Is the AI bubble about to burst?
18th September 2025
Will AI continue to grow, or has it reached the limit of what it can do?
Over the past few years, the talk and hype around AI has been extreme: it quickly went from something that was more fiction than fact to an extremely useful tool that most people I know have utilised in one way or another.
Recently though, the question has been raised as to if the AI bubble – “bubble,” here referring to the economic success AI products have been seeing – is about to burst. With so many businesses relying on AI, we wanted to break down the ways in which this might be true, or might not, to give you an idea of what the future of AI may bring.
The case for a bursting AI bubble
Overinvestment without returns: an MIT study found that 95% of AI pilot projects don’t move beyond initial stages*, with only 5% delivering measurable business value. Some are forecasting that up to a third of generative AI projects will be abandoned*2 by the end of the year.
Inflated valuations: companies like NVIDIA and Microsoft have hit $4 trillion valuations, driven largely by AI hype – with a few companies dominating the tech market, it reflects the landscape that preluded the dot-com crash.
Technical limitations: AI models still struggle with keeping information accurate and unbiased, as well as having extensive energy requirements, leading to sustainability issues.
Expert scepticism: MIT economist Daron Acemoglu argues that AI’s economic impact is exaggerated*3, with AI pioneer Gary Marcus*4 warning that the current development model is unsustainable.
The case against a bursting AI bubble
Long term potential: AI adoption is rapidly growing, and over time it has the potential to become a general-purpose technology, on the same level as electricity or the internet.
Massive investment momentum: Goldman Sachs estimates $200 billion in global AI investment*5 by the end of 2025 – that amount of money doesn’t indicate a decline in AI popularity.
Strategic integration: successful AI projects have focused on integrating AI into workflows rather than chasing hype, with companies already using it in the right way to great effect.
Market correction: analysts suggest we may be in a “correction phase,” rather than a full collapse – the hype may be peaking, but with innovations still being made, the bubble won’t yet burst.
So, is AI on the brink of failure? To us it seems unlikely. As it becomes clearer what the limitations of AI are, the hype will settle down, leading to the bubble shrinking, but with AI technology being so widely adopted and becoming essential to businesses’ workflows, it is doubtful that it will collapse completely.
To be clear, we are not economists – we are purely speculating based on the information we have, as well as our own experiences with AI (and how our clients use it). We could be wrong, and AI could turn out to be just another phase that we move past, but as it stands, we doubt it.
If you have any more AI-related queries, please contact Interfuture Systems.
YouTube: https://youtu.be/1UoV_vlOeCc
*2https://techcrunch.com/2024/10/20/investments-in-generative-ai-startups-topped-3-9b-in-q3-2024/
*3https://mitsloan.mit.edu/ideas-made-to-matter/a-new-look-economics-ai