Post-cloud strategy: Architecting the next enterprise stack

The recent history of cloud strategies has been massively impacted by data sovereignty, a move to collapse tech stacks to become more streamlined and far less complex, which also afforded higher levels of security. The cloud-first approach has become more complex.

As we explored in Parts 1 and 2 of this series, that thinking is rapidly evolving. Rising cloud costs, increasing governance challenges, data sovereignty concerns, and the explosive growth of AI are forcing enterprises to rethink how infrastructure should be designed.

Instead of asking whether workloads belong in the cloud, organizations are now asking a far more strategic question: where should each workload run to deliver the best balance of cost, resilience, control, and performance?

Increasingly, the answer is hybrid infrastructure.

Latest Videos From

Watch full video here:

This final article in the three-part series Post-Cloud Strategy: What Comes After Hyperscale? examines how enterprises are designing intentional hybrid and multi-cloud architectures that combine hyperscale platforms, private infrastructure, sovereign cloud environments, and edge computing into a single operational model. Rather than treating hybrid environments as temporary compromises, organizations are now architecting them as long-term foundations for the next generation of enterprise IT.

Hybrid is becoming a deliberate design choice

For many organizations, hybrid infrastructure emerged accidentally rather than strategically. Legacy systems remained on-premises while newer services moved into hyperscale environments, creating fragmented estates that evolved over time without a clear architectural vision.

Now, enterprises are trying to replace that organic complexity. Irin Rahman, CTO at Audiences, says hybrid-by-design strategies require organizations to fundamentally rethink how environments are built and operated.

“‘Hybrid by design’ really means being intentional about how you build and operate across owned and cloud environments, rather than ending up with a mix of systems that evolved organically over time,” Rahman explains. “One of the biggest challenges organizations face now is complexity, particularly as SaaS providers continue to expand their platforms and businesses end up with significant overlap in capabilities across multiple vendors.”

That complexity is becoming one of the defining infrastructure challenges of the post-cloud era. Enterprises are now operating across public cloud platforms, sovereign cloud providers, private infrastructure, and edge environments simultaneously. Without standardization and governance, operational overhead can quickly spiral out of control.

Joe Baguley, CTO, EMEA at Broadcom, tells ITPro that organizations are moving away from simplistic “cloud-first” strategies toward far more selective workload placement decisions.“Balancing workloads across environments starts with a shift from ‘cloud first’ to ‘right workload, right place’,” Baguley says. “Not every application has the same computing requirements, particularly in a world where sovereignty, performance, and cost pressures are increasing simultaneously.”

That philosophy reflects the wider trends discussed throughout the series. Part 1 examined how rising cloud costs and governance failures were forcing enterprises to reassess hyperscale economics, while Part 2 explored how sovereignty and geopolitical concerns are reshaping infrastructure decisions. Together, those pressures are pushing organizations toward more distributed infrastructure models where workloads are positioned according to operational need rather than infrastructure ideology.

Kate Obiidykhata, general manager, cloud-native at Percona, says hybrid-by-design architectures increasingly depend on open technologies that can operate consistently across multiple environments.

“Implementing a hybrid by design approach in practice involves looking at software, components, or services that can run on any platform, rather than being linked to a specific cloud provider or vendor,” she explains. “This normally means a significant role for open source software like databases and components, then using software containers and an orchestration layer like Kubernetes to deploy those elements.”

The objective is not necessarily to move away from hyperscalers altogether, but to avoid becoming overly dependent on any single environment.

Portability is becoming the foundation of modern infrastructure

One of the clearest lessons enterprises have learned from the first wave of cloud adoption is that portability matters far more than many originally anticipated. Early migrations often tightly coupled applications to proprietary cloud services because they offered speed and convenience. However, those decisions also created long-term operational dependency that many organizations are now struggling to unwind.

Baguley says avoiding vendor lock-in requires enterprises to prioritize open standards and operational consistency from the beginning. “If you want to avoid lock-in, you must stratify your architecture,” he says. “Write to open standards like Kubernetes and use declarative configurations like YAML files. That way, you decouple the application from the underlying infrastructure.”

Containers and Kubernetes have therefore become central to modern hybrid architectures. Kubernetes provides a standardized orchestration layer capable of running consistently across hyperscale environments, sovereign cloud platforms, and private infrastructure. In theory, that gives enterprises the flexibility to move workloads without needing to redesign applications for every platform.

However, Baguley emphasised that Kubernetes alone does not eliminate complexity. “Kubernetes is the modern lingua franca of application portability,” he explains. “But Kubernetes alone won’t save you from operational complexity. If every team is spinning up their own flavour of Kubernetes with different networking, security, and storage plugins, you’ve just moved your ‘best-of-breed’ mess up the stack.”

That warning highlights a growing realization across enterprise IT: portability is not simply about technology. It is also about governance and organizational culture. Leo Derikyants, CEO and co-founder of Mind Simulation Lab, says the key to reducing hybrid complexity requires businesses to “separate the infrastructure from the application itself,” Derikyants says. “You package everything the same way and use standard tools to run it. From a developer’s point of view, it shouldn’t matter where the code runs. On AWS or on your own servers.”

This platform engineering approach is becoming increasingly common across large enterprises. Instead of developers managing infrastructure directly, organizations are building internal platforms that abstract away operational complexity while maintaining centralized governance and security controls.

That consistency is becoming essential as infrastructure environments become more fragmented and geographically distributed.

Workload placement is now driven by economics and control

The days of automatically placing every workload into hyperscale cloud environments are fading quickly. Enterprises are becoming far more selective about where workloads should run based on cost, performance, resilience, latency, and regulatory requirements.

Rahman says, “One of the biggest hidden costs and risks in hybrid environments is unnecessary data movement. Businesses should work on bringing compute to the data, rather than continuously exporting and importing huge data files into new platforms. Every movement incurs an added cost, potentially exposes your business to additional tax and opens up the risk of data breach or leakage.”

That issue is becoming even more significant as enterprises deploy AI systems that generate enormous amounts of data. Large AI training models and inference workloads can create substantial operational costs if data constantly moves between clouds and centralized infrastructure environments. As a result, organizations are increasingly bringing AI processing closer to where data already resides.

Derikyants says “data gravity” is becoming one of the defining principles of infrastructure design.“Wherever your data lives, that’s where everything else will end up, because moving large amounts of data is extremely expensive,” he says. “So it usually makes sense to keep heavy data in your own storage and move compute around it when needed.”

That thinking is also driving renewed interest in private infrastructure for predictable workloads. Steady-state AI inference systems, databases, and large-scale compute environments can often become significantly cheaper to operate on enterprise-owned infrastructure over time compared with hyperscale cloud pricing models.

At the same time, hyperscalers still offer enormous advantages for burst capacity and rapid experimentation. Baguley says organizations need structured workload placement frameworks that align applications with the environments best suited to their operational requirements. “Some applications may benefit from hyperscale environments, while others require the control of private or sovereign cloud, especially where jurisdictional authority is a concern,” he explains.

That balanced approach increasingly defines the future enterprise stack. Rather than abandoning hyperscalers, organizations are integrating them into broader infrastructure ecosystems where workloads shift dynamically according to business need.

AI and edge computing will accelerate the hybrid future

The rise of AI and edge computing may ultimately become the biggest driver of hybrid infrastructure adoption over the next decade. Traditional centralized cloud models are not designed to process every workload efficiently, particularly when applications require low latency, localized processing, or continuous real-time decision-making.

Baguley says enterprise IT is moving toward a highly distributed operational model. “The future of enterprise IT is highly distributed, highly democratised, and incredibly flat,” he says. “The concept of a single, centralised ‘cloud’ is dying. We are moving toward a world of billions of endpoints on oil rigs, in retail stores, in cars, and in space via low-Earth-orbit satellites.”

Percona’s Obiidykhata says Kubernetes will become even more important as AI and edge workloads expand.“Kubernetes is the preferred implementation platform for AI workloads too,” she says. “According to Linux Foundation research, 66% of organizations use Kubernetes to host generative AI inference workloads.”

At the same time, automation is becoming essential for resilience. Enterprises increasingly want hybrid environments that automatically recover from failures and scale dynamically without constant human intervention.

Mind Simulation Lab’s Derikyants says automation and GitOps principles are becoming critical to achieving that level of operational resilience. “With a GitOps approach, your whole system is described upfront, and the infrastructure just follows it,” he explains. “If something breaks like your data centre goes down, the system can bring everything back up in another location and shift traffic automatically.”

Ultimately, the post-cloud era is not about abandoning hyperscale infrastructure. It is about becoming more selective and far more intentional about how enterprise environments are designed.

As explored throughout this three-part series, the future enterprise stack will not belong entirely to hyperscalers, sovereign cloud providers, or private data centers alone. Instead, it will consist of carefully orchestrated hybrid ecosystems designed around flexibility, resilience, portability, and operational control.