After years of focusing on software development and investments, the technology industry has rediscovered the value of semiconductors as it recognizes the critical role AI chips are playing in the burgeoning market for artificial intelligence solutions.
Artificial intelligence hardware—specifically AI chips like graphics processing units (GPUs) and application-specific integrated circuits (ASICs)—are essential to processing any AI task, including learning and inference. These chips are accounting for as much as half the profit generated by AI technology stack, according to a report issued by McKinsey & Company.
“With the growth of AI, hardware is fashionable again, after years in which software drew the most corporate and investor interest,” McKinsey stated. “Our discussions with end users suggest that interest will be strong for both cloud and edge solutions, depending on the use case. Cloud will continue to be the favored option for many applications, given its scale advantage. Within cloud hardware, customers and suppliers vary in their preference for ASIC technology over GPUs, and the market is likely to remain fragmented.”
AI chips will command this large share of market value because they must be customized for specific artificial-intelligence tasks. This requirement for customization will prevent the commoditization of AI chips, bolstering their price tags and helping them retain their value.
“While hardware has become commoditized in many other sectors, this trend won’t reach AI any time soon because hardware optimized to solve each microvertical’s problems will provide higher performance, when total cost of ownership is considered, than commodity hardware, such as general-purpose central processing units (CPUs),” McKinsey added.
For example, accelerator microchips that are designed to run convolutional neural networks are most suitable for image-recognition applications, and thus would be the best type of semiconductors for makers of medical-imaging devices. On the other hand, sellers of sophisticated virtual home assistants might prefer accelerator chips optimized for long short-term memory networks, which are more tailored for speech-recognition and language-translation applications, McKinsey stated.
McKinsey predicts that much of the monetary value of the AI market will reside in “microverticals,” i.e., specific use cases in certain industries. Companies addressing these microvertical markets must offer end-to-end solutions that encompass the entire AI technology stack.
McKinsey divides the AI technology stack into layers that include services, training, platform, interface, and hardware. With hardware accounting for about 50 percent of the stack’s profits, the remainder is represented by services and training.
Stephan Cunningham is vice president, product management at Veritone. Working in concert with core internal teams including industry-specific general managers and engineering as well as directly with clients and prospects, he leads the disciplines and business processes which govern the Veritone aiWARE platform.