Neocloud scale-up will ignite a new M&A cycle in GPUaaS-compatible data-centre infrastructure 

The combined growth in volumes and rationalisation in tenancies from neoclouds is set to trigger a new wave of infrastructure investments in AI-compatible data-centre infrastructure in 2026. 

Neocloud players are in a scaling phase that began in 2025, driven by rapid advances in the sophistication and take-up of AI use cases, which in turn generates additional infrastructure requirements. Some of these players, who offer lean cloud stacks centred on GPU compute/GPUaaS, are already cementing multi-billion-dollar commitments. In the USA, Applied Digital’s two 15-year CoreWeave leases total roughly USD7 billion, signalling a structural shift from opportunistic, short-tenor deployments to long-term contracted capacity. 

This shift is now spreading across Western Europe, where early multi-year partnerships such as Eclairion–Fluidstack are maturing. In 2026, we expect neoclouds to lock in larger pre-leased blocks (10- to 15-year leases, staged MW ramps and indexed escalators), lowering their effective rents while improving revenue visibility. 

However, investors must be aware that the investment case for financing the construction of new data centres, aimed at hosting neoclouds and AI workloads, differs from established hyperscale data centres. 

First, the range of target locations is changing and getting broader. As neocloud workloads are less latency-sensitive, demand is moving away from high-cost FLAP-D hubs toward lower-cost regions such as the Nordics, Iberia and Tier-II cities. These markets offer faster delivery and cheaper, greener energy. Second, from a technical point of view, some neoclouds are considering Tier I/II equivalent reliability for training workloads, allowing lower capex builds and reducing time to market. Third, new risks must be carefully considered and underwritten, particularly in relation to counterparty concentration and technological obsolesence. On counterparty concentration, neoclouds are largely dependent on a limited number of AI leaders and on privileged access to top-end GPUs. On technological obsolesence, the all-important liquid cooling designs remain unstandardised, which brings the risk of high retrofit costs, although modular data centre architectures help mitigate part of this risk.

We will see more ‘4-to-3 mergers’ in European mobile markets 

The European Commission appears increasingly open to mobile-market consolidation. Historically resistant due to concerns over market concentration and consumer impacts, the Commission is now placing greater weight on the need for large-scale investment in networks to support digital transformation, improve coverage and bolster resilience. With this shift in emphasis, the long-stalled debate about whether Europe’s four-player markets should move to three has reopened. 

The UK has already crossed that threshold with the Vodafone–Three merger finalised in 2025. Although the approval process was challenging, the outcome will embolden operators and investors in the remaining major European markets which still operate with four networks. 

France is the most plausible candidate for a deal in 2026. A EUR17 billion consortium bid for SFR was rejected, but the strategic logic remains compelling, and discussions about a revised proposal continue apace. 

Zegona (Vodafone ES) is the natural target in the Spanish market, but none of its rivals are ready to acquire it in 2026. In Italy, a WindTre–Iliad tie-up is the cleanest option, but would still be a slower process than the SFR deal in France. While 1&1’s launch in Germany has been less disruptive than new entries in other markets, a joint-venture structure involving 1&1 and a larger operator, likely Telefónica, is possible. Financial and structural hurdles suggest a timeframe that extends beyond 2026. 

AI will influence valuations for ICT and managed service providers 

For the majority of ICT and managed service providers (MSPs), AI-powered processes are now integral components of the business model, but the role of AI is continuing to evolve. As an early adopter of AI use cases – particularly in customer support and sales – this industry is already polarising into AI winners and losers. 

2025 provided evidence that successful AI adoption by ICT and MSPs is translating into improved valuations across Europe. The KPIs of such companies are being transformed by the impact of AI on their value proposition (custom AI platform + service bundles), service delivery (managed services with AI-automation), compliance requirements (EU AI Act, US frameworks) and talent demand (shift in roles needed from junior to senior roles).  

Analysys Mason expects to see EBITDA multiples increase by 1–2 turns for those ‘AI winners’ that can demonstrate increased sales performance, service throughput and profitability relative to their pre-AI trading average. Conversely, ‘AI losers’ are likely to suffer in comparison to their higher-performing peers, resulting in multiples that are reduced by 2–3 turns. The competitive benchmark performance pressure will be accompanied by client pushback on pricing and demand for innovative solutions, resulting in decreasing profitability and a persistent struggle to become ‘AI future-fit’. The combined effect will create real headwinds: reduced profitability and a persistent struggle to become ‘AI future-fit’.

The gap between the leaders and laggards among software companies will widen 

In 2026, we will see premium multiples paid again for the leading software players, while the less attractive targets will yield much lower prices. This trend will increase pressure on processes as value expectations, set by the leaders, are not met.

2026 will see further widening of the gap between the best software players and the rest. The EBITDA multiples attained by laggards will be in the mid-teens, but the leading industry software players will achieve multiples above 20. 

From a growth perspective, we will see prices on annual recurring revenue (ARR) multiples of above 7x, and potentially more than 9x for the best and fastest-growing software companies. For slower-growing software companies, the valuations will be less than 5x ARR next year. 

Looking ahead to future market developments, AI-driven software companies that can accelerate code development and expedite product launches are expected to experience significant growth and capture substantial market share at an unprecedented pace. For the first time in the history of software, AI-native software businesses will be able to provide this growth without large funding needs because the development costs will be fuelled by rapid growth in monthly recurring revenue (MRR) and the ability to reach product profitability within months.  

Refi or retreat: 2026 will be the year of truth for FTTH 

In 2026, lenders will devote substantial time to reassessing their fibre-to-the-home (FTTH) loan books. There are multiple triggers converging at once: existing facilities are reaching maturity; project finance structures are being converted to brownfield financing packages; platforms are seeking additional capital to deliver roll-out plans; and in some cases, major balance-sheet restructuring will be needed to avoid bankruptcy. 

This crescendo will create a natural moment for lenders to take stock of performance and rebalance exposure. With tens of billions of euros in debt financing at stake in 2026, we expect firmer decisions on who gains continued support and who does not. This is likely to result in a mix of refinancings, selective equity cures, asset sales and, in some instances, formal restructurings. 

Many FTTH infracos have not managed to perform in line with the original banking case. Lenders with exposure to a mix of high performers and underperformers will be less tolerant of under-delivery in a climate of persistently higher interest rates, recent bankruptcies in the sector and the shift to a more mature market phase. 

A credible, evidence-based path to profitability will be a prerequisite. Borrowers will need to demonstrate sustained improvements in build and activation costs, committing to operating efficiency and demonstrating momentum in take-up, not just plans. Winners will secure ongoing backing on tighter terms; underperformers will face capital rationing and strategic alternatives.

A ‘reality check’ will cut valuations for speculative data-centre landbanks 

Over the course of 2026, speculative accumulation of land and power will run out of steam as the market focuses on demonstrable delivery. The relentless boom in AI tools and infrastructure has sustained a rush among developers and investors, who have spent recent years racing to secure sites and grid capacity. Real-estate and energy groups continue to line up projects that take years to come to fruition. Europe’s current 10GW of operational data-centre power is expected to rise two- to three-fold by 2030. Yet the longer-term pipeline is far larger: Italy alone has over 42GW of grid-connection requests for data centres, a scale well beyond what infrastructure, customers, or financing can support. Capital discipline is now coming into greater focus, with investors moving away from the speculative capex of previous years.

Securing large blocks of ‘powered land’ remains difficult in established data-centre hubs, where demand is strong and grid capacity tight. In more remote locations – often targeted for AI builds – the pipeline of powered land now far exceeds near-term demand. Time-to-power is the key differentiator for securing customer contracts. Sites with uncertain grid-delivery dates, complex permitting applications and long construction cycles are being deprioritised by off-takers, and many prospective land positions are unlikely to become operational facilities.

Investors now need to judge whether pipelines will reliably translate into contracted revenue, and the value of any speculative capex incurred on landbanks and site development requires careful assessment. Hyperscalers and neocloud platforms will favour partners with credible landbanks and the proven ability to deliver capacity on time while managing increasingly complex, high-density, liquid-cooled environments in the AI era. In 2026, operational credibility and delivery track record will be decisive factors in valuing data-centre pipelines.