AI Just Broke the Long-Term Hold

I recently attended my 20-year business school reunion at UC Berkeley. It was great seeing old friends and classmates and re-exploring campus after all of these years. Berkeley is as crazy as it ever was, though it has found religion on housing policy as it has grown. New housing was everywhere around campus. If only other Bay Area cities would take note of what’s worked in Berkeley, of all places, and apply it in their own municipalities.

Part of the weekend included a symposium on campus, where Toby Stuart, an entrepreneurship and innovation professor at the Haas School of Business, gave a lecture on AI. The talk referenced his recent Harvard Business Review article, The Future is Shrouded in an AI Fog.

He didn’t offer specifics on how many jobs would be lost or gained because of AI. He didn’t speculate on the winners and losers, an impending SAAS-pocalypse, or how much further NVIDIA’s stock will climb. Instead, he spoke about the unknowns created by AI and how they impact corporate decision-making. His basic premise: AI has the ability to change so much, so quickly, that long-term investing is near impossible today. Not just investing in bonds or stocks, but investing in yourself—your career, your education, anything that once had a somewhat linear path.

So what does this mean for real estate investing?

Shorter Duration Holds

I’m an advocate of short duration investments in real estate at the moment. AI is driving rents in San Francisco to insane levels. $6,000/month for a one-bedroom in a 100-plus-year-old building in a great neighborhood? How about $12,000/month for a two-bedroom in the same building? Or a two-bed, two-bath condo near all of the AI jobs for $15,000 per month? I question how sustainable this is, but at the same time I’m not sure when it will slow down. There’s clearly an AI gold rush happening, but won’t these people put themselves out of a job because of AI?

Nobody knows.

If AI has a massive impact over the next five years, rents continue to climb in San Francisco because it can’t (won’t) build enough housing, and AI drives down interest rates globally through efficiencies and productivity gains, real estate should be worth a heck of a lot more. But there’s political risk in rents growing too fast. You’d have to assume serious pushback against out-of-control rents in a place like San Francisco, and in California at large. Does that negate the drop in interest rates, and with it the drop in cap rates?

San Francisco is also the quintessential boom-bust market. It’s booming quickly. Will the bust come just as quickly, and just as severely? Every investment carries unknowns. Covid reminded all of us of that. But the speed with which the world is now changing demands optionality.

Other Markets

If you’re not bullish on San Francisco from here, the ship has likely sailed on finding a good deal there. AI companies are also taking office space in Silicon Valley, though the dynamics are different. In San Francisco, it’s not uncommon to see massive rents in vintage buildings alongside a tenant paying $700 who has lived there for 45 years. That doesn’t happen in Silicon Valley. Even so, San Jose was just listed by RealPage as the #2 market in the country for forecasted rent growth at 3.3%, behind only San Francisco.

That may not sound like much, but when very few markets are growing at all, it’s a big number. And it’s being driven by demand from AI, technology, and adjacent services. What other markets will see these tailwinds? It wouldn’t surprise me if places like Austin, Dallas-Fort Worth, and Raleigh-Durham post gains thanks to their tech, finance, and healthcare ecosystems. But there’s significant apartment supply still to absorb in those markets, and the fundamentals remain weak. Those are wait-and-see markets. There’s not much benefit to buying in them today unless the yield is high enough to wait. At the first sign of rent growth, though, I expect these markets to accelerate again.

Optionality

Plenty of owners who bought in 2021 and 2022 wish they had more options today. The biggest option-killers are leverage and basis. If you paid too much, took on too much debt, or both, your options are limited. You need things to improve quickly in order to exit. If you paid a reasonable cap rate on real net operating income and have debt that lets you exit when needed without strangling your cash flow, you have optionality.

This is where Toby Stuart’s fog metaphor really lands for me. A 10-year or even 7-year hold is too far out to underwrite with confidence. What cap rate will be reasonable by then? What rent growth will be feasible? What will the regulatory environment look like? I’d rather have a business plan that can be executed in three to five years, based on today’s assumptions, with reasonable leverage that preserves my options—the option to sell if the market turns against me, and the option to hold if the AI runway proves longer and the returns greater.

Some would say that’s just smart real estate investing to begin with. They’d be right. But we’ve all grown accustomed to relatively linear cycles lasting anywhere from about seven years (dot-com bust to GFC) to about twelve years (post-GFC to the 2022 rate hikes). The speed of AI adoption and impact will make the next few cycles quite bumpy until we reach a prolonged cycle—whatever that ends up looking like—once the dust on AI has settled. Until then, I’m looking for maximum flexibility and the shortest duration in real estate.