Megaprojects are central to modern economies. They power cities, connect nations, and underpin industrial growth. Yet despite decades of digital transformation, nine in ten still exceed their budgets or schedules, draining billions from public coffers and private balance sheets. These failures erode trust and delay the benefits that large-scale infrastructure is meant to deliver.
Technology was supposed to change that. Over the past two decades, digital twins, cloud platforms and advanced analytics have become standard on major projects. But cost overruns and delays remain stubbornly common. The reasons are rarely technical. More often, they are structural, rooted in governance inertia, fragmented data and organisational silos that prevent insights from turning into action.
Artificial intelligence is now seen as a potential turning point. Its ability to model complex scenarios, forecast risks and optimise delivery has the potential to transform how major projects are planned and executed. Yet many organisations remain stuck at the pilot stage, unable to translate technical capability into real-world outcomes.
For Muriel Demarcus, Chief Executive Officer of Marsham Edge, the problem is not the technology itself but how organisations use it. “Because culture and incentives haven’t transformed alongside technology,” she said. “We invest billions in systems but hesitate to invest in integration between people, data, and decisions.”
Singapore is showing how integration can work in practice. With clear regulations, collaborative public–private partnerships and open data policies, the city-state has become a proving ground for AI-enabled infrastructure. Its approach offers valuable lessons for Asia-Pacific markets facing rising pressure to deliver projects that are faster, cheaper and more resilient.
Why digital transformation has not fixed megaproject risk
The construction and infrastructure sectors have been early adopters of digital tools. Digital twins now replicate physical assets in real time, cloud platforms enable collaboration across continents, and advanced analytics promise deeper insights. Yet the outcomes have barely changed. Most megaprojects still run over budget and behind schedule, revealing deeper issues beyond technology.
Even the most sophisticated tools can fail without strong governance and integration. Demarcus recalled one project equipped with a state-of-the-art digital twin, yet critical milestones were still being tracked manually in separate spreadsheets. Such mismatches between advanced systems and basic workflows are common, showing how technology can either amplify discipline through proper integration or fuel chaos when applied poorly.
Planning processes also remain too narrow in scope. Many still model only a limited range of risk and schedule scenarios, leaving projects exposed to unforeseen disruptions. Even where AI solutions are introduced, they are often built for generic analytics rather than the specific demands of engineering and construction. This “translation gap” prevents valuable insights from reaching the people making critical decisions.
Digital transformation alone has failed to tackle the root causes of megaproject underperformance. Without cultural readiness, aligned incentives and governance that embeds data into decision-making, technology stays superficial instead of driving real improvement.
AI as a performance engine, not just a dashboard
The next evolution in project delivery depends on shifting AI’s role from visualisation to execution. Beyond dashboards and data displays, AI can become the engine that drives more resilient planning, sharper decision-making and measurable performance improvements.
Marsham Edge’s work shows how powerful this shift can be. In one case, the company built forecasting models for electricity procurement that tracked price fluctuations across multiple markets and variables. Traditional approaches often rely on static long-term contracts or intuition, but the AI model provided a dynamic decision engine that guided clients on when to buy, hedge or hold. The result was millions in annual savings, driven by more precise timing and strategic procurement decisions.
AI has also reshaped project scheduling. Marsham Edge deployed simulation models to test thousands of sequencing scenarios and identify the most resilient approaches for complex construction programmes. The aim was never perfection but resilience, creating plans capable of absorbing uncertainty. The outcome was a measurable acceleration of delivery timelines, with projects moving faster through smarter planning that prevented delays before they emerged.
In another case, AI provided clarity in a data-rich industrial environment. A steel-cutting facility had installed 75 sensors to monitor every possible variable, from vibration to temperature. But machine learning analysis revealed that only three sensors were truly significant in predicting downtime. By focusing analytics on those key data points, downtime dropped by 70%. It was a lesson in the power of targeted insight over indiscriminate data collection.
These examples highlight a critical shift. AI delivers the most value when it is embedded into decision-making and treated as a partner rather than an accessory. As Demarcus noted, Marsham Edge’s approach is to combine engineering logic, data insight and business intuition to ensure technology delivers results in the messy realities of project delivery.
Overcoming adoption barriers at the board and team level
Despite its potential, many organisations struggle to move beyond isolated pilots. Resistance often begins in the boardroom, where complexity, perceived risk and uncertain returns deter decision-makers from committing. For Demarcus, the solution lies in demonstrating tangible results rather than promising sweeping transformation. “Boards don’t resist AI because they dislike innovation; they resist because they can’t see the return on investment. The answer is to demonstrate it through small, tangible wins.’”
Governance is another barrier to scaling. Successful organisations treat AI as core infrastructure from the start, establishing strict data quality standards, ethical safeguards and outcome-focused metrics. Without this foundation, promising proofs-of-concept often remain stuck as presentations, failing to evolve into enterprise-wide capabilities.
Empowering teams to adapt and refine models themselves is equally important. In many organisations, technical changes must pass through multiple layers of approval, slowing progress and limiting impact. Giving field engineers the authority to adjust models based on real-world conditions dramatically accelerates adoption and improves outcomes. As Demarcus observed, both megaprojects and AI initiatives fail for the same reason: when communication between teams breaks down.
Building cultural readiness is also essential. AI adoption involves more than technology; it demands behavioural change. Teams need to trust the tools and know how to apply their insights effectively. Clear communication and targeted training help bridge the gap between technical specialists and delivery leaders, ensuring AI outputs become meaningful action on the ground.
Singapore as a test bed for AI-driven infrastructure delivery
Beyond organisational change, the broader ecosystem plays a decisive role in how effectively AI scales. Singapore’s approach demonstrates how the right ecosystem can accelerate AI adoption in complex infrastructure projects. Its National AI Strategy 2.0 provides regulatory clarity and sets standards that reduce compliance risk. Government agencies collaborate across sectors, while open data policies make high-quality information widely available. Public–private partnerships further encourage experimentation and help solutions mature from concept to deployment.
“Singapore is fascinating: it’s a living lab for coordination. Clear regulatory pathways, cross-agency data sharing, and a culture that balances innovation with accountability,” said Demarcus. “You can’t copy-paste that into more fragmented economies, but you can replicate the principle: start with clarity. Define who owns the data, who benefits, and who’s accountable when things go wrong. Governance, not genius, is what makes Singapore scalable.”
These conditions allow AI to be tested, refined and scaled more quickly, reducing the time and cost of implementation. For other Asia-Pacific markets, many of which are characterised by fragmented regulatory systems and cautious adoption cultures, Singapore offers a valuable model. Replicating its emphasis on governance clarity, accountability, and data accessibility can help unlock AI’s potential even in less mature ecosystems.
Looking ahead: Scaling AI from pilots to core infrastructure delivery
As AI becomes more deeply integrated into infrastructure delivery, organisations will need to evolve beyond pilot projects. Enterprise-grade governance, stronger integration with delivery frameworks and clear ROI models will become essential to justify continued investment. At the same time, regulators and investors will demand greater transparency and accountability as AI takes on a larger role in decision-making.
Regional collaboration will also play a vital role. Sharing standards, frameworks and successful implementation models can help accelerate adoption in markets where ecosystems are still developing. This collaboration can reduce duplication of effort, spread best practices and help organisations scale more effectively.
Success will increasingly depend on a company’s ability to bridge the gap between technical expertise and operational execution. Those that build internal capability to translate complex analytics into actionable decisions will move faster, reduce risk and capture more value from AI. They will also be better positioned to meet the growing demands of regulators, investors and society for cost-efficient, transparent and accountable infrastructure delivery.
The fact that the majority of megaprojects still run over time or over budget has long been treated as inevitable. AI offers an opportunity to change that narrative. By turning vast amounts of data into targeted insights and embedding intelligence into the fabric of decision-making, AI can help projects deliver faster, cost less and perform better. The organisations that master this shift will not only redefine how megaprojects are delivered, but also reshape how economies build and grow in the decades ahead.
