EXECUTIVE SUMMARY
For its hearing on “China’s Pursuit of Next Frontier Tech,” the U.S.-China Economic & Security Review Commission (USCC) is seeking testimony on the
current and potential future state of supercomputing innovation worldwide, with an emphasis on China’s position on the global stage relative to the
U.S. Addison Snell, CEO of Intersect360 Research, provides this written testimony in answer to USCC’s questions for the hearing. Mr. Snell will also
provide oral testimony to answer additional questions at the hearing on March 16, 2017. Information about Mr. Snell, Intersect360 Research, and the
questions asked are in the Appendices of this report. A transcript of the oral testimony will be made available at the USCC’s website, www.uscc.gov.
In this statement, we give an overview of the high performance computing (HPC) industry, including analysis of hardware, software, and industry trends.
Where relevant, market data from Intersect360 Research is included, particularly for the analysis of significant HPC market segmentations. In the next
section, we give a country-level analysis of national supercomputing strategies and programs, for the U.S., China, and other significant countries. In
the closing section we give our analysis, conclusions, and recommendations.
While the U.S. still leads by far in the most straightforward market share metrics of production (vendors, supply-side) and consumption (buyers, demand-side),
industry indicators show the U.S. is falling behind in the leading edge of advancement. Chinese leadership has apparently recognized the relationship between
HPC and economic growth and has set forth on a program to drive the country into a leadership position. The best response to this new challenge is to continue
if not increase national support for HPC at all levels.
National supercomputing efforts are essential to motivating investment at the high end. From that point, U.S. companies excel at seizing opportunities
to drive markets forward. Against these strengths, the top limitations to Exascale deployments are software and skills. If we do build a system, how will
we use it? A key feature of the Exascale Computing Program is its emphasis on co-design, finding end-user stakeholders to collaborate on the design of
next-generation supercomputing technologies, bolstered by government funding. We recommend:
· National initiatives in low-level software tools and programming models, together with stakeholders in industry and academia.
· Government-funded partnerships between industry and academia.
· Ongoing pursuit of next-generation technologies.
Regardless of these recommendations, the HPC market will continue, powering new innovations and ideas around the world. Supercomputers today are close
to a million times more powerful now than they were 20 years ago. In another 20 years, they could be a million times more powerful still. The leaders in
supercomputing will be the ones that do not rest on their achievements, but rather continue to chase the next challenge over each new horizon.