Recognising that not all workloads are best suited for the cloud, businesses have shifted from a “cloud-first” to a “cloud-smart” mindset.
This is because on-premises infrastructure remains essential, offering cost efficiency and consistent performance while addressing increasing regulatory demands around data sovereignty.
Many organisations are therefore expanding their on-premises footprint while also investing in the cloud where it makes sense. According to research from IDC, in Q3, 2024, 88% of cloud buyers were deploying a hybrid cloud or were in the process of operating one.[1]
Mitigating virtualisation challenges
Virtualisation plays a critical role in enabling hybrid cloud strategies. It provides the abstraction layer that allows businesses to manage, move, and optimise workloads across on-premises and cloud environments.
By providing a consistent management framework and visibility across disparate systems, virtualisation gives IT teams the flexibility, control, and efficiency once reserved for hyperscalers.
Many businesses are drawn to virtualisation platforms that offer deep integration and robust feature sets.
However, this high level of integration—while delivering significant technical benefits—can create challenges when organisations wish to change course.
“The more embedded a solution becomes within IT infrastructure, the harder it is to pivot without incurring significant cost or disruption,” explains Nicola Tan, Director of Market Development for Enterprise Software at AMD.
This challenge is compounded when businesses use bundled packages that include advanced features, such as networking or storage virtualisation. In such cases, businesses may find themselves paying for capabilities they do not need, reducing the overall cost-efficiency.
“Another critical factor is the licensing model,” adds Tan. “A move from socket-based to core-based licensing can alter the economics of server procurement. Under socket-based models, organisations can maximise value by deploying high-core-count processors under a single licence. Core-based pricing, however, means that every core contributes to the overall cost, encouraging businesses to optimise around achieving the greatest virtual machine density with the fewest number of cores.”
Optimising licensing through hardware strategy
This shift has pushed many IT teams to reassess their compute strategies. Concurrently, companies like AMD are helping businesses to optimise their hardware and bring down the total cost of their virtualised hybrid environment.
AMD helps businesses tackle rising virtualisation prices by delivering leadership in three ways. First, AMD EPYC™ CPUs can enable more virtual machines per core, reducing the number of cores—and therefore core-based licence costs—required.
Its latest cores can offer up to 2.5x the performance of the previous generation of CPUs that are currently nearing the end of their life. AMD has estimated this can enable enterprises to reduce their virtualisation licence cost by up to 61%.[2]
Second, greater virtual machine density provided by AMD cores means that fewer servers can be required overall to do the same work as a legacy system, helping reduce the data centre footprint.[3]
Finally, the power-efficient AMD design can reduce energy use and simplify management, and can help reduce operational costs such as power, cooling, and real estate.
Tan concludes: “Enterprises should stop thinking of hardware and software as separate cost centres. By selecting faster hardware, you can dramatically reduce virtualisation costs. The key to success is aligning your server strategy with your software licensing model.”
[1] IDC, “Ten Trends That Shaped the Cloud Market in 2024,” February 2025
[2] This scenario contains many assumptions and estimates and, while based on AMD internal research and best approximations, should be considered an example for information purposes only, and not used as a basis for decision making over actual testing. The Server & Greenhouse Gas Emissions TCO (total cost of ownership) Estimator Tool compares the selected AMD EPYC™ and Intel® Xeon® CPU based server solutions required to deliver a Target Performance Metric of 28700 units of integer performance as of April 10, 2025. This estimation reflects a 5 year time frame. Only power costs and software license costs contribute to OPEX. This analysis compares a 2P AMD 32 core EPYC_9375F powered server with a SPECrate2017_int_base score of 1010 (https://spec.org/cpu2017/results/res2024q4/cpu2017-20241105-45389.pdf) compared to a 2P Intel Xeon 32 core Platinum_8562Y+ based server with a SPECrate2017_int_base score of 729 (https://spec.org/cpu2017/results/res2024q2/cpu2017-20240530-43623.pdf) versus legacy 2P Intel Xeon 24 core Gold_6252 based server with a SPECrate2017_int_base score of 287 (https://spec.org/cpu2017/results/res2019q4/cpu2017-20190916-18249.pdf). Calculator. For additional details, see https://www.amd.com/en/legal/claims/epyc.html#q=9xx5TCO-011. (9xx5TCO-011)
[3]AMD, 9xx5TCO-011. For that scenario (28700 SPECrate2017_int_base score), you need 1865 AMD 9375F cores or 4800 Intel 6252 cores. 4800/1865 = 2.6