The resilience test is splitting software from furniture

The current market question is not simply whether geopolitical risk is “risk-off.” It is whether demand that is mission-critical can keep compounding while physical goods absorb a higher cost of movement.

The evidence across the requested scope points to a two-speed answer. Snowflake and Datadog are still describing expanding, embedded workloads. Williams-Sonoma and RH are showing that brands can gain share and protect demand, but only while working harder against tariffs, fuel and freight. The result is a more selective version of the original hypothesis: resilient demand can support parts of the group over the next year, but the transmission mechanism matters.

The market tell

At the September 30 close, SNOW rose 2.8% to $339.56 and DDOG rose 2.0% to $273.85; in after-hours trading, SNOW was $340.93, or 0.4% above its 16:00 ET close, while DDOG was $273.22, or 0.2% below it, as of 19:43 and 19:59 ET respectively.^[1] RH closed at $121.75, unchanged in extended trading as of 18:43 ET, while WSM closed at $228.09, down 0.8% on the day.^[1]

That is not proof of a durable rotation into software. It is a useful tell: the market is currently rewarding evidence of workload expansion more readily than evidence of consumer resilience. RH’s recent path is especially revealing: it closed at $121.75 on September 30 versus $200.80 a year earlier, while WSM closed at $228.09 versus $195.30 a year earlier.^[2] The same broad home category can produce very different outcomes when execution, pricing power and investor expectations diverge.

Why geopolitics matters even when crude flows recover

The latest Middle East signal is not a clean supply normalization. CNBC reported that crude moving through the Strait of Hormuz had returned to a seven-day average of 13.5 million barrels per day—roughly a prewar baseline—but refined-product shipments were only 677,000 barrels per day versus 3.6 million before the war. Combined crude and product flows were about 80% of the prewar baseline.^[3]

That distinction matters for this scope. A crude market that looks more functional can coexist with a fuel market that is still expensive and operationally fragile. The same report said more than 70% of crude crossing Hormuz in August switched tankers near the UAE or Oman, and that roughly 40% of Gulf crude was bypassing the strait through pipelines, versus 17% before the war. Those are shock absorbers, but they depend on escorts, transshipment and infrastructure that can themselves be disrupted.^[3]

Reuters has separately reported that container-shipping rates could test record highs as the Iran-war fuel spike lifts costs, and that vessel traffic through the strait has fallen to only a handful of ships at points in September.^[4] The market implication is less about one oil-price print than about the persistence of an elevated logistics tax.

Software: demand becomes infrastructure

Datadog’s recent earnings-call evidence is consistent with the resilient side of the hypothesis. In Q2 FY2026, management cited a multiyear deal worth more than $30 million in total contract value with a major online media company, including consolidation of four commercial and internal tools. It also described large deployments in banking and health insurance, where observability, security and incident response are tied to uptime, compliance and customer experience.^[5]

Snowflake’s Q2 FY2027 results were even more direct. Product revenue grew 37% year over year for the third consecutive quarter, net-new customer additions grew 32%, and management said existing-customer expansion remained healthy.^[6] Snowflake’s stated mechanism is important: AI brings workloads onto the platform, first-party AI products expand the relationship, and AI activation increases consumption of the underlying data platform. That is a company claim, not an independent forecast, but it is the kind of operating evidence the market tends to distinguish from discretionary demand commentary.^[6]

The counterpoint is that consumption models are not immune to optimization. Snowflake management has acknowledged that customers can optimize usage and that the economics of proliferating agents need to be managed. The bull case therefore requires more than AI enthusiasm: new workloads must become durable production activity rather than a temporary burst of experimentation.

Home furnishings: demand can hold, but the cost curve is louder

Williams-Sonoma’s Q2 FY2026 call offers the strongest evidence against a simplistic “consumer is breaking” narrative. Management said e-commerce comparable sales rose 6.5% and retail 5.5% while the home-furnishings industry was essentially flat; growth therefore came largely from share gains, alongside higher full-price selling. But gross margin fell about 160 basis points, with merchandise margin down about 230 basis points as tariffs flowed through cost of goods sold.^[7]

The company also said Q2 was the peak of tariff pressure and that supply-chain efficiencies and occupancy leverage offset part of the hit. It raised its annual comparable-brand revenue outlook to 4%–6.5% and operating-margin outlook to 17.8%–18.2%, while reporting a $200 million tariff refund, of which $174 million was recognized in Q2.^[7] That is resilient demand, but it is not frictionless earnings growth: policy relief and operational offsets are doing meaningful work.

RH presents a more complicated version. Its Q2 FY2026 materials said revenue increased 2.6% to $922.2 million and adjusted EBITDA included a $55.1 million tariff benefit.^[8] On the call, management described a fifth year of a weak housing market and acknowledged that higher costs and tariffs will affect the business, even as it argues that new formats and brand expansion can create its own market.^[9] The stock’s weaker one-year path suggests investors remain unconvinced that brand strength alone can neutralize housing, tariff and balance-sheet uncertainty.

The remaining home names—ETH, LZB, LESL and TPX—should be treated as a dispersion basket rather than a single read on demand. The September 30 snapshot showed ETH at $25.49, down 0.8%, LZB at $29.64, essentially flat, and LESL at $0.16 after a 26.5% regular-session decline but a 23.4% rebound in extended trading. TPX data in the feed was stale, showing a February 2025 timestamp rather than a current September 2026 observation, so it should not be used as a current market signal.^[1]

What would confirm or break the thesis

The constructive case requires three things to remain true:

1. Software workloads keep converting into production consumption. Watch Snowflake’s net retention, large-customer additions and AI usage, and Datadog’s platform expansion and security adoption—not just headline AI announcements.
2. Physical-goods companies keep gaining share without surrendering all of it to costs. Williams-Sonoma’s full-price selling, freight efficiency and margin trajectory are more informative than a single monthly sales print. RH needs its design, gallery and brand-expansion investments to translate into demand without another large tariff or sourcing shock.
3. Fuel normalization reaches refined products and freight. Crude flows alone are insufficient. Diesel, bunker fuel, insurance, vessel traffic and transshipment costs are the more relevant indicators for furniture and other bulky goods.

The bearish case would be confirmed by a combination of weaker enterprise consumption and persistent logistics inflation. That would remove the current valuation support from software while compressing the margins of home-furnishings companies at the same time. The base case, for now, is more selective: mission-critical software has a better demand cushion, while consumer businesses can remain operationally resilient but are more exposed to the cost and policy path.

This is market research, not investment advice. The evidence is changing quickly, especially around Hormuz, sanctions, fuel shipments and tariff treatment.