How to optimize AI investments through cloud modernization
Cloud-focused services will help with AI workloads. Users also plan to use on-premises storage for AI and should take the opportunity to modernize infrastructure.
We provide market insights, research and advisory, and technical validations for tech buyers.
Published: 24 Apr 2024
As AI initiatives bubble to the top of business priority lists, too many organizations fail to take full advantage of their investments. Often the investments necessary for AI can also foster additional modernization initiatives.
This year's Google Cloud Next conference offered a vision of how organizations can invest to address both their AI and their cloud-modernization goals simultaneously. At the April event, Google unveiled myriad product releases and announcements, mostly focused on simplifying the development, training or use of AI capabilities or AI-based workloads.
Examples of Google's AI-centric cloud-infrastructure announcements included the following:
Upcoming support for the Nvidia Blackwell GPU platform within Google's AI Hypercomputer architecture.
General availability of Cloud Tensor Processing Unit v5p.
A preview of Hyperdisk ML, an optimized block storage cloud service designed to speed up inferencing.
New caching capabilities generally available for Cloud Storage Fuse and in preview for Google Parallel Store.
This focus on AI is no surprise, given Google's history of innovation in AI and the expected surge in investment coming this year. For example, a recent research study by TechTarget's Enterprise Strategy Group found that 54% of organizations expect to have generative AI in production in the next 12 months.
Enterprises seek on-premises storage for AI
While public cloud services such as Google's are poised to benefit from AI investment, Google also introduced new innovations to address the growing interest in deploying AI workloads on premises, whether at the data center or the edge. According to Enterprise Strategy Group research, 78% of organizations said they prefer to keep their high-value proprietary data in their own data centers.
Enterprise Strategy Group also performed a separate analysis of on- versus off-premises deployment decisions. Data sovereignty or governance considerations (cited by 45%) and data already residing on premises (40%) were two of the more common rationales that led cloud-first organizations to deploy a new application within a data center rather than the cloud. By extension, data locality also often determines where associated AI workloads are deployed.
AI projects offer an opportunity to modernize on-premises infrastructure and support a broader set of applications and requirements.
Further highlighting the interest in new on-premises infrastructure to support AI, Enterprise Strategy Group research found that 68% of organizations expect to procure new on-premises storage infrastructure in the next six months. Most of those organizations identify that a generative AI project is fueling all or part of that investment.
AI projects offer an opportunity to modernize on-premises infrastructure and support a broader set of applications and requirements. Businesses should evaluate other objectives that AI infrastructure investments can help achieve.
How vendors package services to ease complexity burden
On-premises modernization has been a priority for several years -- 91% of organizations agree that data center modernization offers a strategic competitive advantage, according to Enterprise Strategy Group research. In recognition of the increased interest in both infrastructure modernization and using AI on premises, Google announced a generative AI search service packaged with Google Distributed Cloud. It will include Gemma, Google's suite of pretrained models, and will be available for preview in the second quarter of this year.
Packaged services such as this one offer an opportunity to speed up time to value with AI projects. In the case of Google Distributed Cloud, it can also offer a broader modernization opportunity. Google Distributed Cloud is designed to deliver Google Cloud services on infrastructure deployed onsite within a data center or edge location. It also offers an air-gapped option for high security environments.
Enterprise Strategy Group asked organizations to identify their preferred options for creating a modernized, cloud-like experience on premises. Using on-premises hyperscale products such as Google Distributed Cloud, AWS Outposts or Microsoft Azure Stack was the most common answer, with 41% of respondents.
The cost and complexity incurred from disparate experiences across locations is unsustainable in an era where hybrid cloud is the dominant operating model. Organizations should prioritize investment in consistency to speed up operations, reduce the burden on internal staff and reduce risk.
Products from VMware and Red Hat also offer options to extend a consistent experience across locations. In addition, Dell Technologies, with its Apex Cloud Platform suite, offers the option to deploy a VMware, Red Hat or Microsoft Azure experience on premises as a service.
Introducing a generative AI search packaged service with Google Distributed Cloud is a savvy move. Enterprise search is considered a popular early use for generative AI projects. Google already has strong experience and brand reputation in the area.
Organizations that use this packaged option with Google Distributed Cloud should see quicker time to value and increased flexibility in terms of where they can begin their AI journey. That journey can and should include speeding up on-premises modernization efforts to a broad set of applications, beyond just AI.
Scott Sinclair is Practice Director with TechTarget's Enterprise Strategy Group, covering the storage industry.
Enterprise Strategy Group is a division of TechTarget. Its analysts have business relationships with technology vendors.
