Is AI in a Gulf-fueled bubble? And more! (with Michael Madowitz)
You asked. Michael answered.
Welcome to our little stack’s first mailbag, a feature we like to call Burning Questions. And readers, your questions did not disappoint. Want to know more about Zohran’s rent freeze, GDP’s flaws, and the vulnerabilities of today’s AI boom?
Our principal economist, Michael Madowitz, has you covered.
Let’s dig in.
Pete Bailey: Will Mamdani’s rent freeze help or hurt the overall NYC housing market for renters?
Michael Madowitz: This is a good one—Anisha Steephen [a fellow at our sister organization, the Roosevelt Institute] wrote about some of the macroeconomic stability effects of rent regulation last month, which are more involved than I initially appreciated. My suspicion is that it may not make as much difference to the overall market as proponents thought before the Iran war, even as the freeze eases pressure on about half the ~2.3 million rental apartments in NYC. New York rents are rising faster than in the rest of the country, and vacancy rates are still quite low, so increasing supply matters a lot.
I’m probably less concerned about downsides to a rent freeze than many economists I respect, but I am extremely worried that housing in NYC (and the country as a whole) is in for a rough year because inflation and interest rates are rising again, which has once again delayed an expected unlocking of the housing market and a housing construction recovery. Borrowing costs are the 800-pound gorilla of construction costs—a lot of policy tweaks here and there can make the market work fractionally better or worse, but these effects are easily swamped by higher interest rates.
Ryan, a NYC-based social policy analyst: Measurements are inherently values-laden given the choices made about what to measure or not measure. For GDP, this can mean that things like unpaid domestic labor or child-rearing are not counted as economic output. What tweaks could be made to how GDP is measured to more holistically capture the experiences and outputs of working families?
Michael: This is so important! First, there is good news here—over the last few years the Bureau of Economic Analysis (BEA), the part of the Census that produces GDP statistics, has started producing Distributional Personal Income data, which lets us see if GDP growth is actually showing up in the incomes of households at different income levels. In addition to building on that program, there are long-standing proposals to create national accounts that value changes in the environment that are well understood.
Of course you’ve already gotten to the most important missing piece of understanding how working families are doing: time, which is a key piece of a good life. Measuring real-time time use at a national level while respecting privacy concerns is a challenge, but one that is increasingly important to understanding well-being in an aging nation with growing care responsibilities.
Arya Harsono, NDC Partnership: You’ve argued that full employment is what makes a just transition possible, that it’s what gives displaced workers leverage instead of just displacement. Do you think AI-driven productivity gains are going to reinforce that leverage, or quietly erode it by automating the next generation of transition jobs before workers ever get there? And if AI ends up compounding the disruption of the net zero transition rather than easing it, who do you think actually has the standing to manage that: Is this a job for labor market policy and unions, or does it require something more like industrial policy that treats jobs and climate as one problem instead of two?
Michael: I’m so happy you brought full employment to this discussion—even though this is a hard question! Full employment, where unemployment is essentially voluntary, is a floor we need to maintain for all economic policy to work, so it’s going to be a key piece of any transition in the economy. I have yet to see anyone who isn’t working for an AI company claim AI will make a net zero transition happen faster—and that’s not that surprising! A net zero transition is closer to running a marathon than playing a game of chess: Figuring out how to generate cheap renewable energy is a solved problem, building enough of it is still the bottleneck. And so far AI is moving us farther, not closer, to where we need to get.
The question of how we allocate productivity gains from AI is partly economic but also democratic. The last generation of tech investment captured the vast majority of the benefits innovation created—in part because investors pursued anticompetitive business models, but also because the web 2.0 world was mostly one-time investments: Instagram is just a website that sells ads next to the content users create.
AI companies have to create the content, and it’s quite expensive, so it’s not obvious there’s a huge surplus they can easily capture themselves. But whether it’s monopoly rents, patents, copyrights, or other sources of business profits, how we allocate the benefits of new tech is determined by government policy, and a democratic economy requires the public to approve these policies. So whether new tech reinforces or erodes worker power is not nearly as mechanical as it can seem in a pitch deck.
Jess Burke: I agree with the broad argument that the economy is still resilient but that policymakers are gradually wearing away the buffers that help it absorb shocks. One pillar of the current AI investment boom has been growing capital from Gulf sovereign wealth funds. If a prolonged conflict with Iran reduces those investment flows, which seems increasingly likely as each day passes, in your view would that simply cool AI, or could it become the kind of additional stress that pushes an already less-resilient economy into something more serious? More broadly, do you view the outsized role of Gulf capital in financing AI as one of the factors that has been strengthening the long-term resilience of the US economy, or as masking underlying vulnerabilities by sustaining an investment boom that may prove difficult to justify if those capital flows slow? (In other words, is it possible that Gulf capital inflows are actually part of that pack-a-day our economy is smoking?)
Michael: I like this question, but I think it’s actually two questions! First, is dependence on Gulf investment a problem? And second, are those inflows going too heavily toward AI investment? The answer to both is a resounding yes, but the deeper questions are whether these trends are new or acute vulnerabilities.
Gulf investment in the US is a long-standing trend, so the vulnerability concern wouldn’t be a concentrated surge of sovereign wealth inflows to US startups, but that it could reflect a decline in diverse foreign investment. So far that prospect is still speculative, but we haven’t exactly been making a lot of friends abroad lately. The latest Treasury Foreign Investment data (only from mid-2025) doesn’t show a Gulf surge—Kuwait is the only Gulf state accounting for more than 1 percent of inflows; and for context, both the UK and Canada account for more than 10 percent—but the lack of a Sell American trade so far doesn’t mean we can rule it out.
Second, current growth in US investment is heavily concentrated in AI—and narrowly in the bet on high-cost, high-performance AI. An AI bubble is an economic vulnerability no matter where the funds come from. It’s a deeper vulnerability because we’re outsourcing our AI strategy and broader economy strategy to a small group(chat) of people rather than driving investment in US AI to a wider range of business models. If we don’t course correct, the next generation of behavioral economics textbooks could very well use VC’s narrow focus on bleeding-edge AI investment as a cautionary tale of “herding.”
That’s a wrap on our debut Burning Questions. Thanks to Michael for being our first expert in the hot seat, and to all of you who sent in questions. For those who were too shy this time, don’t you worry: Burning Questions will be back.
Until next time,
Matt
Fireside Stacks is a weekly newsletter from Roosevelt Forward about bold economic ideas and the people who put them into motion. If you enjoyed this installment, consider sharing it with your friends.