The telecom industry loves to talk about AI, sovereignty, and satellite disruption. But beneath the noise, the real story in 2025 is structural: capital constraints, concentrated infrastructure, overbuilt AI data centres, and resilience that only works if you can afford it.

This report examines what actually happened in 2025 — and what that means for 2026.

The past two years have been defined by recurring disruptions: subsea cable cuts, geopolitical fragmentation, power instability, and infrastructure strain. Rather than focusing only on individual outages, we chose to track recurring structural patterns across the network layer to understand what pressures are building and what they may signal for 2026.

In 2025, six themes emerged repeatedly across operator discussions:

●     Infrastructure concentration remains high.

●     Subsea dependency continues to expose systemic weakness.

●     Capital is tightening while costs are rising, making resilience increasingly dependent on balance sheet strength.

●     AI infrastructure investment is outpacing actual traffic demand.

●     Resilience claims face increasing scrutiny beyond marketing language.

●     Regulatory and geopolitical fragmentation complicate resilience planning.

None of these signal collapse, but they do signal stress. And stress is cumulative.

While this report focuses on the Eurasian connectivity layer, the implications extend beyond the region. Transatlantic cloud ecosystems, U.S. hyperscalers operating in Europe, and multinational enterprises relying on Europe–Asia transit are directly exposed to these same infrastructure dynamics.

Many U.S.-based cloud providers and AI platforms rely on European backbone corridors for intra-European delivery, transatlantic exchange, and Europe–Asia routing. Changes in resilience economics or corridor concentration in Europe, therefore, shape global cloud performance and cost structures.

Trend 1: Infrastructure Is Still Physically Concentrated

What We Saw in 2025

Despite diversification efforts, Europe’s backbone remains heavily centralised around key hubs.

Frankfurt remains described internally as a “critical” interconnection hub — effectively mandatory for ecosystem participation. Paris–Frankfurt terrestrial routes frequently pass through Strasbourg. Historic cable landings cluster in London Docklands.

On Europe–Asia corridors, multiple cables were simultaneously affected because of submarine bottlenecks in the Red Sea and then shared terrestrial fiber routes across Egypt.

This shows that many routes described as “diverse” are only different on paper — in reality, they often share the same physical paths.

Why It Matters

When routes share physical corridors, failures become correlated.

Inside Europe, rerouting usually increases latency but preserves connectivity.

Intercontinentally, especially Europe–Asia, rerouting capacity is far more limited.

The result is asymmetrical resilience: regional robustness and intercontinental fragility.

2026 Outlook

If corridor diversification does not accelerate:

●     Intra-European outages likely continue to degrade performance rather than cause collapse.

●     Europe–Asia remains exposed to concentrated transit chokepoints.

In practical terms, this means that if a major transit corridor is disrupted in 2026, multiple operators could experience simultaneous performance degradation. Rather than full blackouts, the more likely outcome would be sustained latency increases, temporary congestion on alternative routes, and reduced service quality for cross-border traffic. For enterprises relying on low-latency international connectivity, this could translate into operational slowdowns and have an immediate financial impact.

Trend 2: Limitations of the Subsea Routes Create Structural Risk

What We Saw in 2025

More than 90% of the world’s intercontinental traffic is moved via subsea cables. Even with terrestrial routes in place, many choose traditional routing via subsea. Most Europe–Asia traffic still depends on subsea corridors transiting geopolitically unstable regions or avoiding higher operational costs.

Operators reported that:

●     While subsea repair times remain significantly longer than terrestrial repairs is not news, 2025 showed record delays. Terrestrial fiber cuts are typically resolved within 8–12 hours, while subsea cable repairs can take a minimum of two weeks, depending on vessel availability and location. 2025 showed some submarine cables being out of service for more than 250 days.

●     Demand for terrestrial alternatives increased following high-profile outages, even though current capacity means they serve as resilience routes alongside major subsea corridors rather than full substitutes.

●     Subsea capacity supply has not kept pace with diversification demand.

Simultaneous disruptions across multiple systems highlighted the illusion of redundancy when infrastructure converges on shared terrestrial crossings.

Why It Matters

Subsea outages are not short-lived inconveniences. They can last months.

Customers continue paying during outages. Service credit mechanisms are limited or nonexistent. This is a feature of the industry — and rarely discussed publicly.

True redundancy requires:

●     Physically diverse paths.

●     Matching capacity on backup routes.

●     Capital discipline to maintain both.

In reality, few operators invest equally in primary versus fully diverse backup paths, because building equivalent redundancy across multiple corridors is expensive.

2026 Outlook

Terrestrial routes are likely to see continued strategic interest in the short term, particularly following recent subsea disruptions. However, their role will remain constrained by economics and physical infrastructure limits.

Whilst terrestrial capacity has a sticker price which makes it appear significantly more expensive than subsea, often 2x, this is misleading. It is often protected by default, so it spends less time non-operational. When it does fail, it is repaired much quicker, and service credits are payable during outages. When you compare like-for-like in terms of the number of subsea circuits necessary to maintain comparable performance, terrestrial connectivity often has a lower cost.

Large portions of global traffic cannot migrate away from subsea infrastructure because many intercontinental routes—such as Europe–U.S. and Asia–U.S.—have naturally no terrestrial alternatives and rely entirely on submarine cables. While some Europe–Asia traffic can be rerouted via terrestrial corridors, the capacity of these routes remains limited relative to subsea networks. In practical terms, this suggests selective budget reallocation toward secondary terrestrial paths where they are geographically viable—particularly for enterprises with significant Europe–Asia exposure.

However, large-scale migration away from subsea infrastructure is not feasible. Instead, operators are likely to adopt hybrid resilience strategies, maintaining submarine cables as the primary backbone for intercontinental connectivity while selectively adding terrestrial redundancy where possible for high-value or latency-sensitive traffic.

Trend 3: The Capital Squeeze

What We Saw in 2025

Wholesale prices continue declining. Internal references cite annual erosion around 10%.

At the same time:

●     Hardware costs increased.

●     Electricity costs remain elevated.

●     Hardware lead times extended (from ~8–10 weeks to ~26 weeks in some cases).

●     Capital intensity rose (a project that previously required roughly €50k of investment would now require closer to €75k to generate similar revenue).

Private equity-backed infrastructure players continue aggressive deployments, often overbuilding each others routes, failing to add true network diversity, whilst resulting in excessive price compression on competitive routes.

Smaller operators face margin compression with rising capital requirements.

Why It Matters

Resilience is capital-dependent. If prices fall while infrastructure costs rise, only well-capitalised players can maintain or expand resilience investments; at the same time, investors risk never seeing a return on that money while companies are building Ponzi schemes seeking bigger investment after previously increased investment. 

The squeeze on resilience investment forces trade-offs:

●     Delay upgrades.

●     Reduce diversification.

●     Consolidate.

●     Exit.

Capital discipline improves efficiency — but limits flexibility.

Infrastructure costs — hardware, energy, deployment — are rising, while wholesale prices continue to decline. This widens the gap between cost and pricing. Operators with access to capital or strong balance sheets can absorb margin compression and maintain aggressive pricing while continuing to invest in redundancy. Smaller or undercapitalised providers face a harder trade-off between staying price competitive and funding resilience.

If this dynamic persists, selective consolidation becomes more likely. Fewer operators in key corridors reduce competitive diversity and increase dependence on larger balance sheets. In the short term, customers may benefit from lower pricing. Over time, however, reduced diversity can increase concentration risk and limit infrastructure investment options.

2026 Outlook

Two plausible paths:

1. Price stabilisation allows reinvestment and resilience improvement.
2. Continued erosion accelerates consolidation and increases concentration risk.

Resilience may become less about engineering and more about balance sheets.

Infrastructure cannot be permanently subsidised by investor optimism; at some point, pricing and capital requirements must rebalance.

If consolidation accelerates, customers may face fewer provider options in certain corridors, reducing pricing leverage. Smaller regional ISPs could become more dependent on larger backbone providers, increasing supply chain concentration. Over time, less competition may slow infrastructure innovation and limit service differentiation.

This dynamic is particularly relevant for private equity and infrastructure funds with exposure to European telecom assets. Many of these investments were underwritten on expectations of stable long-term pricing, predictable bandwidth growth, and continued access to low-cost capital. Persistent wholesale price erosion challenges those assumptions by compressing margins while operators must still invest in new routes, optical upgrades, and resilience. If consolidation accelerates, asset valuations, competitive positioning, and long-term return assumptions may shift accordingly.

Trend 4: AI Data Centre Overbuild

What We Saw in 2025

Massive capital inflows targeted AI infrastructure:

●     Governments pursuing “sovereign AI.”

●     Private equity pouring money into AI data centres.

●     Hyperscaler-driven buildouts.

However, it remains unclear how much AI investment will translate into backbone traffic growth. In many AI data centres, most data moves within the facility itself rather than across international internet routes, meaning AI investment does not automatically lead to more backbone traffic.

Simultaneously:

●     AI facilities consume substantial power and water.

●     Countries like the Netherlands or Ireland face grid strain due to data centre concentration.

●     Housing and other infrastructure compete for electricity allocation.

Why It Matters

AI may dramatically increase local compute density — but not necessarily long-haul backbone traffic.

There is a risk of overbuild:

●     Too many AI data centres.

●     Overestimated traffic projections.

●     Infrastructure financed on hype cycles.

Telecom has seen this before: cloud, fiber overbuild, debt-fuelled expansion.

2026 Outlook

Likely outcomes:

●     Some AI data centre projects remain underutilised. This could result in delayed returns on infrastructure investment and more cautious capital deployment in 2026. For enterprises, this may mean slower rollout of AI-driven services than expected. For end users, it likely translates into fewer immediate improvements in AI-powered applications despite heavy investment headlines.

●     Selective consolidation. Consolidation may reduce the number of independent infrastructure providers in certain regions. In the short term, pricing pressure may continue, but over time, reduced competition will limit operational or route diversity and negotiating leverage for enterprise customers. End users are unlikely to see immediate price increases, but market diversity may narrow.

●     Increased scrutiny of power allocation and environmental impact. Greater scrutiny could slow approval timelines for new AI facilities, particularly in power-constrained regions. This may limit the rapid expansion of high-density computing in certain markets and shift new projects to regions with available energy capacity. For businesses, this could affect where latency-sensitive workloads are hosted.

AI will reshape access networks more than international backbone.

In practical terms, AI-related investment is more likely to influence last-mile and metro infrastructure — where end-user applications run — rather than dramatically altering long-haul international capacity. Consumers may notice improvements in local network performance before seeing meaningful changes in global backbone resilience.

For investors and analysts modelling exponential backbone growth tied to AI, current traffic observations suggest a more gradual demand curve than some capital allocation narratives imply.

Trend 5: Resilience vs. Marketing

What We Saw in 2025

“Resilience” became a dominant industry narrative.

Operators frequently market redundancy and resilience. Yet:

●     True redundancy requires equal capacity on alternate routes.

●     Many operators build backup routes with lower capacity.

●     Capacity-matching redundancy doubles capital exposure.

In one example, five routes existed into Taiwan; four failed; one remained operational due to prior overinvestment in route diversity.

Why It Matters

Resilience is not a press release – it is an engineering choice.

Without:

●     Capacity symmetry,

●     Supplier diversity,

●     Geographic divergence,

… redundancy is cosmetic.

The industry risks resilience theatre — talking more than building.

2026 Outlook

Customers will increasingly ask:

●     Are backup routes equal capacity?

●     Are routes physically separated?

●     Are suppliers independent?

In 2026, this is likely to translate into stricter redundancy requirements in enterprise contracts, including equal-capacity failover clauses and greater scrutiny of route independence. Operators unable to demonstrate true physical separation may face longer sales cycles or pricing pressure. For customers, this could mean modest increases in connectivity budgets for high-availability services, but improved performance stability during disruptions.

The net effect may be fewer low-cost, lightly protected configurations and a gradual shift toward more robust — and slightly more expensive — connectivity tiers.

Trend 6: Regulatory Fragmentation and Sovereignty Gaps

What We Saw in 2025

In 2025, RETN observed:

●     Europe’s heavy dependence on U.S. hardware vendors.

●     Cloud ecosystem concentration.

●     Regulatory filtering at access or backbone layers in multiple countries (including sanctions regimes, export controls, and divergent compliance requirements).

●     Increasing geopolitical complexity across 40+ operating jurisdictions.

Data may be stored locally. Infrastructure often is not locally controlled.

In practice, data sovereignty does not automatically translate into infrastructure sovereignty. Core networking hardware, optical systems, and cloud ecosystems used across European networks remain heavily sourced from U.S.-developed platforms. While data may be stored locally, the control layers — hardware supply chains, software ecosystems, and cloud architecture — often remain externally dependent.

Why It Matters

Data sovereignty without infrastructure sovereignty is partial independence.

Geopolitical tension can:

●     Disrupt hardware supply.

●     Restrict vendor relationships.

●     Alter regulatory obligations overnight.

2026 Outlook

Short term:

●     Dependency persists. Sovereign cloud initiatives and selective government migrations will expand. Any broader shift away from U.S.-origin hardware or cloud ecosystems, if it occurs, is likely to be gradual and concentrated in specific segments — particularly government workloads and certain regulated services. Core backbone hardware ecosystems remain deeply integrated, making rapid substitution complex and capital-intensive.

●     Fragmentation increases compliance overhead. For operators, this likely translates into higher procurement and compliance costs as vendor diversification and parallel certification requirements expand. For end consumers, immediate service changes are unlikely, but cost pressures at the infrastructure level may gradually feed into pricing or reduce the margin available for network upgrades.

Medium term:

●     Policymakers will remain busy.

●     Governments likely to increase scrutiny of critical infrastructure ownership.

●     Sovereignty rhetoric will intensify faster than infrastructure replacement.

In practical terms, increased scrutiny may slow cross-border acquisitions, lengthen regulatory approval timelines, and complicate vendor selection for new infrastructure projects. For operators, this could mean higher legal and compliance costs and slower deployment cycles. For enterprises and end users, immediate service disruption is unlikely, but infrastructure expansion in certain regions may progress more slowly, potentially affecting long-term capacity growth and investment pace.

Conclusion: The Industry Needs Realism

The telecom industry cycles through narratives: cloud, AI, satellite, sovereignty.

Underneath those cycles lies infrastructure economics.

2025 revealed:

●     Concentrated physical topology.

●     Subsea fragility.

●     Capital compression.

●     AI over-optimism.

●     Regulatory complexity.

●     Resilience gaps between rhetoric and engineering reality.

None of this is catastrophic, but it is cumulative.

The common thread across these trends is not collapse, but compression. Physical concentration persists. Subsea exposure remains unresolved. Margins are tightening. Capital intensity is rising. Supply chains are lengthening. The system continues to function — but the margin for error is narrowing.

In 2026, the organisations best positioned to navigate connectivity risk will likely be those who:

●     Stay focused on core network and service competencies.

●     Invest in capacity-matched redundancy and transparent resilience planning.

●     Avoid hype-driven overexpansion.

●     Be transparent about what levels of redundancy and service guarantees are realistically supported by pricing.

Resilience is not a feature you switch on — it is the result of deliberate investment, redundancy, and trade-offs.

Methodology

This report is based exclusively on qualitative analysis by RETN conducted in March 2026, including executive and operational leadership perspectives. Insights were derived from direct statements, cited operational examples, and numerical references. No external datasets, third-party research, or inferred metrics were introduced.