Investors are pouring billions into AI, but has supply now outstripped demand? Mike Smeed looks at how AI funding has become misbalanced
VCs have poured investment into artificial intelligence (AI) at break-neck speed. Atomico’s State of European Tech Report reveals that by the end of 2024, over $11bn will have been invested in AI companies by European investors, with our US counterparts deploying $47bn in the sector.
The figures paint a vivid picture of a flourishing AI market. Yet, the majority of this capital has been channelled into the development, not deployment, of the technology at scale.
Fear of missing out has led to an overfunding at the AI tooling layer, resulting in a (somewhat predictable) misalignment between dollars invested and consumer demand. At the same time, investors have largely overlooked the fundamental hardware and compute layer required to support the continuation of the AI boom. All signs point towards a need for a market correction.
An AI ceiling
Branded the ‘next industrial revolution’ by Jensen Huang, CEO of Nvidia, it’s safe to say that the power and digital infrastructure required to continue advancing AI at its current rate will be staggering. And with supercharged AI usage comes an insatiable demand for water, electricity and other resources.
Billions of dollars have already been invested in the physical infrastructure of AI, but this capital allocation pales into insignificance when studying the total investment in the sector. In reality, we’re effectively playing catch-up, building the foundations of the AI boom after it has taken off.
Without a considered redistribution of capital – one that acknowledges the need for improved physical infrastructure – we risk reaching an AI ceiling in spite of the vast amount of capital advancing innovation in the sector.
The VC redirection
Fortunately, VCs are taking note. Today, we are witnessing the beginnings of a market correction in how funds are deployed, with investors shifting their focus away from the crowded middleware layer and towards business-critical, above and below-the-stack opportunities.
The pressing need for more data centres has opened up a host of new investment avenues, particularly in early-stage companies redressing the significant environmental consequences associated with our ever-increasing use of AI.
Major institutions are raising mega-funds to capitalise on opportunities in these underfunded verticals, as evidenced by Blackrock and Microsoft’s AI partnership to invest in data centres and power infrastructure, one of the biggest investment vehicles ever raised on Wall Street.
This is not to say opportunities at the application layer should be discarded. Investment appetite remains intact. But I’m seeing many VCs adjusting their approach: prioritising long-term defensibility alongside first-mover advantage and, crucially, resisting the temptation to follow the hype. The next decade of AI innovation is dependent on robust physical infrastructure. An investment thesis that centres on longevity, as well as financial returns, is key to safeguarding the billions of dollars already deployed in the sector.
Mike Smeed is managing director at Inmotion Ventures