Brian Potter is a structural engineer, Senior Infrastructure Fellow at the Institute for Progress, and author of The Origins of Efficiency. He spent two and a half years at the failed construction startup Katerra, which raised $2 billion attempting to revolutionize building through factory prefabrication methods.

The conversation explores why construction productivity has stagnated while other industries have dramatically improved, examining the recurring failures of prefabrication attempts from Sears mail-order homes through modern startups. Potter draws on insights from Why Nothing Works by Mark Dunkelman to explain how American politics shifted from a Hamiltonian impulse favoring government capability to a Jeffersonian suspicion that has created regulatory barriers to infrastructure development.

Potter discusses the tension between local opposition to development and broader economic benefits, the surprising success of AI datacenter construction, and potential solutions ranging from federal permitting reform to changing cultural attitudes about growth and progress.

Katerra's $2 Billion Prefabrication Failure

Katerra attempted to revolutionize construction through factory-built prefabrication, an idea that 'is perennially popular' with someone trying it 'every 10 to 20 years' - Brian

The company scaled operations before achieving product-market fit, building what Potter believes was 'the largest CLT plant in the world' for cross-laminated timber panels

'Almost as soon as they got it online, they were sort of trying to get rid of it because it no longer fit what they were trying to do' - Brian

COVID disrupted operations, but Potter notes the fundamental problem was 'constantly trying to sort of figure out what that needed to be' rather than having a proven product to scale

The Prefabrication Myth Across History

Sears sold nearly 100,000 mail-order houses from 1908-1940 based on 'knockdown boat' assembly concepts, but ultimately lost money on mortgages during the Great Depression

Ford's Model T mass production success in the 1930s inspired widespread belief that 'the same ideas should work' for complex houses assembled like cars

Sweden builds 80-90% of single-family homes and 40% of apartment buildings using prefabricated construction, yet 'their costs are not like low' and exceed US housing prices

The recurring thesis that 'if I move my process into the factory, it will become much more efficient' has failed repeatedly across decades of attempts

From Hamiltonian Power to Jeffersonian Paralysis

Why Nothing Works by Mark Dunkelman describes competing tendencies: Hamiltonian 'robust muscular government' versus Jeffersonian suspicion that views government as 'a big danger to its citizenry'

The Power Broker biography of Robert Moses marked a turning point where 'people started becoming much more suspicious of the government's power and authority' in the late 1960s-early 1970s

Environmental laws like the National Environmental Policy Act, while 'good rules,' also 'serve to restrict what the government is able to do' and enable litigation to halt projects

'We've been on the end of several decades of this Jeffersonian impulse reigning supreme' leaving government agencies unable to accomplish major infrastructure projects - Brian

AI Infrastructure: The Exception That Proves the Rule

Current AI datacenter buildout represents 'one of the great infrastructure construction efforts in history' comparable to 'Manhattan Project or Apollo' in capital deployment

Datacenters succeeded initially because they 'paid a lot property taxes, but they didn't demand all that much in the way of new government services' unlike population-dense developments

'We can still deploy infrastructure and resources to build capabilities really, really astoundingly quickly if there's motivation to and if stuff is not blocking the way' - Brian

Growing opposition now emerges as new datacenters stress power and water infrastructure while 'AI capabilities are making so many people nervous'

The Local NIMBY Problem: Concentrated Harms vs Diffuse Benefits

'A city overall will benefit from population growth' through 'bigger, stronger economy, more division of labor' but costs are 'concentrated right next to that one spot' - Brian

Federal housing czar would face limitations since 'a lot of restrictions on building things are at the state or regional or local level' through city supervisors and design review boards

Federal government has 'a lot more ability to influence' large-scale energy infrastructure through permitting reform for transmission lines, pipelines, and solar installations

California's transformation from growth-celebrating state with population tracking signs to anti-growth policies in the 1960s exemplifies how affluence breeds opposition to development

AI's Promise for Construction and Automation

'AI is going to make dramatically new types of automation possible' reducing 'tasks that you used to be having to do manually' - Brian

Current AI excels at 'any information processing task' but robotics capabilities 'are not as good as driving a robot around yet' compared to chatbot performance

'If you could automate a big fraction of construction work, you would be able to drive down the cost of building' and 'finally address this construction productivity problem'

Potter speculates a 'robot GPT-3 moment' could emerge from training models on human teleoperation data, potentially requiring 'many millions of dollars' but 'maybe less than you might expect'