Why we’re still building essential retail in the AI age
Summary
- AI is now the economy’s center of gravity. It accounts for about 45% of the S&P 500, the largest segment of investment-grade debt and 86% of U.S. venture dollars.
- That concentration is a reason to own essential retail. Investors loaded with AI exposure need durable, contractual income, and almost no new centers are being built.
- Publix-anchored centers have already passed the rate test. Across 32 sales since 2019, the 10-year rose about 370 basis points while cap rates held essentially flat.
- The thesis holds whether AI transforms the economy, grinds along or busts. The biggest AI investors are still buying grocery-anchored centers, and we’re still building them.
I’ve always been a sci-fi guy. And one thing you can count on in movies, comic books and novels is that, sooner or later, artificial intelligence wakes up and decides humanity is the problem. Think HAL 9000 in 2001: A Space Odyssey, Ultron in the Avengers movies, or pretty much any Agent roaming around The Matrix. And who hasn’t joked about Skynet from the Terminator movies when talking about an AI breaking out of its sandbox? I’m not losing sleep over Skynet just yet, but I am paying close attention to how much of our economy now rides on its real-world cousins.
The money chasing AI stopped being a tech story a while ago. It’s in the stock market, where AI-linked companies now make up about 45% of the S&P 500. It’s in the bond market, where AI borrowers have become the single largest segment of investment-grade debt.1 And it’s in the economy itself, where, depending on who’s counting, AI spending is driving somewhere between a fifth and nearly half of U.S. growth.
So here’s the question I’ve been asking myself as someone who builds grocery-anchored shopping centers for a living: with this much money chasing AI, will investors still want to own what we build?
Skynet doesn’t have much need for groceries, but the rest of us still do. The same goes for a haircut, a physical therapy session, or dinner out on a Friday night. I still believe a great corner with a grocer, a restaurant, a bank or a wellness concept on it may be one of the most underappreciated counterweights to the AI trade. That’s not because AI will fail. In fact, count me among the people who think it will beat every forecast. It’s because essential retail holds up in all three futures: the one where AI works so well it starts replacing the people who used to do the work, the one where the boom simply keeps going, and the one where it busts. Our own data on Publix-anchored centers, which held their value through the sharpest rate rise in four decades, suggests investors agree.
Come With Me If You Want to Live…
General admission to the market now includes the AI rollercoaster. Investors may never have chosen the ride, but their index funds, bond funds and pensions chose it for them. AI-linked companies now account for about 45% of the S&P 500’s total value, a record for any single theme and roughly twenty points higher than when ChatGPT launched in late 2022.1 SpaceX went public in June at a valuation of about $1.8 trillion, the largest IPO ever. Anthropic has filed confidentially to follow, and OpenAI is reportedly close behind.2
Then there are the bricks and sticks that house the chips. The five big cloud companies, Alphabet, Amazon, Meta, Microsoft and Oracle, are expected to spend close to $780 billion on capital projects this year, nearly double what they spent last year, and Wall Street expects more than $1 trillion a year from them after that.3 To put that in terms of my own world: the entire country spends somewhere around $28 billion a year building new retail space.4 Five companies are outspending that by more than twenty to one.
And then there’s the borrowing. Through July, the hyperscalers had issued about $132 billion in bonds, already more than in any prior full year and about 11% of all U.S. investment-grade supply.3 When the biggest companies on earth flood the bond market, the rest of us compete for what’s left, real estate developers included. That’s the corporate debt wave I wrote about last month, and part of why long-term rates sit where they do.
Private money is even more concentrated. In the first half of this year, AI companies took 86% of every venture capital dollar invested in the U.S., about $356 billion out of $413 billion.5
Economists argue over the exact share, but estimates of how much of U.S. growth now comes from AI investment run from about a fifth (Morgan Stanley)6 to about a third (ING)7, and higher on narrower measures in some quarters. The center of gravity for the American economy is squarely in the AI universe.
The data center build-out won’t last forever, because trees don’t grow to the sky. I don’t know how long this ride will last. But for me, the question isn’t whether AI is a “bubble.” It’s whether, during this run, investors will still have the room and the appetite to own anything else.
My gut says they will, and the reason is the concentration itself. The pension funds, sovereign wealth funds, insurance companies and family offices that buy shopping centers are the same institutions holding all this AI exposure. Logically, that should send them looking for a way off the ride, and toward durable, predictable income in particular. Grocery centers, strip retail and single-tenant buildings provide just that: rent from tenants selling things people need every week, under long leases, in a format almost nobody is building anymore. And investors seem to agree. More than 80% of retail investors surveyed by JLL this year still target grocery-anchored centers,8 and grocery-anchored sales volume jumped 42% last year to nearly $11 billion.9 These buyers aren’t waiting for rates to fall. They’re buying the cash flow, not the financing.
I’ll Be Back
When an investor buys a grocery-anchored shopping center, they’re buying more than a building. They’re buying a predictable, long-term stream of rent, and the value proposition is how much they believe in the tenants behind it.
A center like ours is anchored by a grocer such as Publix on a long-term lease, surrounded by tenants that sell everyday needs: a restaurant, a bank branch, a pharmacy, a hair salon, a coffee shop, a fitness studio. Most of those leases are net, meaning tenants also pay their share of taxes, insurance and maintenance, so the landlord’s income isn’t eaten away when those costs go up.
The grocer is the anchor, the engine for the retail machine. Most families shop for groceries every week (if not more), and every trip is a chance to grab lunch, pick up a prescription or get a haircut next door. The shopper, like a certain cyborg, always comes back. That’s why anchored, credit-tenant centers outperform. JLL puts grocery-anchored vacancy at 4.0%, versus 6.3% at centers without a grocer, with rents running about 4.4% higher.9 The anchor supports the daily trips, and that traffic attracts the other essential retailers.
Then there’s supply, or the lack of it. The United States now spends about $28 billion a year, adjusted for inflation, building new retail space, down from roughly $70 billion in 2007.4 In the second quarter, CBRE recorded the least new retail space delivered since it began tracking,10 and the total stock of retail space is growing just 0.4% a year.11 As a developer, I know the headwinds: rates, construction costs and macro uncertainty mean only the best projects get done.
Single-tenant net lease buildings, such as a bank branch or a quick-service restaurant on a long lease, offer a simpler version of the same thing: something close to a bond with a building attached. Even with AI pulling attention elsewhere, investors are still buying this thesis. Starwood paid about $2.2 billion for the net lease platform Fundamental Income and added another 48 properties in the first half of this year.12
None of that income depends on a frontier model release or an earnings call. But AI’s appetite for capital is one of the forces keeping interest rates high, and real estate is supposed to be the asset that suffers when rates rise. So does this passive income actually hold its value when rates spike?
Judgment Day
In commercial real estate, the number everyone watches is the cap rate: a property’s annual net rent divided by its price. A 6% cap rate means a buyer is paying about 16.7 times the property’s annual net income. When cap rates go up, values go down, and the usual assumption is that cap rates move in tandem with interest rates. If an investor can earn 5% risk-free on a Treasury, why accept 6% on a shopping center?
The good news for us is we’ve already played out this exact scenario. Since 2019, the 10-year Treasury has climbed from about 1.3% to about 5% on trend, the sharpest rate rise in four decades. The steepest part of that climb came in 2022, when the Fed moved aggressively to contain inflation. More recently, record borrowing, including the AI build-out, has helped keep long-term rates elevated. If cap rates moved in lockstep with rates, values should have fallen.
Working with our good friends at Berkeley Capital Advisors, we gathered Publix-anchored shopping center sales they tracked across North Carolina, South Carolina, Georgia, Florida, Tennessee and Alabama from April 2019 through August 2026, 32 in all. Over that stretch, the 10-year trend rose about 370 basis points, from 1.27% to 4.96%. The Publix cap rate trend went from 5.96% to 5.82%. Values actually improved in this period. Plot each sale’s cap rate against the 10-year on the day it sold and the best-fit line is essentially flat: a full percentage point move in the 10-year has lined up with about a two basis point move in cap rates. The asset set the price, not the rate.
Splitting the sales at the Fed’s first rate hike in March 2022 tells the same story. Before the hike, ten sales averaged a 6.01% cap rate with the 10-year at 1.67%, and a median price of $272 per square foot. After it, 22 sales averaged a 5.82% cap rate with the 10-year at 4.12%, and the median price rose 16% to $315 per foot.
The cleanest example is a single property. Weaverville Plaza, near Asheville, sold in February 2020 for $23.5 million at a 6.87% cap rate, with the 10-year at 1.60%. It sold again in January 2026 for $30.5 million at a 6.00% cap rate, with the 10-year at 4.13%. Same center, rates up 253 basis points, value up 30%.
To be fair, 32 sales is a meaningful sample, but this certainly isn’t a robust statistical study (I’ll leave that to others). And measured from the low point in 2022, Publix cap rates have risen about 82 basis points, though that’s still less than the 100 to 200 basis point rise UC Berkeley’s Fisher Center cites for commercial real estate overall.4 But investors appear to share my view on the asset class. Buyers want income they expect to grow, with almost no new development competition, from tenants who aren’t going anywhere.
And the good news is that none of this strategy depends on rates coming back down. The Fisher Center’s base case puts the 10-year at 5.25% over the next year.4 These centers don’t need rate relief to hold or increase their value. They’ve already shown they can hold it, and grow it, while rates rose.
No Fate But What We Make
Sarah Connor’s classic line works here. I have no idea which reality will emerge in the AI age, but I believe the essential retail asset class will endure under any plausible outcome.
In the first future, the one I believe in, AI succeeds, and succeeds broadly at scaling up productivity. Frontier models keep improving, agents start doing real work, and efficiency gains show up in the data. That’s good news for the economy and (maybe) hard news for a lot of knowledge workers. But people still eat, shop and gather. They still need a haircut, a dentist, a prescription and somewhere to get dinner, and none of that comes through a chatbot.
On the grocery front, will AI change how people buy groceries? Honestly, I think we answered this question with the advent of Instacart and DoorDash and the answer is yes, it already has. Online orders now account for roughly a fifth of U.S. grocery spending, and in July, all of the year-over-year growth in grocery sales came from online while in-store sales fell 2.6%.13 AI shopping agents will likely push that further. But look at how those orders get filled. Strip out ship-to-home, which most regional grocers don’t offer, and online is about 16% of grocery spending, most of it picked off the shelves of a physical store and handed to a driver or to a customer in the parking lot. For a grocer like Publix, the store is the warehouse, and the anchor’s rent doesn’t change when the shopper sends an agent instead of pushing a cart. What I’m watching as a landlord is tenant mix. We’re focusing on tenants that need a physical trip: food, fitness, medical and personal services. So I expect cap rates to hold, because the investor’s use case doesn’t change.
The second future is the one we’re living in now. The AI boom keeps going, capital keeps pouring in, and rates stay elevated, but Skynet takes a little while longer to become self-aware. Online grocery keeps growing, yet roughly four of every five grocery dollars are still spent in a store,13 and the weekly trip remains the rhythm of most households. In this world, the pressure on real estate doesn’t come from changing habits. It comes from the cost of capital, but the good news is that we’ve already run that test. Publix-anchored cap rates held through a 370 basis point move in the 10-year, and investors kept buying. I expect cap rates to keep holding.
The third is the bust. Capex gets cut, AI equities reprice, and the wealth effect at the top of the economy goes into reverse. This world hurts everyone, including retail. But a bust would likely bring lower rates and a flight to safe, passive income. In that world, durable, contractual rent is exactly what capital runs toward, and cap rates have room to compress.
Salvation
On September 28, the 10-year closed at 5.24%, its highest level in a year. That same day, Brixmor and Everview agreed to buy Slate Grocery REIT for $2.34 billion, all cash, with no financing condition, and the Abu Dhabi Investment Authority joined as a strategic investor.14 Norway’s sovereign wealth fund has committed $500 million to a neighborhood retail venture with Asana Partners,15 and Bain and 11North bought ten Southeast centers, seven of them Publix-anchored, for $395 million.16 Sovereign funds like Abu Dhabi’s and Norway’s are among the largest holders of AI equity on the planet. They aren’t betting against AI. They’re balancing it.
So, back to the question I started with: with this much money chasing AI, will investors still want to own what we build? I believe so. Not because I doubt AI, but because the more capital crowds into one idea, the more valuable it becomes to own something that doesn’t depend on it. And essential business retail is the natural “physical” complement to the AI virtual world.
We’re bringing a new Publix-anchored center online in Florence, South Carolina, next month, and we have many more of these developments planned throughout the Southeast over the next two years. It’s the same playbook we’ve followed for decades: great corners, credit tenants, long leases. Skynet may end up running the world someday. But in the fight between Skynet and shopping carts, I’m putting my money on the carts.
References
Cited in text
1 Kobeissi Letter data, as reported in “AI Swallows Wall Street: Stocks Hit Record 45% of S&P 500 Market Cap,” BeInCrypto via Yahoo Finance, 2026. https://finance.yahoo.com/markets/stocks/articles/ai-swallows-wall-street-stocks-094052523.html
2 Reuters reporting on the SpaceX initial public offering, priced June 11, 2026 ($75 billion raised; $1.77 trillion valuation), and on Anthropic’s confidential IPO filing, June 1, 2026.
3 Vanguard, “The AI Buildout Comes to the Bond Market,” August 19, 2026, based on Bloomberg data through July 31, 2026, as reported by Chief Investment Officer. https://www.ai-cio.com/?p=106712
4 Hartung, Christopher. “Looking Past the Fog.” Presentation to the NC State Pappas Real Estate Development Program Industry Council, September 2026. Fisher Center for Real Estate + Urban Economics, UC Berkeley. Source for retail construction spending (U.S. Census Bureau, 2017 dollars, 2Q 2026), the rise in commercial real estate entry cap rates since 2022, and the 10-year Treasury base case. This presentation also shaped the macro framing of this piece. Used with permission.
5 PitchBook-NVCA Venture Monitor, Q2 2026, as reported by SiliconANGLE, July 9, 2026. https://siliconangle.com/2026/07/09/pitchbook-us-venture-funding-hits-412-7b-first-half-ai-deals-dominate/
6 Morgan Stanley, “AI Spending: A New Engine for the Global Economy,” Thoughts on the Market (podcast), July 2026, as reported by 24/7 Wall St., July 22, 2026. Estimate is net of imported equipment. https://247wallst.com/investing/2026/07/22/morgan-stanley-ai-spending-could-hit-1-4-trillion-by-2028-but-60-leaves-the-us/
7 Knightley, James. “How much is AI contributing to US economic growth?” ING Think, 2026. https://think.ing.com/opinions/how-much-is-ai-contributing-to-us-economic-growth/
8 JLL, 2026 Retail Investor Survey, as reported by REBusiness. https://rebusinessonline.com/retail-real-estate-enters-new-era-of-confidence-say-jll-survey-takers/
9 JLL, 2026 Grocery Tracker (2025 transaction volume, vacancy and rent premium), as reported by Chain Store Age and Bisnow. https://chainstoreage.com/jll-grocery-anchored-centers-notch-highest-visitation-rates
10 CBRE, U.S. Retail Figures, Q2 2026, as reported by Chain Store Age and Modern Retail, August 2026. https://chainstoreage.com/cbre-retail-rents-increase-24-q2
11 Marcus & Millichap, “U.S. Retail Real Estate Outlook 2026,” August 2026. https://www.marcusmillichap.com/research/special-report/2026/08/us-retail-real-estate-outlook-2026-why-fundamentals-remain-strong-despite-economic-uncertainty
12 Starwood Property Trust, Form 10-Q for the quarter ended June 30, 2026 (Fundamental Income acquisitions).
13 Brick Meets Click, Grocery Shopper Survey, as reported by Digital Commerce 360 (Q1 2026) and Talk Business & Politics (July 2026). https://www.digitalcommerce360.com/article/online-grocery-sales/ and https://talkbusiness.net/2026/08/the-supply-side-annual-e-grocery-sales-expected-to-top-363-billion-up-4/
14 Brixmor Property Group, Form 8-K, Exhibit 99.1, “Brixmor Property Group & Everview Partners to Acquire Slate Grocery REIT for $2.34 Billion,” September 28, 2026. https://www.sec.gov/Archives/edgar/data/0001581068/000110465926111144/tm2626458d1_ex99-1.htm
15 Asana Partners, “Asana Partners and Norges Bank Investment Management Launch Strategic Neighborhood Retail Venture,” press release, July 7, 2026. https://www.prnewswire.co.uk/news-releases/asana-partners-and-norges-bank-investment-management-launch-strategic-neighborhood-retail-venture-302819614.html
16 Commercial Observer, “Bain Capital, 11North Partners Buy Retail Center Portfolio for $395M,” August 2025. https://commercialobserver.com/2025/08/bain-capital-11north-partners-buy-retail-portfolio/
Market and company data referenced
Ten-year Treasury yields, 2019–2026, including the September 28, 2026 close. Board of Governors of the Federal Reserve System, H.15 Selected Interest Rates, via FRED series DGS10. https://fred.stlouisfed.org/series/DGS10
Publix-anchored shopping center sales in North Carolina, South Carolina, Georgia, Florida, Tennessee and Alabama, April 2019 to August 2026 (32 transactions). Berkeley Capital Advisors, used with permission; analysis by The Morgan Company.
Weaverville Plaza sale terms (February 2020 and January 2026) and Palmetto Place (Florence, SC) are drawn from The Morgan Company’s records and Berkeley Capital Advisors transaction data.
Prior article referenced in text: Morgan, Trey. “Borrowing Short, Holding Long: Why we changed our financing and not our business.” The Morgan Company, August 30, 2026. https://morgancompany.com/borrowing-short-holding-long-why-we-changed-our-financing-and-not-our-business-by-trey-morgan/
