For a better experience, click the Compatibility Mode icon above to turn off Compatibility Mode, which is only for viewing older websites.

Housing for Chips

Below is the Nowak Metro Finance Lab Newsletter shared biweekly by Bruce Katz.

 

Sign up to receive these updates.

 

January 16, 2024

Scientific Data and Computing Center (SDCC), formerly known as the RHIC and ATLAS Computing Facility | Photo by the U.S. Department of Energy

​(co-authored with Ben Preis and Michael Saadine) 

Housing has dominated state and local headlines, the federal campaign trail, and even this newsletter of late. But you wouldn’t know it if you were moving within private capital circles. Rather, the talk of the town in the public markets, amongst lenders, and at private equity conferences is data centers: housing for chips.

The advent and growth of artificial intelligence, and broader technological growth before it, has brought with it significant downstream effects. Tech companies, including Microsoft, Google, and OpenAI, need data center space to store their computing capacity and capability. Data centers are a very unique type of real estate: they need cooling, they need power; they don’t need windows, they need fewer bathrooms. Today, there is an almost insatiable demand for data centers to power artificial intelligence models, and those data centers have an almost insatiable demand for power. All at once, private capital and the world of innovation have rallied to build these centers and provide them power. For years we have hand-wrung about whether nuclear energy is a worthwhile investment as we pursue a greener economy; Microsoft and Meta are now reviving and building nuclear power plants to fuel their data centers. President Biden just signed an executive order making federal land available for development of data centers to be leased by the Department of Defense and Department of Energy[1].

Blackstone, the country’s largest private investor in real estate, has a $70 billion data center portfolio, with $100 billion of acquisition and development in its pipeline[2]. Microsoft is spending $80 billion on data centers this year[3], and a future single project in partnership with OpenAI could cost $100 billion[4]. The industry is projected to approach $800 billion in size by 2034[5].

At the median cost to produce a housing unit of $250K, the $170 billion of capital associated with Blackstone alone could produce 680,000 housing units. Now, Blackstone and other scaled private equity firms have done plenty of investing in housing. But by and large, and especially lately, they have vastly preferred buying and renovating or buying and holding housing to funding excess production of it. While housing starts increased over the pre-COVID “zero-interest-rate-phenomenon” time period, they have plummeted since rates were raised to battle inflation. And even at their peak, we were not producing enough to fill our deficit.

What a shame that the constructs of our housing market make it unexciting, and even financially irresponsible, for these capital sources to build new housing. What could be a more rewarding challenge than innovating the most important kind of real estate around: people’s homes? Instead housing production is years behind other industries, overregulated, and unsophisticated. Artificial intelligence models, with all the uncertainty about their continued growth and ultimate role in society, work their way to the front of the private capital line while housing is a lost cause.

We have talked a lot about treating the housing crisis like the crisis it is. Instead of housing, data centers are the opportunity du jour. Perhaps it’s time to treat the housing opportunity like the opportunity it is.

A New Housing Opportunity

Traditionally relied-upon recent estimates of the country’s housing shortfall range from 3.7 million units[6] to 5.5 million units[7]. A new study takes it multiples further, saying the deficit is 20.1 million homes[8]. No matter the starting point, millions of additional units are considered severely inadequate, or obsolete[9]; our housing stock is older than it has ever been and requires significant outlays for repairs. Still millions more units are located in areas of extreme climate risk[10], and we have recently witnessed how devastatingly quickly entire communities can be wiped away.

These tens of millions of units of potential need do not even account for consumer preferences – how many new units would be demanded if it was relatively frictionless to move closer to a new job, a favorite hiking trail, or an amenity-rich urban or town center, or away from uninsurable climate risk? What if it was easy to upgrade to newer housing to reduce repair, maintenance, and operating costs?

The net-new housing opportunity, excluding the massive existing housing acquisition industry, easily breaches $1 trillion with the obvious needs in front of us and is likely multiples of that when considering what a better-functioning market would provide consumers. These numbers are intimidating, but if they could represent an attractive investment opportunity, they would constitute the most exciting industry in the private sector – as building new housing has been in decades past. Far bigger than even the most aggressive estimates of data center needs.

A $1 trillion (and likely more) additional housing need sounds intimidating, but if data centers can do it, housing should be able too as well.

What Stands in the Way

By now, the fundamental challenges to a functional housing system are well-documented. However, the root causes underlying our problems are complex and interconnected.

We know the production of housing is overregulated. What causes local overregulation? Is it bureaucratic incompetence caused by misaligned incentives in our federated political system? Is it disagreement on how to measure housing needs and set goals? Is it because people are overprotective of their homes and neighborhoods? If so, is that instinct driven by too much net worth being tied up in homes, societal issues like crime and racism, or all of the above?

We know our land is underutilized. Is it because of the regulation enumerated above? Is it because of the public sector’s inability to account for its most valuable land? Is it because we overtax structures and undertax vacant land? Is it because of a concentration of ownership by groups that can hoard land for decades without action? Is it too difficult to utilize anything but expansive, greenfield land?

We know construction costs are severely high. Is this solely because of interest rate movements? Because of supply chain and labor costs? Have we built a system riddled with complex programs, middlemen, and fees that add up to be insurmountable? Why haven’t we significantly innovated our construction methods in decades?

Capital is choosing to allocate to other sectors, or buying and holding instead of developing sufficient net-new housing. Where does the math break down? Is housing risk mispriced? Is the capital breakdown circularly connected to the overregulation, land costs, and construction costs mentioned above?

It is historically easier to innovate in a relatively new segment, like data centers, than one with centuries of legacy institutions, methods, and norms. New AI models, new chips, new developments to house them, and new energy to fuel those sites capture the excitement and energy of the capital markets. The technology sector’s hyper concentration allows it to aggressively take on constraints to deployment.

The housing system is a mess. We have a population that is suffering under high costs, but allergic to potential change. We have a complex web of institutions, intermediaries, consultants, and advocates working in service of narrow slices of the system instead of a collective North Star. We have well-intentioned subsidies whose usefulness is weighed down by complexity and cost. As a result, capital sources simply look the other way.

An Ecosystem Approach to Create the New Housing Opportunity

Our housing ecosystem needs a radical remaking. We are realists about the need for incremental change in the meantime, which is why the National Housing Crisis Task Force issued a federal policy agenda in November and is hard at work on a state and local action plan highlighting bottom-up innovations.

But any earnest effort to solve the housing crisis must mean deeply rethinking the systems that have gotten us here. Should homeownership be as emphasized as it is? What level of housing density works best for our cities, our suburbs, and our towns? Should we be constructing differently? Which entities and intermediaries represent pure bloat? Who has the power in the system? Are other countries doing it better?

The entire ecosystem needs to be mapped, with an eye towards bottlenecks and breakdowns. The most high-leverage points need to be examined. Stakeholders from federal, state, and local governments, to philanthropies, to financial institutions, to developers need to convene and examine their roles and connectivity.

We need to rebuild the system to shift from hand-to-hand combat against a growing crisis, to creating the next great housing opportunity.


[1] https://www.whitehouse.gov/briefing-room/presidential-actions/2025/01/14/executive-order-on-advancing-united-states-leadership-in-artificial-intelligence-infrastructure/
[2] https://www.datacenterdynamics.com/en/news/blackstones-prospective-data-center-pipeline-hits-100bn-on-top-of-70bn-portfolio/
[3] https://www.bisnow.com/national/news/data-center/microsoft-will-ramp-up-data-center-spending-to-80b-urges-trump-to-boost-american-ai-dominance-internationally-127401
[4] https://www.forbes.com/sites/cindygordon/2024/03/31/microsoft-and-openai-partnering-on-stargate-a-100b-us-data-center/
[5] https://www.forbes.com/councils/forbestechcouncil/2024/12/23/five-trends-that-will-define-the-data-center-industry-in-2025/
[6] Freddie Mac
[7] National Association of Realtors
[8] https://www.iza.org/publications/dp/15447/the-understated-housing-shortage-in-the-united-states?utm_source=substack&utm_medium=email
[9] Improving America’s Housing 2023, Joint Center for Housing Studies of Harvard University
[10] https://www.cbsnews.com/news/climate-change-home-prices-housing-realtor-com/?utm_source=chatgpt.com


Bruce Katz is the Founding Director of the Nowak Metro Finance Lab at Drexel University. Ben Preis is the Director of the National Housing Crisis Task Force and a Senior Research Fellow at the Nowak Lab. Michael Saadine is a Senior Advisor to the Nowak Lab and Managing Partner at Invisible Group, an interdisciplinary real estate investment platform.