The growing demand for high performance computing (HPC) is creating an emerging battleground between cloud services providers and other IT infrastructure providers, with both sides of the market vying for market share.
This is according to data and analytics firm GlobalData, which argues that a confluence of developments, including a surge in data generated by internet of things (IoT) sensors and the emergence of new data-rich applications, some of which use artificial intelligence (AI), will drive demand for higher performance computing solutions over the next five years.
“Traditional HPC use cases including financial services, scientific and medical research, and natural resource exploration and extraction remain as important as ever,” said Chris Drake, principal analyst at GlobalData. “However, they are being joined by new applications from an expanding range of industries, including those that leverage AI and large volumes of data.”
Such demand is broadly a good thing for the market, right across the channel ecosystem. However, it remains to be seen precisely how industry will choose to consume HPC services as it becomes more popular.
HPC services can be consumed in a number of ways, some of them delivered by hardware infrastructure employed directly by users themselves and others provided as-a-service by cloud vendors.
Certainly, it is expected that the surge in HPC demand will contribute to an increasingly vibrant and competitive market for AI servers and HPC infrastructure, including supercomputing tech.
At the same time, cloud service providers are increasingly bulking up their as-a-service offerings with cloud-delivered HPC services.
As such, both IT infrastructure vendors and cloud service providers are upbeat about future enterprise demand for new, more powerful computer processing capabilities, despite market disruptions resulting from the ongoing pandemic.
This upbeat outlook is reflected in the development and enhancement of server portfolios, which are being beefed-up to include the latest processor chips and hardware accelerators, to ensure maximum support for the most demanding, compute-intensive workloads, according to GlobalData.
It is anticipated that, alongside the emergence of new, compute-intensive applications, the industry will see a growing focus on more flexible ways of consuming IT solutions, including the use of as-a-service consumption models.
Indeed, GlobalData expects flexible consumption models to become an important means by which server vendors will drive future adoption of AI and HPC servers, extend their research into new market segments and compete against the hyperscale cloud providers.
“The hyperscalers have been strengthening their own ability to offer HPC and supercomputing as a service and, in June 2021, Microsoft announced the general availability of its new cloud supercomputing offering for AI and HPC workloads, powered by NVIDIA A100 Tensor Core GPUs,” Drake said. “Microsoft Azure joins Amazon Web Services, Google Cloud and Oracle, which all now offer NVIDIA A100-based cloud services.
“Growing demand for more powerful processors is also driving competition among chip manufacturers such as Intel, AMD and NVIDIA. NVIDIA is making a multi-pronged attack to become a major player in the enterprise IT market, starting at the very high end,” he added.