HPC industry analyst firm Hyperion Research has updated its market sizing for the HPC industry to include “AI-centric servers” from non-traditional HPC suppliers, such as Nvidia, Cerebras, SambaNova and SuperMicro, which has increased the overall HPC market by 36.7 percent in 2023 and is projected to add $13.6 billion by 2028, Hyperion said.
“As AI is being integrated into traditional HPC workflows, and HPC-class machines are being purchased to adopt AI into enterprise data center applications, we are seeing a great increase in new buyers and new suppliers that have entered the HPC market,” said Earl Joseph, CEO at Hyperion Research.
The changes to HPC market sizing and tracking are based on multiple Hyperion surveys conducted over the last two years and represent a major adjustment to the overall HPC market, heavily driven by the use of AI, Large Language Models, and purchases of large GPU/accelerator-based systems. HPC systems are used as the backbone for many AI and AI-related applications, which has driven the overall HPC market to a new level of growth, the firm said.
Market Segment Definitions
AI-centric Servers (new revenues added to the previous HPC market sizing)
These are on-premises AI-centric HPC servers that are provided by non-traditional HPC suppliers, such as NVIDIA, Cerebras, SambaNova, SuperMicro, etc., frequently at non-traditional HPC user sites, such as large enterprise sites adding AI capabilities.
- These servers are designed primarily to run AI and AI-related workloads.
- These servers are a subsegment of the overall HPC market but haven’t historically been accounted for within prior HPC market numbers.
HPC & Advanced AI Servers (previously counted in the HPC market sizing)
These are on-premises servers that are used for highly computational or data-intensive tasks:
- Hyperion Research uses the terms technical computing and high performance computing (HPC) to encompass the entire market for computer servers used by scientists, engineers, analysts, and other groups using computationally and/or data-intensive modeling and simulation applications. An on-premises system primarily used for HPC workloads (at least 50 percent) can be referred to as an HPC/AI system, or simply an HPC. These servers also include AI-centric servers used by traditional HPC end-users and are typically sold by traditional server vendors.
- In addition to scientific and engineering applications, technical computing includes related markets/applications areas such as economic analysis, financial analysis, animation, server-based gaming, digital content creation and management, business intelligence modeling, and homeland security database applications.
- Systems acquired by cloud service providers for the purpose of hosting cloud workloads are excluded, as Hyperion Research separately tracks spending for HPC usage in cloud environments.
- The Advanced AI servers previously tracked are primarily sold by traditional HPC system vendors going to traditional HPC sites.
Speak Your Mind