FACT WIKI · Concept
AI infrastructure
AI infrastructure
AI infrastructure consists of the hardware and software needed to create, deploy and manage applications and workloads powered by artificial intelligence.
Roles and functions
AI infrastructure helps developers create and deploy applications such as virtual agents, facial and speech recognition, and computer vision. It forms part of a broader AI stack that also includes frameworks, tools and services supporting the AI lifecycle.
How it works
Edge AI runs models on local devices such as cameras and sensors. This allows organizations to obtain immediate insights without sending the processing work to cloud infrastructure.
A hybrid deployment approach combines public cloud services for scalable model training with infrastructure on an organization's premises for reliable, high-volume inference—the running of AI models.
Scope and limitations
In industries such as finance and insurance, regulations can constrain where data is stored and processed. IBM identifies these constraints as an important consideration for AI infrastructure in on-premises and private data centers.
Background and related concepts
IBM describes AI sovereignty as a goal of infrastructure customization: giving organizations direct control over their AI models to support organizational independence, security and compliance.
References
Related stories
FactSnap stories are secondary summaries.