Enterprises Hit Governance Gaps as 76% Run GPU Workloads in Production
Updated
Updated · InfoWorld · May 21
Enterprises Hit Governance Gaps as 76% Run GPU Workloads in Production
10 articles · Updated · InfoWorld · May 21
Nearly three-quarters of enterprises are training machine-learning models and 76% already run GPU workloads in production, pushing agentic AI from pilots into live operational workflows.
That shift is exposing cloud architectures built for application deployment, not governed AI execution: almost all organizations must migrate more than 25% of their data to support model pipelines.
Roughly half of respondents still rely on public AI tools, while fewer than a quarter have enterprise-wide governed AI deployments on a shared framework, leaving audit, policy and compliance controls uneven.
Multicloud sprawl deepens the strain, with many companies managing six to 20 cloud accounts and mixed infrastructure tools, making retraining, inference, identity and logging harder to align.
The report argues the main scaling constraint is architectural fit rather than model choice or build-versus-buy, with durable growth requiring AI-native governance, observability and compliance from day one.
With Colorado's AI Act looming, is your company's 'Shadow AI' about to become its biggest legal liability?
AI project failures exceed 80%. Is this a crisis of technology or a failure of architectural vision?
The $758B AI Challenge: EU AI Act, Shadow AI, and Securing Enterprise Infrastructure by 2026
Overview
With the EU AI Act set to become effective in August 2026, organizations—especially multinational enterprises—are under immediate pressure to overhaul their AI governance frameworks. The rapidly approaching compliance deadline is forcing urgent action, as the Act introduces stringent operational requirements like full data lineage tracking, human-in-the-loop checkpoints, and risk classification documentation. These demands require a fundamental shift in how AI systems are developed, deployed, and monitored, creating substantial compliance and governance challenges. Multinational corporations are finding it difficult to integrate these requirements into unified compliance programs, highlighting the complexity and urgency of the situation.