7% mortgages are back. The Iran-war cost shock is splitting home cyclicals from AI software.

Seven months into the U.S.-Iran war, the market tell is no longer the warhead — it is the 30-year mortgage. Freddie Mac’s survey printed 7.03% for the week ending Sept. 24, the first trip back above the 7% threshold since January 2025, pushed there by surging Treasury yields and the energy shock tied to the conflict.^[1] That single number is doing most of the work in splitting our eight-name research scope into two widening camps: an AI-driven software pair whose demand keeps accelerating, and a home-furnishings cluster that is absorbing the interest-rate, freight, and fuel bill on the chin.

Call it roughly a 60/40 proposition that the software side of this split holds into next year and the housing-levered side gets pushed out another two quarters. The 40 is real, and both legs are expensive to get wrong, so let’s put the evidence on the table.

The freight and fuel shock is the quiet main character

The Strait of Hormuz — the chokepoint that carries roughly a fifth of the world’s oil — has been a shooting zone since the war began, and the knock-on effects are showing up far from the Gulf. Analysts at Reuters flag that ocean container rates could test record highs as the Iran-war fuel spike works through the system,^[2] and the cost of shipping a container from Asia to the United States has already doubled since the conflict started.^[2] Williams-Sonoma said in its Q2 call that higher oil prices were pressing on transportation and supplier costs, with fuel near current levels embedded in its guidance — and it reiterated that the tariff impact, peaking in Q2, would moderate through the back half.^[3]

None of this is lost on the builder complex. Homebuilder confidence fell to a one-year low in September, and Fannie Mae’s forecast pencils single-family starts down roughly 7% for 2026.^[1] Freddie Mac’s chief economist notes the housing market is still supported by a solid labor market — but the affordability math is the constraint, and 7% puts it back at the edge of the lock-in effect that froze existing-home turnover for two years.

The home-furnishings camp is absorbing the cost

This is where the scope bifurcates. Ethan Allen reported fiscal 2026 profit down more than 20% on softer sales, cushioned by high margins and a debt-free balance sheet.^[4] La-Z-Boy is pulling upholstery assembly out of Mexico — closing a San Luis Río Colorado plant and its Tijuana Joybird plant, moving output into existing U.S. plants with $17.6 million of one-time charges.^[4] Leslie’s, the pool retailer, is reportedly preparing for a Chapter 11 filing with a possible Nasdaq delisting looming.^[4] That last one is the extreme tail, but it frames the distribution: the discretionary consumer is present but value- and price-sensitive, and the winners are taking share at the expense of margin rather than earning margin on a rising tide.

RH is the sharpest expression of the trade-off. Demand at the luxury furniture maker is outpacing other home-furnishing companies — an analyst noted demand has “come at a pretty big cost to margin”^[5] — and management is leaning into the RH Estates catalog relaunch, with galleries transitioning mid-November toward 75-85% of the business converted by December.^[6] The international drag from European flagship openings, which pressured profitability, has no further European openings planned in 2027, which management says should help reduce that drag.^[6] RH’s math is a bet that the new-product launch can generate its own demand independent of the housing cycle — historically how the company has moved the business 15 to 20 points.

The software pair is running the other direction

On the other side of the line, Snowflake’s numbers are unambiguous in direction. Product revenue grew 37% year over year in its latest quarter — the third straight quarter of acceleration — with non-GAAP operating margin expanding 400 basis points to 15%, and management raised fiscal 2027 product revenue guidance by over 500 basis points to 36% growth.^[7] The prior quarter showed the same profile: product revenue up 34%, net revenue retention at 126%.^[7] The narrative is that AI consumption is a flywheel pulling platform demand through the model, not a discretionary line item customers can defer.

Datadog is a more mixed read. The underlying quarter was strong — revenue up roughly 36% to about $1.1 billion, called a key AI beneficiary — but shares fell on reports that a major AI client reduced usage, which is the exact vulnerability buried in this basket.^[8] One analyst’s framing captures it: “Expansion Is Broadening, but the Multiple Is Still Priced for Perfection.”^[8] That is the honest descriptor of both software names — the demand is real, but the market has already paid for it, and a single concentrated customer trimming consumption is enough to move the stock double digits.^[8]

What to watch next

• The Hormuz diplomatic channel. A senior Iranian official told Reuters the most realistic path is allowing navigation through the Strait of Hormuz in exchange for the U.S. ending its naval blockade — and oil fell Friday on optimism over that potential solution.^[9] If a reopening happens, freight-normalization is the ballast under every home-furnishing margin for 2027.
• The software reports. Datadog is scheduled for Nov. 5 before the open and Snowflake Dec. 2 after the close (estimated dates).^[10] The question on both: is any large AI customer slowing consumption, or is the flywheel still accelerating?
• Home-cycle confirmation. Williams-Sonoma reports Nov. 18, La-Z-Boy Nov. 17, Leslie’s Dec. 1, and RH Dec. 10 (estimated).^[10] Watch for whether management is building any housing-recovery into guidance — Williams-Sonoma explicitly said it is not.^[3]
• The 7% line. If 30-year mortgages hold above 7% into year-end, the housing-levered names likely grind sideways into spring. A drop back below 6.9% on de-escalation would change the setup quickly.

The honest read: this is a demand story that is really a rates-and-freight story wearing a demand costume. The software names are priced for continued acceleration with thin margin for error, and the home names are priced for a recovery that the 7% mortgage just pushed further out — unless the Hormuz channel delivers the de-escalation that reprices both.