Why Enterprises Can’t Afford To Ignore Cloud Optimisation In 2025

Cloud infrastructure has evolved far beyond its early use cases. Enterprises now need to scale on demand, maintain performance under unpredictable workloads and support increasingly specialised computing requirements driven by Ai, data analytics and high-performance computing.

These demands have made cloud optimisation a critical priority, not just for operational performance, but also for cost control and competitive resilience.

According to Flexera, nearly half of all enterprise workloads are now hosted in the public cloud. In parallel, 72 per cent of IT decision-makers are prioritising optimisation efforts to reduce costs and increase efficiency.

Infrastructure Needs Are Becoming More Specific

Traditional general-purpose computing is no longer enough to meet the demands of today’s cloud environments. As Ai and HPC workloads become more widespread, many legacy cloud infrastructures struggle to deliver the required performance and scalability.

This issue is compounded in multi-cloud environments where agility, performance and security all need to be tightly balanced. The growing intensity of compute workloads is driving the need for cloud-native architectures that offer more than just elastic provisioning.

Without purpose-built infrastructure, running Ai workloads at scale can also lead to unnecessary energy use, inflating costs and impacting sustainability targets. These pressures are coming from both internal ESG policies and external stakeholders with increasing expectations for green accountability.

Security As A Core Priority

Beyond compute performance, security has become a key differentiator in cloud infrastructure decisions. Cloud-optimised hardware often includes features like confidential computing that encrypts data while in use, reducing the risk of unauthorised access from both physical and virtual threats.

As cyber threats escalate, organisations must treat cloud security as essential. This includes protecting virtualised environments from zero-day exploits and isolating sensitive processes to mitigate the risk of attacks on memory or kernel-level processes.

Leaving cloud environments unprotected is no longer a viable option. Threats from both individual hackers and state-sponsored actors require infrastructure that includes built-in protections, not just software-based solutions added later.

The 2025 Cloud Optimisation Playbook

To meet the demands of modern workloads, IT leaders should align infrastructure selection with four core principles:

• Performance: Cloud instances must be able to handle everything from lightweight web applications to memory-intensive analytics and transactional systems. The infrastructure should match workload profiles rather than applying a one-size-fits-all model.
  
  
• Cost And Efficiency: Cloud instances with higher compute density allow more virtual machines or containers per server, reducing the physical footprint and power consumption. This enables organisations to meet workload demands while lowering operational costs.
  
  
• Security: Confidential computing and memory encryption are now critical. Enterprises should ensure their cloud environments offer built-in protections for data in use, not just at rest or in transit.
  
  
• Ecosystem Compatibility: Using x86-based architectures makes it easier for developers to manage workloads across platforms, maintain application stability and migrate without service interruptions.

Modernisation With Measurable Impact

Selecting the right cloud infrastructure involves weighing the cost per unit of work. Traditional infrastructure models often led to overprovisioning or under provisioning, both of which add risk and cost.

Cloud-optimised hardware shifts this equation, enabling more efficient use of resources. For example, Oracle Cloud and Google Cloud have announced services powered by 5th Gen AMD EPYC processors.

Oracle’s latest E6 standard bare metal instances reportedly deliver up to 33 per cent more compute and memory and double the network bandwidth compared to E5, offering significant improvements in performance. Google Cloud’s C4D virtual machines claim up to 80 per cent higher throughput per vCPU, driven by AMD’s Zen 5 architecture.

These improvements not only support scalability, but also enhance the sustainability of enterprise cloud operations by lowering energy usage per workload.

Optimisation Is A Continuous Process

Cloud optimisation is not a one-off exercise. As demands shift, workloads evolve and technologies advance, organisations must continuously revisit their infrastructure strategies.

By modernising cloud environments, businesses can build a more secure, scalable and efficient digital foundation – critical for supporting future innovation and maintaining a competitive position in 2025 and beyond.

Peter Chambers is Managing Director, APAC at AMD.

Last Updated on June 6, 2025 by Peter Chambers