huge returns for those lucky enough to invest in the AI titans early: $1,000 USD invested in Nvidia before their IPO would at its peak be worth $8.3m USD which AI bulls - quite understandably - feel is a rate of return unlikely to be repeated in the sector!
On the face of it the sheer amount of money being thrown at AI suggests there is simply too much momentum for it to be a flash in the pan. Leviathan tech companies, such as Amazon, Meta, Microsoft, Alphabet are investing heavily; spend on AI infrastructure for 2025 across these companies will be in the region of $400bnUSD, one of the largest spending cycles in history.
Much has been said about exactly how the money is being invested in the current cycle. Many private AI firms have been able to raise billions on vapour-wear – that it so says no MVP, or indeed product at all – rather an idea and a lot of promo.
There is also some interesting circular finance with AI firms investing in mutual investment and partnerships; a diagram of where the investment flows look a lot like a plate of spaghetti and following where the money has gone will quickly give you a headache. This creates huge risk from the inter-reliance, and even a quick review of how the epic amount of investment creates revenue loops that can artificially inflate valuations.
There’s also a question of who is propping up the AI valuations, with some big tech firms create opaque structures to shirk the spend off balance sheet, which begs the question who foots the risk if it goes wrong.
There is also the question of the rate of adoption of AI. Certainly, the vendor landscape is complex and in need of significant consolidation, moreover many projects languish in POC stage, and the ROI often difficult is ascertain. However, in my view, this is symptomatic of any new technology, and a more balance view should be taken of AI’s potential, which is ultimately what the investment cycle depends on – a considered bet on where the technology will be in the medium and short term.
Of course, this relies on the customers deriving value from deploying the technology. Few FTSE of NASDAQ firms have a strategy that omits AI, and it seems that it offers significant promise to reduce cost and risk across most industries. Indeed, financial services is postulated as one of the industry’s most likely to be disrupted by AI, a Softcat survey in 2025 of technology leaders found that 48% selected AI as a priority, and Gartner found an 88% increase in spending in relation to AI.
Let us not downplay that huge disruption that AI offers, it is hard to argue that it is not a genuine technological breakthrough. ChatGPT (albeit that it doesn’t generate profit!) is universally accepted as a productivity tool from school kids to CEOs, across pretty much every industry and business function. Frankly, being able to justify even few % in productivity gains, the effect on most businesses’ bottom lines would prop up current AI valuations. Further the huge progress made by advances in GPU, custom chips, and model efficiency ensure the future viability – it would be a disaster if the theoretical use of AI was hindered by under-powered infrastructure, investing someone ahead of market demand is, in the crude reality of day, a good thing.
Granted there some significant breaks to adoption that hinder progress. Of particularly important within our own sector, there is the elephant in the room of regulation – or lack thereof! Worldwide we are still only in the early stages of working how if and how we apply rules to the usage of AI.
There is a broader question of ethics, and how we ensure AI is used responsibly, with early promising specialist tech solutions for governance and assurance. There are significant issues in ESG, and particularly in the huge environmental cost of AI both in the significant power needed, and in the depreciation of physical infrastructure. Whilst these issues exist may businesses are reluctant to fully release the throttle on AI – rather they are taking a pragmatic “wait and see” approach and are following in the slip stream of early adopters. In my day job supporting innovation in 2000 financial services’ firms, I see much anxiety around firms keen to be neither first nor last in the AI arms race!
Technology is by its nature cyclical and investment thesis are always a “best guess” basis. We have moved on from the Tulip crisis of 1637 – we have, thankfully, an almost unlimited market for AI which sadly did not exist for the amateur investors that purchased futures in bulbs with little to no demand.
For a more recent example, the crypto old guard chuckle somewhat when we read about AI being over heated – Bitcoin lost 80% of its value in 2018, falling from $19,783 to $3,200 before then reaching an all-time high of $126,000 in 2025. The technology lost no potency even if the valuations had got ahead of reality.
Indeed, If I had a pound for every time I heard crypto was dead, I’d have retired a long time ago; I can’t help but think the same is true of the current AI nay-sayers. Whilst some correction in AI tech stocks is no bad thing, it doesn’t mean that either the technology has failed, nor that future demand is anything but strong. The advent of quantum computing is likely to put rocket fuel into AI, and indeed the share prices of those tech firms who stand to benefit from it.
