EXECUTIVE SUMMARY
This Intersect360 Research report presents the 2023 market for on-premises (non-hyperscale) servers used for High Performance Computing (HPC) and artificial intelligence (AI) and constituent server vendor revenue shares, with comparison to 2022. “On-premises” in this context includes co-located systems; the report spans all HPC-AI systems outside of hyperscale.
This report tracks revenue shares for Dell, Eviden (formerly Atos), Fujitsu, HPE, IBM, Inspur, Lenovo, Nvidia, Penguin Solutions, and Supermicro. Additionally, the report discusses Huawei, NEC, and Sugon, without reporting exact shares.
In general, the HPC-AI market includes multi-node, networked systems or cloud instances running parallel applications, which require a focus on performance or scalability in some dimension (e.g., processing, memory, I/O, networking), plus associated storage, software, networks, services, and other necessary components. The market size does not include spending on internal budget items such as power consumption or staffing; these are tracked in separate surveys. This methodology also does not include single-node desktops or workstations, embedded computing, or other edge devices.
A major component of the HPC-AI market is AI-focused infrastructure purchased by hyperscale organizations. Hyperscale companies are those with internet-driven business models, spending hundreds of millions to billions of dollars per year in total IT infrastructure. Amazon, Apple, Google, Meta, and Microsoft are the largest examples, but the hyperscale segment also includes non-U.S. companies such as Alibaba, ByteDance, Naver, and Tencent, as well as new AI-focused cloud services such as CoreWeave, Denvr Dataworks, and Lambda. The hyperscale segment is excluded from the analysis of the on-premises HPC-AI server market that is the focus of this report. The hyperscale market has its own vendor dynamics that are discussed qualitatively.
Intersect360 Research reports available in this series include the following segmentations of the non-hyperscale portion of the market:
- Products and Services: servers, storage, networks, software, service, cloud, other
- Vertical markets: academia, national security, national research labs, national agencies, state or local governments, bio sciences, chemical engineering, consumer product manufacturing, electronics, energy, financial services, large product manufacturing, media and entertainment, retail, transportation, other (presented both with and without hyperscale as a vertical market)
- Regions: Americas, EMEA, Asia-Pacific
- Server class (HPC-AI server revenue): entry-level, midrange, high-end, supercomputer
- HPC-AI server vendor market shares (total on-premises HPC-AI, current year only, not forecast)
Non-hyperscale HPC-AI spending, including the use of cloud computing for HPC-AI, increased 9.0% to $48.4 billion in 2023. The largest component of this is on-premises HPC-AI servers (including co-located systems). Fueled by more expensive configurations that incorporate GPUs for AI workloads, spending on on-premises HPC-AI servers grew by 11.6% year-over-year, reaching $14.1 billion in 2023.
The number one vendor for on-premises HPC-AI systems continued to be HPE in 2023, followed by Dell; these companies have held the top two spots every year since 2014. However, this seeming stability belies major movements in the market. Nvidia has exploded to become a top-tier vendor for HPC-AI systems. If hyperscale sales were included, Nvidia would already be the top vendor. Including only on-premises, non-hyperscale systems, Nvidia is third, having nearly tripled in revenue year-over-year. Supermicro, another vendor that has benefited tremendously from hyperscale sales, had another big growth year and is fourth.
TABLE OF CONTENTS
EXECUTIVE SUMMARY 2
TABLE OF CONTENTS 4
INTRODUCTION 6
What Are HPC, AI, and Hyperscale? 6
THE HPC-AI INFRASTRUCTURE MARKET: ON-PREMISES VS. HYPERSCALE 8
Combined HPC-AI Market View 8
Table 1: HPC-AI Infrastructure, On-Premises vs. Hyperscale ($M), 2023 vs. 2022 9
Figure 1: HPC-AI Infrastructure, On-Premises vs. Hyperscale ($M), 2023 vs. 2022 10
Figure 2: HPC-AI Infrastructure as Proportion of Total On-Premises and Hyperscale Infrastructure ($M) 11
NON-HYPERSCALE HPC-AI SERVERS: VENDOR MARKET SHARES 12
Non-Hyperscale HPC-AI Server Classes, 2023 vs. 2022 12
Table 2: HPC-AI Server Market, Excluding Hyperscale ($M), 2023 vs. 2022, by Vendor 13
Figure 3: HPC-AI Server Market, Excluding Hyperscale ($M), 2023 vs. 2022, by Vendor 13
Figure 4: HPC-AI Server Market, Excluding Hyperscale ($M), 2023, by Vendor 14
HPC-AI System Vendor Discussion 15
CONCLUSIONS 20
APPENDIX A: METHODOLOGY 23
APPENDIX B: HPC, AI, AND HYPERSCALE: DEFINITIONS AND INTERDYNAMICS 25
What Are Artificial Intelligence and Machine Learning? 25
What Is HPC? 27
What Is Hyperscale? 27
What Makes Hyperscale “Hyper”? 28
Hyperscale and Business Computing 29
Hyperscale and HPC 29
How Hyperscale is Unique 31
APPENDIX C: HPC MARKET DYNAMICS MODEL AND FUNDAMENTAL FORECAST ASSUMPTIONS 33
Market Maturity 33
Fundamental Market Dynamics Model 34
Figure C1: Traditional HPC Market Dynamics 35
Fundamental Market Assumptions 36
Fundamental Drivers 36
Fundamental Market Dampeners 37
Model-Based Assumptions 38
Basic Market Drivers 38
Basic Market Dampeners 39