Advanced Computing in the Age of AI | Wednesday, February 28, 2024

Machine Learning Fuels a Booming HPC Market 

Enterprise infrastructure investments for training machine learning models have grown more than 50 percent annually over the past two years, and are expected to shortly surpass $10 billion, according to a new market forecast.

Intersect360 Research notes that while machine learning has achieved “a very high growth stage,” automation technology covered under the rubric of artificial intelligence has not yet matured into what analysts consider a “market.”

Still, machine learning is adding revenue to the booming HPC market, particularly for training workloads. “Budgets are shifting,” said Intersect360 Research CEO Addison Snell. “This is indicative of additional revenue entering the market and a further commitment to machine learning algorithm development.”

Just over 60 percent of those surveyed said they are running machine learning programs within or in addition to HPC workloads. An additional 10 percent of those polled said they plan to add machine learning programs over the next year.

Those projections are in line with expected growth in the market for hyperscale infrastructure as enterprises seek to apply AI and other automation tools to big data. Intersect360 Research previously estimated the global hyperscale market reached $57 billion in IT spending during 2018. The 30-percent year-on-year spending increase reflects enterprise efforts to ramp up data analytics and machine learning capabilities.

“Hyperscale influences product development, shapes the workforce and fuels major trends like AI,” Snell noted.

So much so, the market analyst continued, that the top eleven vendors accounted for a whopping 82 percent of IT spending in 2018. The market watcher predicts that consolidating hyperscale market will top $100 billion in revenues with the next five years.

Those bullish revenue projections are closely tied to scale, meaning core components like computing, storage and networking “tend to cluster at the high end of the scalability spectrum, often at the extreme high end,” the market tracker noted.

The survey findings confirm earlier analyses of hyperscale trends. For example, Synergy Research Group reported earlier this year that the number of hyperscale data centers has reached more than 430, with another 132 in the pipeline. About 40 percent of major cloud and Internet data center sites are based in the U.S., it said.

Soaring demand for hyperscale datacenters has also fueled an ongoing IT skills shortage. Among the reasons are a data deluge, distributed applications and other real-time capabilities. Hence, infrastructure vendors report growing pressure to complete projects on tighter schedules. Those pressures have exacerbated a skills shortage while raising concerns over datacenter security.

About the author: George Leopold

George Leopold has written about science and technology for more than 30 years, focusing on electronics and aerospace technology. He previously served as executive editor of Electronic Engineering Times. Leopold is the author of "Calculated Risk: The Supersonic Life and Times of Gus Grissom" (Purdue University Press, 2016).