We Already Ran This Experiment

Key Takeaways

• BIM improved coordination but not project cost
• Change orders remained primary profit driver for contractors
• Software neutrality can't overcome misaligned incentives
• AI promises outcome‑based pricing, but needs contract reform
• Owners must drive performance contracts for real savings

Summary

A16z argues that construction software, still built on 1997 code, is holding back industry efficiency, while the author counters that the sector already experimented with BIM at scale without reducing waste. From 2007 onward, billions were spent on BIM tools, richer models, and coordinated workflows, yet budget overruns and rework costs stayed flat. The author attributes this to misaligned economic incentives across architects, engineers, and contractors, which turn rework into revenue rather than a failure. He concludes that AI and value‑based pricing can only succeed if contract structures shift risk and reward to owners who bear the lifecycle costs.

Pulse Analysis

The last decade saw the AEC industry pour billions into Building Information Modeling, hoping that three‑dimensional, data‑rich models would slash waste. While clash detection and internal coordination improved, the headline metrics—budget overruns and rework expenses—remained stubbornly high. This paradox stems from the fact that the tools merely digitized existing processes without reshaping the financial motives that drive them.

Every stakeholder in the construction chain is compensated in ways that reward change rather than stability. Architects earn fixed fees, leaving them indifferent to later revisions. General contractors profit from managing change orders, and specialty consultants bill hours regardless of design errors. Consequently, richer BIM models become instruments for documenting scope and defending claims, not for preventing costly modifications. To unlock genuine efficiency, the industry must redesign contracts so that risk and reward flow to the party—typically the owner—who ultimately suffers from overruns.

Artificial intelligence and large language models now offer capabilities that BIM could not achieve, such as parsing unstructured metadata and automating code‑compliant designs. However, these advances will only translate into $100 billion‑plus of value if they are tied to outcome‑based pricing structures. Owners need to demand performance‑linked agreements, and legal frameworks must enable shared risk. When software, incentives, and contract design align, the sector can finally move from incremental productivity gains to transformative cost reductions.