This article is part of the VB Special Issue titled “Fit for Purpose: Tailoring AI Infrastructure.” Explore all the stories here.
Enterprise data infrastructure is rapidly evolving, driven primarily by the demands of data-intensive generative AI.
Are You Ready for AI Agents?
As generative AI reshapes enterprises, leaders face the challenge of optimizing their data across cloud, edge, and on-premises solutions. They require immediate access to data while ensuring its security.
To navigate this complexity, many organizations are adopting hybrid models. These models leverage the strengths of cloud, edge, and on-premises setups. According to IDC, 85% of cloud buyers are either implementing or planning to deploy a hybrid cloud.
Priyanka Tembey, co-founder and CTO at Operant, states, “The pendulum between the edge and the cloud has shifted over the past decade. There are numerous use cases where computing can benefit from being closer to the edge or through a hybrid combination.”
The Shifting Data Infrastructure Landscape
Historically, cloud computing has been linked to large hyperscale data centers, but this perception is changing. Dave McCarthy, research VP at IDC’s cloud and edge services, explains, “Organizations now realize that the cloud is an operational model that can be deployed anywhere.”
With the maturation of cloud technologies, businesses are reevaluating their architectures to better integrate hybrid cloud and edge computing, especially to maximize AI’s potential.
As Miguel Leon, senior director at WinWire, notes, “The combination of hybrid cloud, edge computing, and AI is profoundly transforming the tech landscape. As AI becomes integral to businesses, its connection with hybrid models will deepen.”
The Role of Edge Computing
IDC research predicts edge computing spending will hit $232 billion this year, addressing several challenges not solved by cloud alone. A primary concern is latency-sensitive applications. McCarthy highlights that quick responses are crucial in sectors like manufacturing, where delays can impact production lines.
“Edge computing processes data closer to its source, cutting down latency and enhancing business agility,” Leon adds. It also enables AI applications requiring rapid data processing, such as image recognition.
Edge computing proves beneficial in environments with limited connectivity, such as IoT devices that frequently fluctuate between coverage areas. For example, autonomous vehicles rely on AI functionality even when networks are unavailable.
With data generation soaring—estimated at 328.77 million terabytes daily and projected to exceed 170 zettabytes by 2025—managing data efficiently becomes critical. McCarthy observes that as remote data accumulation grows, so do the costs of transferring it to centralized locations. However, an edge computing system can effectively determine which data to retain.
Compliance challenges arise as data sovereignty regulations come into play, making edge solutions essential for meeting legal requirements regarding data location.
Scalability Challenges
As organizations move AI projects from trial phases to full deployment, scalability is a significant concern. McCarthy points out that a surge in data can overwhelm core infrastructures, much like the early internet’s need for content delivery networks (CDNs) to cache data closer to users. “Edge computing will perform the same function for AI,” he says.
Advantages of Hybrid Models
Different cloud environments cater to various needs. For instance, public clouds excel in scaling during peak usage, while on-premises and private clouds provide enhanced control and security over proprietary data. The edge ensures resiliency and optimal performance.
“The advantage of a hybrid cloud is the ability to select the appropriate solution for each task,” McCarthy explains. Hybrid models see diverse applications; in financial services, mainframe systems integrate with cloud environments to maintain data centers for banking while utilizing cloud services for customer interactions. In retail, local systems can independently handle transactions during outages, ensuring continuity as AI systems are implemented to analyze customer behavior.
Tembey emphasizes that a hybrid approach allows AI processing on-device, at the edge, and in larger cloud environments, safeguarding sensitive information.
However, hybrid environments can introduce management complexities, especially in multi-vendor settings. To address this, cloud providers are extending their platforms to support on-premises and edge deployments, while OEMs and ISVs increasingly partner with cloud providers.
Interestingly, 80% of IDC survey respondents indicated plans to migrate some public cloud resources back to on-premises solutions. McCarthy remarks, “While cloud providers pushed for a fully cloud-based future, the reality is different.”
This evolving landscape underscores the need for organizations to adapt to a hybrid infrastructure that meets their specific challenges, especially as AI continues to reshape how businesses operate.
Ranking
