Why We Invested in Paces
For most of human history, the built world has been shaped by the location and movement of labor and other resources. In the AI era, the world will be shaped by the location and movement of electrons.
Location, location, location.
Even laypeople outside the real estate industry understand that location defines real estate. But what do we mean by location? Why do developments emerge in some places but not others? Why is real estate (of any asset class) worth more in San Francisco, California than in Omaha, Nebraska?
On a macro level, the value of a location comes down to two dimensions: What are the resources available in that location, and what is its access to various forms of mobility (to access other resources located elsewhere)?
In every prior technology revolution, different resources and modes of mobility have become the currency and constraint, shaping the built environment.
In the Neolithic (agricultural) revolution, arable land led to resource surplus and population growth, creating the first “urban” environments. When these towns linked to each other via waterways, enabling trade and access to even more surplus resources, they were able to develop into major port cities like Rome. It’s no accident that the majority of cities that are economically important today (and where real estate value is highest) originated as ports.
In the Industrial Revolution and Information Age, similar dynamics — especially the availability and movement of raw and finished goods and labor — shaped development of the US Steel Belt as well as Silicon Valley.
A new set of constraints
In our view, the availability and movement of labor and goods will diminish in importance over time in the AI era. As work increasingly becomes automated, the importance of its location declines. As transit becomes autonomous, costs decrease and the mobility of physical goods becomes less constrained.
With those constraints cleared, a new bottleneck emerges: energy.
This is a paradigm shift when we think about real estate development and value, and our view is that it will have a massive impact on which investors and developers win or lose over the next decade. But first, let’s explain why.
The hungry, hungry AI model
The massive energy consumption of data centers required to support AI has been well documented, and the comparisons are mind-boggling: An AI search, for example, is ten times more energy intensive than a traditional search; Training an AI model is equivalent to streaming Netflix for nearly two centuries. Conservative forecasts warn that AI will consume as much energy as small but economically important countries like Japan or the Netherlands.
Importantly, the increase in energy demand is taking place in the context of multiple other megatrends, including a boom in domestic manufacturing, electrification of transport and buildings, and the transition from centralized coal and natural gas power generation to decentralized power with a focus on renewables.
The economy is effectively moving from being powered by a diverse set of “energy” resources, including labor (onshore and offshore), petroleum and natural gas, all moving across a diverse set of modes of transit including pipelines, tanker trucks and railroads, to just one type of energy — electricity— moving along just one mode of transit: an aging grid.
Taken together, the stress on grid operators is not only the increasing total power generation capacity required to support these trends (which need to double over the next few decades, no small feat in the context of the last few decades of flat to declining capacity) but even more critically, the increasing need for transmission.
Development realities
The US transcontinental railroad took just 6 years to build. The US interstate highway system took nearly four decades. With the rise of NIMBYism, ballooning of construction costs and massive shortages in skilled labor, we’re not optimistic that the annual decline in new transmission lines built per year is suddenly going to reverse.
Meanwhile, new energy generation and storage projects are already sitting longer and longer in interconnection queues. Grid operators don’t publish equivalent data on the load side, but anecdotally we know the situation is worsening. It can take years for data centers to get the interconnection approvals they need, and even relatively minor power upgrades for apartment buildings to install electric vehicle charging stations can take months to complete.
Managing interconnection and permitting risk
When our team met Paces earlier in 2024, we were immediately impressed for all the typical reasons VCs get excited by in early software companies: compelling team (relevant start-up experience, ex-Meta AI), great metrics (5X YoY growth), glowing customer feedback. Their first product, focused on site selection, was being used by some of the top solar and wind developers to vet potential sites based on permitting and interconnection risk using Paces’ proprietary data and projections.
But even more than that, the Paces team had the same premonition and vision about how the next decade-plus would unfold in the built world that we did — that energy availability and connectivity was acutely painful in a small number of sectors today, but that this pain would spread. And therefore, building software to manage site selection, due diligence (and permitting) and the pre-construction process with a focus on energy risk would not only be valuable, but critical — a necessary tool for developers, investors, and even utilities and municipalities — across every type of development, from power generation to communities of new homes.
We couldn’t be more excited to team together with James and Charles as they build the best-in-class platform to manage development risk for all grid-connected projects. Welcome to the Navitas portfolio!
