The artificial intelligence (AI) landscape is evolving at a rapid pace. The key issues organizations face in 2025 include greater financial accountability, Agentic AI, domain-specific and small language models, and best practices to promote scalability, according to GlobalData, a leading data and analytics company.
GlobalData’s latest report, “2025 Enterprise Predictions: Artificial Intelligence,” reveals that in 2025, organizations will take a more mature approach to implementing GenAI, developing more realistic mid- to long-term roadmaps and identifying KPIs to measure the impact of projects and conduct ROI analysis to evaluate success.
Rena Bhattacharyya, Chief Analyst and Practice Lead for Enterprise Technology and Services at GlobalData, comments: “Many organizations have been experimenting with GenAI with varying degrees of success. What was an opportunistic stance to GenAI in the past, will evolve to a more structured and strategic approach that incorporates analysis of where the technology can make the greatest business impact.”
GlobalData predicts Agentic AI will accelerate GenAI adoption in the enterprise, thanks to the capacity to accelerate productivity across a range of tasks. Enhanced memory capabilities enable the systems to develop a greater sense of context, including the capacity for planning.
Beatriz Valle, Analyst for Enterprise Technology and Services at GlobalData, notes: “With improved reasoning capabilities, all these GenAI agents can better understand and anticipate the intention of human users, translating them into a series of steps, taking the initiative and acting independently.”
Looking ahead, enterprises should expect technology developments to make on-premises and edge deployments of AI a more realistic alternative for 2025. For example, the emergence of small language models not only lowers the cost and complexity of deploying GenAI, it also means that more organizations can consider running GenAI on-premises or at the edge.
Bhattacharyya concludes: “The transition to AI at the edge has significant implications for real-time applications that require low latency connectivity, since they perform better when processing is done closer to the point of data collection. Furthermore, AI processing at the edge can assuage data privacy and regulatory concerns, since potentially sensitive information does not need to be transported across large distances or regional boundaries. However, the move to edge processing has connectivity implications that will require organizations to reevaluate their networking requirements and overall cloud strategies.”
