Artificial intelligence (AI) is entering a new stage of growth - moving beyond development and into widespread, real-world use.
This stage, known as “inference,” represents the practical application of AI across industries, where systems are actively deployed to perform tasks rather than simply being trained.
A key development within this shift has been the rise of Agentic AI - an evolution that is accelerating adoption and expanding the potential of AI across both consumer and enterprise environments.
What is Agentic AI?
Agentic AI refers to systems that can autonomously carry out workflows and tasks, often with minimal or no human intervention. Unlike earlier forms of AI that generate outputs based on prompts, these systems are designed to take action - coordinating processes, making decisions, and executing tasks continuously.
This marks a significant shift in capability. Rather than simply providing insights, AI is increasingly able to “do the work” - opening up a much broader set of use cases across areas such as:
- Enterprise productivity
- Commerce and customer service
- Advertising and personalisation
- Everyday digital assistance
Major technology firms are already investing heavily in this space, developing AI agents designed to operate continuously and integrate into daily workflows.
From Cost Centre to Core Infrastructure
As AI adoption scales, its role within organisations is also evolving. What was once seen largely as a cost centre is now rapidly becoming core infrastructure, underpinning revenue generation and operational efficiency.
Recent developments suggest this transition is happening faster than expected:
- AI usage is growing at an accelerated pace, driven by new agent-based applications
- Companies are increasingly adopting “Agentic-first” strategies
- AI platforms are being embedded across customer-facing and operational functions
This shift points to a more sustained and structural phase of AI investment.
Why This Phase Matters
The transition to Agentic AI is not just about new applications, it is reshaping the entire AI ecosystem.
- Increased Demand for Compute Power
Agentic AI systems operate continuously and perform multiple tasks simultaneously. As a result, they require significantly more processing power than earlier AI applications.
This is driving:
- Higher overall compute demand
- More complex and distributed workloads
- A need for different types of processing infrastructure
- A Shift in Infrastructure Requirements
Earlier stages of AI, particularly model training, were heavily reliant on specialised processing hardware. In contrast, Agentic AI focuses more on coordinating and executing tasks in real time, which changes the underlying infrastructure mix.
Key implications include:
- Greater importance of traditional processing systems to support ongoing workloads
- Increased demand for memory and storage to retain context, history, and data
- A broader range of technologies playing a role in AI delivery
- Growing Importance of Data Storage and Memory
Agentic AI relies heavily on access to persistent data - such as user history, prior actions, and contextual information - to function effectively.
This is driving demand for:
- Large-scale data storage systems, particularly in cloud and data centre environments
- Memory technologies that support fast access to data
- Infrastructure capable of handling rapidly growing volumes of information
In this environment, data is increasingly seen as a critical input, often described as the “fuel” needed to power AI systems.
- The Expanding Role of Data Centres
The rapid growth in AI usage is also transforming the role of data centres.
While earlier AI development was concentrated in large, centralised facilities, the rise of Agentic AI is expected to drive greater demand for distributed infrastructure, including facilities located closer to end users.
This is because:
- Agent-based systems require real-time responsiveness
- Lower latency becomes critical for performance
- Workloads are increasingly spread across multiple locations
As a result, co-location and edge data centres are likely to play a larger role in supporting AI deployment, particularly for enterprise applications.
- Sustained Investment Supporting Growth
One of the key factors underpinning the continued expansion of AI is the scale of investment flowing into the sector.
Large technology companies are committing significant capital to build out AI infrastructure, including:
- Data centres
- Compute capacity
- Supporting technologies
Importantly, this investment is being supported by strong operating cash flows, suggesting that the current growth cycle is financially sustainable rather than speculative.
Forecasts indicate that investment in AI infrastructure could reach substantial levels over the coming years, reinforcing the long-term nature of this trend.
- A Broadening Opportunity Set
As AI moves into the inference and Agentic phase, its impact is expected to extend across a wider range of industries and technologies.
Rather than being concentrated in a narrow set of applications, AI is becoming a foundational capability influencing:
- Technology infrastructure
- Energy and power systems
- Data storage and management
- Enterprise software and services
This broadening effect highlights how AI is evolving into a general-purpose technology, with implications across the global economy.
Looking Ahead
The emergence of Agentic AI represents an important inflection point in the evolution of AI.
As adoption accelerates, key themes to watch include:
- The continued shift from experimentation to real-world deployment
- Rising demand for infrastructure, particularly compute and data storage
- The expansion of AI into new industries and use cases
- Ongoing investment supporting long-term growth
While still at an early stage, this next phase has the potential to reshape how businesses operate and how technology is embedded into everyday life.
