The Rise of AI-as-a-Service and Its Impact on Data Centers
By: Kayla Matthews on October 7, 2019
More companies than ever want to bring AI into their workflows and decision-making processes. Thirty-seven percent of all organizations have already implemented some form of AI — a 270 percent increase over the last four years.
However, these companies don't always have the skilled workforce or computer resources and infrastructure needed to bring these AI solutions on-site.
Instead, companies are turning to cloud-based AI-as-a-Service solutions to meet their needs. But not all data centers — who host these solutions — are prepared for the massive uptick in processing power and data storage that AI demands. This shift in demand may mean big changes for the data center industry.
AI and Big Data Come to the Cloud
AI-as-a-Service is similar to Software-as-a-Service (or AIaaS and Saas, respectively). With SaaS, companies build a tech solution — for example, a project-planning utility — and then license these solutions to other companies. Unlike traditional software, the program is either partially or completely web-based, rather than located entirely on a user's computer.
With AIaaS, tech companies create some AI-based solution — like a big data analysis tool or AI-powered antivirus — and license it out. Some element of the solution — a processing power-hungry algorithm or big data set — will be stored on the cloud, because the average office desktop just doesn't have the computing resources to support it.
The AI developers don't usually have the processing power or storage needed to support all of their clients. Instead, they turn to data centers.
The AIaaS style of tech offering is rapidly becoming popular. Amazon Web Services, which accounts for more than half of Amazon's profits, reported a 250 percent increase in the use of AI tools during 2017. As AI companies become a larger sector of the market and the tech behind AI becomes more refined, this trend will probably continue.
As companies shift away from on-site physical storage to cloud-based data storage and utility computing, data centers will need to be ready to meet the demand for increased storage and computing power needs. Faster methods of storage — like all-flash storage — and increased processing power will likely go from being nice-to-have options to AI necessities. Data centers will also need more and more space to store the massive data sets that companies draw from when implementing big data analysis.
As for data centers that don't prepare for AIaaS? More than 30 percent may be no longer viable by 2020, according to IT advisory firm Gartner.
The Future for Data Scientists
Some fear that this turn to AI will displace data scientists and that companies looking to save money on skilled workers will turn to fully-automated solutions. In most industries, however, it's become clear that current AI tech can't replace skilled workers. Grim statistics are occasionally reported, but AI hasn't made major waves in terms of employment.
Ninety-four percent of data science graduates have found work since 2011. At the moment, there are no real indicators that AI will change that. A field like data science, which relies primarily on analysis and communication, can't reasonably be automated. Instead, AI will be a kind of tool used to make data scientists more efficient.
Even in manufacturing — the field considered most susceptible to automation job loss — employers are turning away from fully-automated AI technology. Instead, factories are investing in collaborative robots that work alongside people, using AI to provide assistance and useful data. The dream of an automated factory is always a few years away, but AI and robotics tech never seem to improve enough to get us any closer. The same can likely be said about automated offices and data centers.
How AI-as-a-Service will Impact Data Centers
Data centers will need to adapt to a much higher level of demand than they've seen in the past. Companies want AI-based solutions to their biggest challenges, but not every business has the computing power and data storage needed to perform tasks like big data analysis on-site. Data centers and cloud computing will be called upon to bridge that gap.
AI will probably be most disruptive at the management and policy level, where big data insights can drive changes in decision-making processes. At the employee level, AI will most likely be seen in the form of AI-powered tools and utilities. Data scientists may need to adapt to a new kind of data science, but their jobs are probably safe.
About Kayla Matthews
Kayla Matthews writes about data centers and big data for several industry publications, including The Data Center Journal, Data Center Frontier and insideBIGDATA. To read more posts from Kayla, you can follower her personal tech blog at ProductivityBytes.com.