Cloud Is No Longer Neutral
How Infrastructure, Energy and Jurisdiction Are Reframing Europe’s Digital Architecture
For more than a decade, cloud computing was presented as neutral infrastructure — borderless, scalable and abstracted from geography. Compute became a utility. Data flowed across regions. Location was reduced to latency. The promise was frictionless abstraction.
That abstraction is dissolving.
Artificial intelligence is re-materialising the cloud. What once appeared as virtual capacity is now visibly grounded in hardware concentration, energy density, cooling systems and legal jurisdiction. The cloud is no longer merely a software layer. It is becoming a structural layer of power — physical, legal and geopolitical.
The decisive shift is not technological. It is architectural.
As Charles Sansbury recently observed:
“The shift toward Private AI is a direct result of companies having moved data too quickly to the cloud — what I call the ‘cloud hangover’. Organizations are now realizing that the real value of AI lies not in public data, but in their own private data that they want to keep within their security perimeter.”
Charles Sansbury, CEO Cloudera
Source: Interview with SiliconANGLE, October 2025
Sansbury’s use of the phrase “cloud hangover” is telling. It implies that the earlier phase of cloud migration — often guided by a blanket “cloud-first” doctrine — underestimated structural consequences. The first era of cloud was about convenience and speed. The emerging era is about control, resilience and long-term design.
The Rematerialisation of Cloud
Traditional SaaS workloads — CRM systems, ERP platforms, collaboration tools — required distributed compute capacity. AI workloads do not. Generative AI, large language models and inference at scale demand concentrated clusters of GPUs and specialized accelerators. These clusters cannot be located anywhere. They require proximity to high-capacity grids, advanced cooling and stable political environments.
Cloud is no longer a floating abstraction. It is an industrial installation.
Jensen Huang has repeatedly framed the AI race as a race to build “AI factories”:
“Every company will become an AI company. But AI cannot scale without a fundamental rethinking of infrastructure, power density and cooling. The race for AI is ultimately a race for the physical location of compute.”
Jensen Huang, CEO NVIDIA
Source: GTC Keynote 2024/2025
This is not marketing rhetoric. It is architectural reality. AI compute reshapes land use, energy allocation and industrial policy. The selection of a cloud provider increasingly implies a choice about where workloads are physically anchored — and under which national grid they operate.
The neutral cloud has become materially situated.
Jurisdiction as Architectural Component
In the previous era, compliance was often treated as an overlay — a legal box to be ticked after architectural decisions were made. That sequencing no longer holds.
Jurisdiction now shapes architecture itself.
The United States’ CLOUD Act permits extraterritorial data access under certain conditions. The European Union’s regulatory environment — structured around GDPR, NIS2 and the EU AI Act — embeds sovereignty, risk management and rights protection into infrastructure design. These are not minor policy differences. They produce different architectural logics.
Under the U.S. model, scale and data monetization dominate. Under the EU model, protection and controlled data environments are primary. An enterprise operating in Europe cannot treat cloud selection as a purely technical decision. It is a legal positioning.
Sebastiano Toffaletti articulated this clearly:
“Sovereign digital infrastructure is not an abstract ambition; it is a precondition. Without our own physical infrastructure in Europe, our administrations and industries remain structurally dependent on external actors operating under different rules.”
Sebastiano Toffaletti, Secretary-General, European DIGITAL SME Alliance
Source: Policy vision paper, “The year ahead: 2026 will make – or break – Europe’s tech sovereignty”
Cloud architecture is therefore no longer neutral with respect to governance. It encodes legal exposure and strategic dependency.
Energy as the Invisible Boundary
Perhaps the most underestimated constraint is energy.
The scalability of cloud once appeared linear: more demand could always be met with more servers. AI disrupts that assumption. A large language model query consumes multiple times the energy of a traditional web search. At scale, inference becomes a persistent load rather than an occasional spike.
A recent strategic briefing from the World Economic Forum warned:
“The growth of AI depends entirely on managing the energy nexus. Without careful coordination, AI risks choking on its own demand for electricity and water. Data architecture is already being dictated by grid availability rather than user proximity.”
World Economic Forum, Strategic Intelligence Briefing
Source: “The AI–Energy Nexus Will Dictate AI’s Future,” December 2025
Regions such as Amsterdam, Dublin and Frankfurt — major European data centre hubs — are experiencing grid congestion and regulatory pushback. In some cases, new data centre projects face delays due to power allocation limits. Energy is no longer an operational afterthought. It is a strategic constraint.
This shifts board-level thinking. A “cloud-native” strategy now has direct implications for ESG commitments. Workload placement decisions increasingly involve considerations such as carbon intensity, peak-load management and energy pricing volatility.
Energy-aware design may become a core architectural discipline: moving workloads dynamically to regions with available or greener power, designing AI inference strategies that reduce redundancy and reassessing whether all compute must reside in hyperscale environments.
The cloud’s scalability is no longer governed solely by software. It is governed by thermodynamics.
The Return of Architecture
In response, the binary logic of “cloud versus on-premise” is dissolving.
Enterprises are rediscovering architecture as a strategic function. Hybrid models, edge deployments and private AI stacks are not regressions to the past; they are attempts to rebalance scale with control.
Sergio Gago described this convergence:
“In 2026, the boundaries between cloud and data center will definitively blur. The question is no longer ‘cloud or on-premise’, but how to design a shared, efficient architecture where energy efficiency becomes the primary KPI.”
Sergio Gago, Global Chief Technology Officer, Cloudera
Source: Cloudera Vision Report 2026, “The Era of Convergence”
Architecture is no longer driven solely by cost optimization or deployment speed. It is shaped by sovereignty, sustainability and risk distribution.
Public cloud offers elasticity and rapid scaling. Private or edge deployments offer control, latency advantages and predictable cost structures for stable workloads. The emerging strategy is not “cloud-first”, but “fitness-for-purpose”. Sensitive data determines compute location. Energy availability influences workload distribution. Governance frameworks shape data flows.
Architecture is returning as a discipline of long-term design rather than short-term convenience.
The European Question
For Europe, the neutrality of cloud was always partly illusory. The majority of hyperscale infrastructure remains under non-European corporate and legal control. The first phase of cloud adoption created efficiency gains — but also dependency.
If cloud is no longer neutral, Europe faces a structural choice.
A credible European cloud proposition would require:
- Open ecosystems that minimize vendor lock-in
- Strict compliance with local legal frameworks
- Energy-aware infrastructure design
- Strategic alignment between telecom, grid planning and data centre expansion
- Clear governance over AI workloads and private data assets
Autonomy does not mean isolation. It means architectural agency — the ability to determine where data resides, who can access it and under which energy and legal conditions it operates.
If the cloud is no longer neutral, neutrality cannot be the strategy.
Photo credit:
Altair Media – Conceptual image generated with AI
Caption:
The cloud was once marketed as weightless and borderless. In reality, AI runs on concentrated hardware, cooling systems and national grids. Cloud is no longer neutral — it is infrastructure.
