$64 Trillion and the Path to $140 Trillion

According to the EY report, global infrastructure faces a historic challenge. Although enormous sums have flowed into digital tools and modernization in recent years, productivity benefits often remained localized at the level of individual projects. Investments were not consolidated across programs or portfolios. The result is a global deficit of $64 trillion, which could climb to a need for $140 trillion in new investments by mid-century.

These numbers are not abstract. In the European context, they mean pressure to complete transport corridors, modernize energy grids in line with the Green Deal, or renovate water management infrastructure assets. Even in the Czech Republic, we struggle with long-term delays in key construction projects — from the D35 highway through railway modernizations to the preparation of high-speed rail lines. This is exactly where agentic AI can offer concrete solutions.

Why Is Infrastructure Stagnating? Data Fragmentation and Missing Coordination

EY warns in its study that infrastructure productivity is not hindered by a lack of innovation, but by the fragmentation of systems, data, and stakeholders. Most large projects today depend on manual coordination and periodic reporting. Data is locked in separate silos, limiting visibility across the entire construction lifecycle — from planning through implementation to maintenance.

One of the most expensive consequences of this fragmentation is rework — corrective work, which according to industry standards can reach up to 15% of total project costs. Translated to a global scale, this represents hundreds of billions in lost value annually. These losses are not caused by individual engineers' mistakes, but by systemic communication failures between the tools used by individual teams.

Agentic AI as an "Intelligence Layer"

While classical artificial intelligences often function as isolated tools — for example, for document analysis or risk prediction — agentic AI represents autonomous software agents capable of performing complex tasks across different systems without the need for constant human supervision. In the infrastructure context, EY describes this concept as intelligence layer, i.e., an intelligence layer that "sits" above existing systems and connects them.

Instead of a project manager having to manually click through budgeting software, CAD tools, schedules, and construction diaries, agentic AI continuously monitors these sources in real time. When it detects a discrepancy between the plan and actual costs, it automatically routes it to responsible persons. When it identifies a delay risk caused by the supply chain, it immediately alerts all stakeholders.

It is important to emphasize that the goal is not to replace people, but to free them from time-consuming administration. As Aurelia Costache, advisory partner and AI lead at EY Romania, states: "The real benefit of agentic AI lies in clearer insights, not just automation. Proper deployment supports faster project delivery while keeping people firmly in the driver's seat."

Five Pillars of Governance: Without Trust It Won't Work

For successful deployment of agentic AI in such a sensitive area as critical infrastructure, technology alone is not enough. EY therefore introduced in the report a governance framework based on five critical areas:

• Accountability and legal liability — it must be clear who bears responsibility for decisions mediated by an agent.
• Transparency and explainability — agents must be able to defend their conclusions, not function as a "black box."
• Data sovereignty and business confidentiality — especially in infrastructure, this concerns sensitive information about asset condition, budgets, and national security.
• Human oversight and competence — experts must understand what the AI is doing and be able to intervene.
• System resilience — the entire ecosystem must be designed so that the failure of one agent does not endanger the whole project.

This framework is not just theoretical. It is in line with the requirements of the EU AI Act, which emphasizes precisely transparency, human oversight, and accountability in deploying artificial intelligence in critical infrastructure. For Czech construction firms, transport network managers, and public investors, this means that any deployment of agentic AI will have to meet strict European standards — which is also an opportunity to build an advantage over competitors who ignore these rules.

What Does This Mean for the Czech Republic and Europe?

While the EY report has a global scope, its conclusions are also relevant for the Czech and European scene. The European Union plans massive investments in renewable sources, power grids, and transport infrastructure by 2030. The implementation of these projects, however, runs into the same barriers: fragmented data between ministries, regions, suppliers, and supervisors.

In the Czech Republic, examples of complex projects with a high risk of delays and budget overruns are precisely transport construction projects. Agentic AI could theoretically help with risk prediction based on data from previous projects, automation of budget controls, or coordination among subcontractors. In practice, however, it will be crucial for such tools to first prove themselves in pilot projects — for example, on large transport construction projects managed by the state or during modernization of railway corridors.

Technologically, agentic systems are already available today — companies such as Microsoft, Google, or specialized developers offer platforms for building autonomous agents. For the Czech market, however, the challenge remains to adapt these solutions to local legislation, language requirements, and the specific environment of public procurement.

Conclusion: Productivity Is Essential, Not Optional

Agentic AI is not a miracle cure for all infrastructure problems. Without proper data, clear processes, and human expertise, even the smartest agents can be helpless. The EY report clearly shows, however, that a productivity shift is essential if the world is to have a chance to fill the enormous investment gap. Losing 15% of costs to rework is simply unacceptable in an era when every dollar counts.

For Czech readers and professionals, the report is a signal that the future of construction and infrastructure will belong not only to heavy machinery, but also to smart data coordination. And that agentic AI may be precisely the connecting link that has been missing for so long.