Joe Baguley, CTO EMEA, Broadcom, outlines why cloud strategy, private/hybrid architecture, and continuous optimisation are becoming the make-or-break layer for compliant AI at scale.
Europe has declared its AI ambitions through the AI Act and accelerated cloud adoption, but ambition alone will not deliver impact. That responsibility now sits with the infrastructure underpinning AI workloads.
For those designing, operating and investing in AI systems, the priority is clear: cloud platforms, data centres and the wider digital backbone must be built to scale securely, operate reliably and meet stringent regulatory expectations. Without this foundation, Europe’s most sophisticated AI initiatives risk faltering before delivering value. The stakes go beyond any single organisation; they concern Europe’s credibility as a global technology leader.
How infrastructure shapes AI outcomes
Businesses are investing heavily in generative AI, automation and AI-driven decision-making, expecting transformative results – from operational efficiency to new services. The reality is that infrastructure underpins everything in AI deployment. Algorithms or data alone are not enough. AI workloads demand compute capacity, seamless data access, and robust compliance controls, all while managing costs effectively. Without an effective cloud foundation, the way systems are built, maintained, and optimised will determine whether these investments succeed or become another silo.
It will also affect the EU’s ability to achieve its strategic objectives and support the growth of AI across Europe. The stakes are high: 48% of EMEA IT leaders report wasting at least 25% of their cloud spend, and 90% prioritise cost predictability.
Infrastructure can either accelerate AI adoption or create bottlenecks, leaving organisations grappling with underutilised investments, performance issues, spiralling costs and serious questions about regulatory compliance and sovereignty. In fact, 51% of global organisations are moving workloads back to private cloud over security or compliance concerns, underscoring the importance of robust, well-governed infrastructure in realising AI’s potential.
Building for scale and resilience
AI workloads are dynamic, evolving with data and demand. Infrastructure must be equally agile – scaling flexibly to avoid constraints and ensuring rapid, secure data access. A system slowed by inefficient storage or fragmented data environments directly impacts the speed and reliability of AI insights.
Operational readiness extends beyond technical performance. It requires resilience, security, and the ability to handle demand surges. Organisations that prioritise these capabilities can maximise the value and reach of their AI initiatives, turning infrastructure from a constraint into a business enabler.
Resilience is not just an operational consideration but also a regulatory requirement. EU legislation for financial institutions, such as the Digital Operational Resilience Act (DORA), mandates resilience in every aspect of the financial services information technology infrastructure, with an emphasis on functions supporting critical services. The scalability of any AI application for the financial industry will need to factor not only the likelihood that it will support a critical service within the meaning of DORA, but also the regulatory and compliance consequences that emerge from that determination.
Implementing scalable AI strategies
For IT leaders, the question is no longer whether to invest in AI infrastructure but how to do so in a way that supports scale, cost control and resilience. With 93% of organisations favouring private cloud for critical applications due to its financial visibility and predictability, there is a clear shift towards solutions that combine flexibility with strong governance.
Private and hybrid cloud strategies can offer the agility needed for high-demand AI workloads while meeting regulatory and sovereignty requirements, and are often considered alongside hyperscaler models where cost, control and compliance requirements vary by workload.
Focusing on scalable infrastructure helps ensure AI initiatives can grow without limitations. Here is a practical approach to assess and align infrastructure for AI adoption:
- Assess and align infrastructure
For organisations looking to adopt AI more widely, the first step is to assess current infrastructure against projected AI workloads, identifying gaps in compute capacity, data accessibility and cost management. Building or expanding infrastructure with a focus on scalability helps ensure AI initiatives can grow without hitting bottlenecks. - Prioritise data integration and compliance
AI thrives on data, yet fragmented or siloed information can hinder both performance and compliance. Ensuring seamless data integration, secure access and audit-ready pipelines is fundamental. Leaders should prioritise architectures that support interoperability, secure storage and high-speed processing, enabling AI models to deliver actionable insights rapidly and reliably. Leaders should also assess how they plan to apply the technology to comply with applicable rules and sector standards.
Use cases captured by the EU AI Act are likely to require specific controls and governance linked with the data and the algorithms as they flow through the infrastructure. Requirements such as DORA and NIS2, which are linked to sectors, are likely to prioritise organisational and technical controls on the infrastructure, the supply chain and the supply of data. Sovereignty will remain a political priority, especially for public sector or critical infrastructure customers. Therefore, the ability to demonstrate independence from foreign interference in operating an AI infrastructure may become a key consideration in public procurement. - Continuous progress
AI infrastructure is not a set-and-forget investment. It requires ongoing tuning, testing and optimisation to remain aligned with evolving workloads and regulatory expectations. By adopting a proactive, forward-looking approach, enterprises can keep AI deployments effective and compliant.
Europe’s changing AI landscape
The need for continuous optimisation goes hand in hand with navigating a fast-evolving regulatory landscape that is redefining how AI is developed and deployed. It also affects the rules and obligations associated with specific applications or sector verticals. For European organisations, these pressures are particularly pronounced.
The EU AI Act is a landmark piece of legislation that aims to create a consistent set of rules for AI use across member states. Its influence is already shaping enterprise priorities, while political initiatives aiming to promote cloud and AI utilisation are underway.
In this complex environment, compliance is now a strategic imperative that can determine the success of one’s efforts. Businesses should ensure their infrastructure embeds governance, risk management, and transparency to meet regulatory demands and foster trust with customers, investors, and regulators.
Deploying AI in a non-compliant manner, either because of infrastructure choices or a lack of effective controls, risks reputational harm as well as financial penalties and legal action. By integrating compliance into infrastructure design, organisations can turn regulatory requirements into an operational discipline that supports trustworthy, ethical AI.
Ensuring Europe leads in AI
Europe’s AI future will not be won through superior algorithms or access to data alone, but by infrastructure readiness. With forward-looking regulation, Europe has a genuine chance to secure a leading role in the global AI landscape, but this outcome is not guaranteed. Enterprises must elevate infrastructure from a technical consideration to a strategic priority, ensuring operational resilience and scalability match regulatory ambition. Europe has set out its vision clearly; whether it becomes reality will depend on the foundations being laid today.
