Companies blindsided by the high cost of cloud computing are reexamining its use to attain a higher ROI.
For more than a decade, enterprises have run workloads on cloud providers, starting with AWS, and later with Microsoft Azure and Google Cloud. The initial attraction was usage-based pricing for cloud computing, which businesses expected to be a better deal than buying and managing physical application servers in the data center.
The rush to the cloud that drove double-digit revenue growth for cloud providers accelerated during the pandemic. Companies quickly launched video conferencing applications to connect employees, partners and customers forced to work from home. Supply chain disruptions brought more use of cloud computing because of long delays in receiving physical servers.
The speed with which companies headed to the cloud left little room for strategic thinking, experts said. As a result, many companies found themselves paying for more cloud resources than they needed, duplicating applications on multiple clouds and running data-intensive analytics that would be less expensive on premises.
Professional services firm KPMG recently released a 2022 technology survey that found that 67% of the 1,000 executives polled hadn't achieved a substantial ROI from their cloud spending. Those same executives said their companies were reevaluating their use of the cloud to get more bang for their buck.
There's no doubt that cloud computing is here to stay, and companies plan to continue using it -- 3 in 4 respondents to the KPMG survey said their organizations are continuing to migrate strategic workloads to the cloud.
Nevertheless, the ongoing cloud reset contributed to a slowdown in cloud spending. IDC expects global cloud spending to reach $830.5 billion this year -- a 17.5% increase from 2021, but less than the 18.3% rise last year. IDC predicts cloud spending growth will fall further in 2023, to 16.3%.
The shift to a more cost-sensitive approach to running applications in the cloud or on premises is apparent in how companies are splitting data-intensive machine learning (ML) models among the cloud, private data centers and colocation facilities, said Thomas Robinson, vice president of strategic partnerships and corporate development at Domino Data Lab. The company offers data scientists the Domino Enterprise MLOps platform for developing, deploying and managing ML models in any location.
Domino Data Lab customers no longer head to the cloud first to run their ML models, Robinson said, explaining, "We see organizations being more judicious."
For example, pharmaceutical companies developing ML models to assist with cancer research learned that feeding petabytes of data into a model running in the cloud became cost-prohibitive.
"That spins the meter very, very quickly," Robinson said. Running a research model could cost tens of millions of dollars on a public cloud -- 20% to 40% more than on premises.
Datatron Technologies is another MLOps platform vendor that sees customers rethinking cloud use and moving workloads on premises. Approximately 20% of Datatron's customer base comprises retailers.
Because customers increasingly have workloads running in data centers as well as public and private clouds, Domino Data Lab and Datatron are building UIs that provide visibility to ML models running in multiple locations.
Domino Data Lab plans to launch its UI in January. Datatron doesn't have a firm date, but expects to launch the technology in the first half of next year.
"Having this capability across multiple clouds is something we are in the process of putting together," said Victor Thu, Datatron's president.
During the past eight months, hybrid cloud service provider Involta has seen an increase in the number of companies looking to reduce higher-than-expected cloud costs for running workloads on AWS, Azure and Google, said Josh Holst, Involta's vice president of cloud services. Besides price, companies also want to reduce application latency and have more in-house control over their data.
To get all three, companies are talking to Involta about moving workloads to its colocation facilities and using the public cloud as a backup, Holst said. Manufacturers and financial services companies seem more aggressive than others in pulling workloads out of the public clouds they moved to during the pandemic.
4Voice used to provide only cloud-based VoIP that it delivered from AWS to small and medium-sized businesses. Over the last year, nearly every potential customer has asked for a hybrid model that keeps the telephony system in the cloud, but voicemail and transcription services in house.
"They wouldn't even ask that question before," said Amruth Laxman, founding partner and CEO of 4Voice.
Enterprises' approach to the cloud is changing, but experts do not expect a mass exodus from public clouds. Instead, companies are changing architectures as they learn more about the differences in cost between public clouds and private data centers.
"Choosing between CSPs [cloud service providers], colocation and on-premises data centers will become an application-by-application choice," Enterprise Strategy Group said in a recent report on application infrastructure modernization trends.
Enterprise Strategy Group is a division of TechTarget.
Antone Gonsalves is the news director for the Networking Media Group. He has deep and wide experience in tech journalism. Since the mid-1990s, he has worked for UBM's InformationWeek, TechWeb and Computer Reseller News. He has also written for Ziff Davis' PC Week, IDG's CSOonline and IBTMedia's CruxialCIO, and rounded all of that out by covering startups for Bloomberg News. He started his journalism career at United Press International, working as a reporter and editor in California, Texas, Kansas and Florida.
Have a news tip? Please drop him an email.