AI Edge attracting new investment and driving new financing structures

AI inference, proximity, reasoning and agentic services all depend on latency and resilience as the business case for new AI Edge data centres located around and close to major metros.

Into this space there is a whole new data centre category emerging. And it is attracting new investors. 

As European nation states move to attract investment it is metro markets that are expected to experience significant growth due to applied AI workload demand. 

This post AI model training stage will bring proximity AI, inference, reasoning and agentic AI into play. 

The open questions include: What is AI user proximity? Will it see 20MW-30MW and even 70MW facilities developed close to cites? What are the risk factors?  For investors? For Operators? Will this shift investor attention away from vast campuses to a proliferation of hundreds of standalone AI Edge data centres? 

This edge is a new proposition for investors and operators. Instead of the mobile base stations and micro data centres for smart cities, IoT and 5G that defined edge computing several years ago – new AI edge data centres are discussed in terms of many tens of megawatts being located in or close to major population centres. 

Such developments also raise questions on connectivity such as what distances represent practical proximity? 100m? 100KM?

Sebastien Bonneau, Partner at law firm, McDermott Will & Schulte LLP  says the new Edge is defined by the position of traditional data centre cloud services which grew big across FLAP+D. As such it represents an exciting proposition for traditional investors and new market entrants. 

‘Successful AI Edge infrastructure will rely on the fact that there are companies willing to buy storage and processing services for AI – that’s the rule of thumb. The question is in which locations? In some places it is not yet clear if there exists an offer to pay for these types of data centre services. Alongside storage and processing, AI edge requires high quality local connectivity and power availability. How interconnectivity is delivered is a major consideration. These are key factors in driving the investment behind the concept of AI edge.’ he says.

Pace of change is an additional consideration. AI Edge data centres will serve markets where the sheer pace of AI innovation is not vertical, but linear and horizontal to the data centre. 

The same but different

How AI Edge plays out presents a major opportunity for data centre investment. But despite similarities with cloud data centres this infrastructure will have marked differences.

Harqs Singh is CTO at InfraPartners, a specialist in prefabricated AI data centre solutions.

He points to the post LLM training stage of AI being where monetary value is being created. 

‘What developers of AI inference Edge DCs who are looking to attract funds need to consider and what investors of AI Edge need to understand is that these data centres are where AI is monetized. Models are trained and then loaded into inference sites for users to prompt and leverage for tokens and intelligence. Inference is required for general intelligence models but also for industry specific models and sovereign AI capabilities which, by definition, are local. Inference at scale will require distributed high-performance infrastructure which is responsive to user needs, both humans and machines. This incorporates low latency and security considerations. To be able to deliver distributed AI inference at scale a standard rapid deployment model will be the only way, and this is where prefabricated data centre solutions provide the greatest value.’ 

An example of an AI edge firm which has attracted investment for European expansion is US based Metrobloks. In a statement financier Eurazeo, through its fund EZORE, this year announced a €100 million strategic partnership with Metrobloks. 

It said: ‘This collaboration aims to develop next generation data centres at the edge of city centres, providing essential infrastructure dedicated to low-latency applications in artificial intelligence and cloud fields requiring proximity to end-user and high technology standards.’

The firm’s strategy is to acquire land, develop and operate data centres in strategic FLAP-D Tier 1 markets as well as in emerging Tier 2 markets like Milan, Madrid, Lisbon, Berlin, and Manchester. Pierre Larivière and Riccardo Abello, Co-Heads Real Estate at Eurazeo, said: ‘The aim is penetrating the rapidly expanding data centre sector, capitalizing on its high potential.’

For InfraPartner’s Singh, this potential is evidenced by artificial intelligence representing ‘the largest capital deployment in human history.’ 

‘Early investors who allocate resources strategically stand to realize significant long-term returns. To capture this value, infrastructure must be designed for flexibility. The ability to deploy the latest GPUs—such as NVIDIA’s cutting-edge accelerators—within a flexible, upgradeable data centre is critical. As AI technology evolves rapidly, static infrastructure risks obsolescence and diminished returns. Power will be the only constant at each site, the key to maximizing return on that power lies in deploying the most advanced compute possible—something only achievable through a prefabricated upgradeable design,’ he says.

Singh believes the key element to address AI challenges is infrastructure that keeps pace with technological change while optimizing capital efficiency.

Also on the infrastructure side delivering the necessary power to the AI presents another investment opportunity. 

Ben Pritchard, CEO, at energy solutions provider AVK says: ‘The new edge represents a fundamental shift in how we approach data centre power infrastructure. AI workloads will drive unprecedented energy demands and push compute closer to end users. We're seeing a critical need for decentralised, resilient power solutions that can scale rapidly without compromising sustainability.’

‘What excites me most about the new edge is its potential to transform data centres from energy consumers into community energy hubs. Through intelligent microgrid design, edge facilities can balance their own operational needs while supporting local energy infrastructure – creating a more distributed, resilient grid that benefits everyone.’

For Pritchard the investment opportunity is clear. ‘As hyperscalers chase nuclear solutions that remain years away from deployment, companies building practical, scalable microgrid solutions for the new edge are positioning themselves at the centre of the next wave of digital infrastructure innovation.’

Bonneau points out that ‘the industry is very entrepreneurial at this point in time.’

‘Investors now know they can’t always play at the upper end, so they have to be flexible. They are interested in less vanilla investment. The industry is entrepreneurial – that is what is exciting – even traditional investors know there are not too many opportunities to go after. And when they see opportunities, they know they are competing – and not always selected – so there is a focus also on smaller, more risky, more innovative approaches to serving the market.’ 

Neil Cresswell, Vice Chairman, VIRTUS Data Centres, says “It’s a rapid pace industry and we’re seeing a faster shift than initially anticipated. Inferencing workloads are rapidly expanding across Europe’s major cities, not only the core markets, but also the rapidly expanding tier II markets.” 

Players seeking faster returns 

In AI Edge data centre investments traditional approaches remain with some developers seeking equity to buy the land – then borrowing to develop. But there are also many different players entering the market who are seeking faster returns – as opposed to traditional players looking at customary time to value. 

The brings new pools of capital from different types of investment player who are innovating on deal structures. 

‘Deal structures are more varied. We’re seeing more joint venture deals’ says Bonneau, ‘but while the incentives are there the need remains to find land, to find power, to find money, and find someone with industry expertise.’