Dr. Atif Ansar is the co-founder and Executive Chairman of Foresight, a global provider of data driven project delivery technology.
getty
The great AI buildout of 2026 isn't just a tech trend; it is one of the largest, most coordinated capital deployments in human history. In 2026, we may see a projected $562 billion poured into AI infrastructure in a single calendar year.
Beneath the surface of this gold rush lies a staggering structural failure: Our delivery models were never built for this velocity.
While the capital is flowing, the concrete isn’t drying fast enough. Research reveals a sobering reality for the industry: Nine out of 10 data center projects (90%) run late. These aren't minor hiccups; we are seeing average schedule delays of 34%.
Consider the context. Often, these aren't underfunded startups; they are titans in the industry. From what I've seen in the industry, vacancy rates are dropping, creating a desperate "need it yesterday" environment. Ultimately, the problem isn't a lack of money or demand. It is a visibility crisis.
The lesson for 2026 is clear: AI data centers aren't missing their marks because developers lack talent. They are failing because the sheer scale of the buildout has rendered traditional management tools obsolete.
The industry is attempting to scale at light speed using delivery models inherited from an era when errors could be absorbed. Today, time-to-market is the key differentiator that matters. In this high-stakes environment, a month delay isn't just a shifted deadline—it is vanished revenue that can never be recovered.
To illustrate my point, I'll turn to a common illusion well researched in psychology known as the Müller-Lyer illusion. The illusion consists of two identical horizontal lines, but the addition of arrows at their ends creates a powerful trick of the eye. On one line, the arrow points inward, while on the other, the arrow faces outward. Even when you know two lines are the same length, your brain insists one is longer because of the arrows at their ends. Experts who have seen this illusion a thousand times still can’t "think" their way out of it. They need a ruler.
Project delivery works the same way. Decades of experience don't exempt a leadership team from bias; often, that experience reinforces it. The schedule, filtered through layers of management, tells the CEO what they want to hear—until the variance becomes too massive to hide. By then, the window for timely intervention has slammed shut.
When asked about project timelines, many CEOs offer a standard deflection: "Our project control team handles that." This perspective treats the schedule as a clerical document rather than a financial one.
However, those who want to win the AI race need to reject this delegation. They must view the schedule as a balance sheet tool. When time is viewed through the lens of IRR, valuation and lease commitments, behavior can shift.
So, how do you shift to this new perspective? The most successful organizations I've observed are adopting a governance model borrowed from risk management: the three lines of defense.
The power of this framework is that it eliminates the single point of failure by creating layers of accountability:
1. Execution: On-site teams oversee the day-to-day schedule. They have the autonomy to move fast, but they operate within clear standards. They own the work.
2. Review: This is a centralized PMO or project controls team that acts as a "reality check." Their explicit job is to challenge assumptions and counterbalance optimism bias. They can identify risks while they are still manageable, ensuring that "green" dashboards aren't hiding "red" realities.
3. Approval: Senior leadership and members of the C-suite aren't just bystanders; they must approve critical milestones and baseline changes. They must treat delivery as a portfolio-level trade-off between time, cost and risk, ensuring every construction decision aligns with the balance sheet.
In the current AI buildout, the second line of defense, review, is almost always the weakest. Most organizations have talented boots on the ground and engaged executives at the top, but they lack the independent, systematic "middle layer" required to surface problems while they are still solvable.
Without a robust second line, you aren't managing a project—you’re just waiting for the next surprise. Turning this around can be greatly aided by data-driven project delivery platforms harnessing AI, machine learning and advanced analytics to translate siloed data into a single source of truth. With these real-time insights, owners, operators and contractors can proactively identify risks, prevent overruns and deliver on time.
The $562 billion that is projected to funnel into AI this year is predicated on the assumption that the industry can actually build what it has promised—on time and on budget.
The systems currently governing that delivery are buckling under the weight of it. Addressing this blind spot requires better technology, certainly—but more importantly, it requires a leadership shift. Predictable delivery starts the moment the C-suite stops viewing the schedule as a project update and starts viewing it as a core fiduciary responsibility.
Forbes Technology Council is an invitation-only community for world-class CIOs, CTOs and technology executives. Do I qualify?
此内容由惯性聚合(RSS阅读器)自动聚合整理，仅供阅读参考。 原文来自 — 版权归原作者所有。