TL;DR Cloud, on-prem, or both? Enterprise infrastructure is now defined by balancing agility, control, cost, and compliance. Most organisations operate a hybrid approachdriven by the realities of modern workloads, regulatory pressure, and operational maturity. Your architecture should be informed by honest risk assessment, clear ROI, and operational readinessnot trends or vendor claims.
Enterprise reality: the complexity behind “cloud, on-prem, or both?”
Last month, a FTSE 100 CTO told me they’d just finished a six-month ‘cloud-first’ initiative, only to discover their AI workloads performed better on-premises, their compliance team preferred hybrid, and their finance director was questioning cloud costs entirely. Sound familiar?
This is the new normal. Few organisations have the luxury of a single answer. Most are somewhere between cloud and on-premand the best are learning to move with both.
What the data shows: hybrid and multi-cloud are standard
- 89% of enterprises now use multi-cloud as standard, with most running critical workloads in more than one environment (Flexera 2024 State of the Cloud Report).
- Cloud spend wastage remains high: 27–28% of budgets are lost to unused or underoptimised resources (Flexera 2023 and 2024).
- Skills and compliance: A lack of skilled staff—especially platform engineers, SRE specialists, and FinOps practitioners—plus rising demands for data sovereignty and governance, are now the main barriers to effective scaling.
- AI and PaaS adoption are accelerating: Over 75% of organisations experiment with or run AI/ML in production, fundamentally changing workload placement, performance, and cost considerations.
- Security and FinOps: Security and cloud cost management now top the agenda, with over half of enterprises operating dedicated FinOps teams to enforce financial discipline and transparency.
This is not just a technology problem; it is an operational maturity challenge that requires systematic measurement and relentless optimisation.
The Kubernetes paradigm shift: hybrid as the new operating model
Kubernetes has redefined how enterprises approach hybrid and multi-cloud. What once required deep provider lock-in can now be run anywhere: public cloud, data centre, or edge. In 2025, 40% of organisations now run all production workloads on Kubernetes (KubeCon London 2025 Survey).
- Local cloud stacks (Rancher, OpenShift, VMware Tanzu) offer a private cloud experience on-premises, with integrated CI/CD, self-service, and observability.
- Sovereignty and locality: On-prem and sovereign cloud models support compliance with regulations like GDPR, while maintaining developer experience.
- Operational consistency: Workflows remain consistent, whether cloud or on-prem, so teams can focus on outcomes, not environments.
- New skill requirements: Running Kubernetes at scale is complex. Platform engineering, SRE, and DevOps are now core disciplines.
Kubernetes is not just a technical upgrade; it signals an organisational shift towards hybrid-first, platform-driven infrastructure. According to the CNCF, Kubernetes is driving containerisation of mission-critical workloads, increasing developer productivity, and reshaping operational models for IT. For more on future trends, see the detailed analysis in Kubernetes in 2025: Top 5 Trends and Predictions.
Hybrid in practice: enterprise case studies and measurable results
Santander adopted a hybrid model with Red Hat OpenShift, retaining regulatory workloads on-prem while scaling customer apps in the cloud (Red Hat OpenShift success story). In Spain, their data lake programme delivered 10x faster time-to-market and 20x cost reduction (Santander Spain Data Lake).
Deployment frequency and risk: High-performing teams leveraging hybrid and cloud-native tools now deploy code multiple times daily, with up to 200% more deployments and markedly reduced failure rates (DORA 2023).
Other enterprise cases: Over 80% of large UK banks now blend on-prem and cloud, balancing data sovereignty with operational flexibility.
Decision framework: step-by-step for engineering leaders
Adopting “cloud, on-prem, or both” is not a one-off. Leaders must continually re-evaluate architecture, team capability, and business drivers. Use this framework for structured migration:
- Assess workloads: Map every workload by sensitivity, compliance, performance, and cost profile
- Build a cross-functional platform team: Platform engineering, SRE, security, and FinOps, all at the table
- Pilot migration: Select a non-critical workload and implement with progressive delivery, monitoring DORA and SPACE metrics
- Expand and optimise: Roll out hybrid patterns iteratively, using continuous feedback to adjust placement and process
- Set time-bound milestones: Review quarterly, using clear metrics for deployment frequency, failure rate, lead time, and developer experience
- Model TCO over 3–5 years: Include infrastructure, support, migrations, skills, and opportunity cost (Cloudian Hybrid Cloud Strategy)
Emerging patterns: AI, sovereignty, skills, and platform engineering
- AI and data gravity: AI/ML workloads increasingly remain close to their data, shifting training jobs on-prem or into sovereign clouds for cost, privacy, and performance. Organisations must regularly review whether cloud placement or repatriation is best for each class of workload.
- Data sovereignty: GDPR and similar regulations drive demand for localised infrastructure, multi-region compliance, and sometimes outright cloud repatriation.
- Platform engineering: Now seen as an enterprise core discipline, platform engineering teams design and operate hybrid platforms, create standardised developer experiences, and enforce cross-cloud observability and security (Gartner report, Platform engineering guide).
- Skills and organisational change: The greatest barriers are rarely technical. The most urgent hiring needs for successful hybrid transformation are experienced platform engineers, SRE specialists, and FinOps practitioners. Upskilling and ongoing learning are now the default, not the exception.
Building adaptive infrastructure strategies: what matters next
There is no universal architecture for the enterprise; only disciplined, measured adaptation. Infrastructure strategy is never “set and forget”. Leaders who revisit architectural choices, track ROI, and invest in both skills and platforms will deliver on the promise of “cloud, on-prem, or both” and future-proof their operations for what comes next.
Immediate next action: Start by auditing your current workload placement against these emerging patterns. Most organisations discover immediate optimisation opportunities, from right-sizing AI workloads to plugging skills gaps or improving compliance.
For further reading and strategy:
Leave a Reply