A bridge, dam or water system built today with a 100-year design life will likely still be operating well after 2120. Infrastructure often outlasts its original design horizon. Yet the engineers making capital decisions now need credible information about climate, demand and technology decades beyond what current science can reliably project. This is the reality beneath every major infrastructure investment. Most leadership teams still plan as if the future is knowable, when it is not.
The Assumption That Broke
For decades, infrastructure design relied on the idea that historical climate and demand data could predict future conditions. That assumption no longer holds. Climate is shifting in ways the records cannot capture, and the impacts are already measurable. Under high emissions, nearly half of global transport assets could see storms that were once-in-fifty-year events arrive far more often by midcentury. By 2100, that risk rises further as warming accelerates. Infrastructure built to a static, historical risk profile is being engineered for a world that is already disappearing.
Two Ways to Get It Wrong
Leadership teams responding to this uncertainty tend to fail in one of two predictable directions, and both share the same underlying mistake. Underbuilding: designing to the historical baseline without accounting for how conditions will change. Most leadership teams respond to this uncertainty in one of two ways, both flawed. Underbuilding—designing to historical baselines—leads to expensive retrofits, service disruptions and political battles for more funding when the original design fails early. Overbuilding commits capital to capacity that may never be used, creating stranded assets and poor returns. Both mistakes come from the same place: designing as if the future is knowable, when it is not. Valuable options that would be difficult or expensive to add later, rather than either committing fully to one predicted scenario or building generic flexibility that never gets used. This is a meaningfully different design philosophy: not "an asset that can become anything," but an asset where the owner has consciously protected the interfaces, land, and structural capacity needed to modify it in plausible, high-value ways as conditions actually unfold.
Long-term adaptation usually falls into four paths: maintain the asset as is, harden it against greater risk, relocate it when the site becomes untenable, or accept reduced service or abandonment when no other option is justified. Standards like Canada's CSA S910.1:25 for dam climate vulnerability show how this thinking is moving into practice. Instead of prescribing a single solution, it sets a repeatable process for evaluating exposure, sensitivity and adaptive capacity across the asset's life, bringing climate uncertainty directly into the decision rather than treating it as a problem to solve before acting.
The Governance Discipline Most Leaders Skip
Adaptive infrastructure only works if someone is accountable for using the preserved options when needed. Most organizations skip this step. A proper handover should include an options register, clear triggers for action, ongoing monitoring and scheduled adaptation reviews—not just as-built drawings filed away after construction. In practice, this approach delivered value in a reservoir project in Mombasa, Kenya. By planning for new climate information over time, the team could invest in reliable, incremental stages instead of a single large build, reaching more vulnerable communities with less upfront capital.
What This Means for Infrastructure Leadership
A project team may oversee an infrastructure build for five years. The asset will operate through many technology cycles, climate shifts, funding environments and leadership changes. Durable infrastructure does not come from claiming certainty about 2075. It comes from building governance, documenting options, defining triggers and ensuring accountability, so future leaders can use the flexibility preserved by today's decisions.
Most organizations still plan for a single predicted future. Few build in the options needed for the futures that may actually arrive.
