The market’s AI question has changed

The latest geopolitical shock is not simply a risk-off episode. It is a test of whether earnings growth and resilient demand can support high-growth technology when the cost of energy and capital is moving against it.

On Friday, October 16, Brent crude touched $110 a barrel as attacks around the Strait of Hormuz intensified and Russia-linked supply risks persisted. U.S. crude finished above $100. The Nasdaq-100 fell 2.5%, while the Philadelphia semiconductor index lost 3.8% for an eighth consecutive session. The market’s message was unusually specific: demand may be real, but the funding and operating assumptions behind the AI buildout are no longer being treated as frictionless.^[1]

What changed: geopolitics moved into the cost of growth

The immediate catalyst is the energy channel. Reporting on the week’s move linked the oil spike to the closure of the Strait of Hormuz after the U.S.-Israel attack on Iran, attacks on commercial shipping, and damage to Russian refining infrastructure in the Ukraine war. Refinery capacity in the U.S. and Europe was already described as near its limits for diesel production, leaving less short-run flexibility if supply is disrupted further.^[1]

That matters for growth stocks in two ways. First, higher fuel costs can keep inflation elevated and make another Federal Reserve rate increase more likely. Second, the companies most exposed to the AI capital cycle are spending and borrowing against long-lived infrastructure while the discount rate is becoming less forgiving. NBC reported that the Nasdaq-100’s selloff erased more than $1 trillion in value and identified ballooning AI debt, high rates and oil as overlapping concerns.^[1]

The geopolitical risk is also becoming more physical for the technology system itself. An NBC analysis reported at least 27 attacks affecting data centers across five countries, reinforcing that power, connectivity and facilities are strategic infrastructure rather than invisible back-office inputs.^[2]

The company evidence does not say that AI demand has disappeared

The strongest counterpoint to the selloff is operating evidence from the software platforms.

Datadog’s latest available transcript points to broad customer demand rather than an AI-only spike. Management said customers continued moving forward with AI activity, that more than 6,500 customers were sending data from at least one AI integration, and that those customers represented about 80% of annual recurring revenue. The same call described GPU monitoring, AI-stack observability and security products aimed at making AI workloads more reliable and efficient.^[3]

Snowflake’s latest available call was similarly constructive. Q2 fiscal 2027 product revenue grew 37% year over year, the company reported a third consecutive quarter of acceleration, and management said AI was bringing workloads onto the platform, increasing use of its core data services and expanding its role in enterprise workflows. It also reported a 15% non-GAAP operating margin for the quarter.^[4]

Those statements support the first half of the hypothesis: resilient demand and product adoption can support revenue growth. They do not settle the second half: whether the economic value of that growth will outrun energy costs, financing costs and the capital required to build capacity.

The market is separating platforms from infrastructure-heavy exposure

Friday’s tape made that distinction visible. In the requested scope, DDOG fell 7.2% on Friday’s regular session close, SNOW fell 8.4%, and CRWV fell 11.9%; their post-market prints remained below those closes at roughly 19:57 ET. By contrast, AN rose 1.2%, while WSM was down less than 0.5%. RH fell 4.1%, ETH fell 1.7%, and LZB fell 1.3% at the regular close.^[5]

These are observations, not proof of a single causal chain. The contemporaneous reporting does, however, connect the sharp technology selloff to oil, rates and renewed skepticism about AI financing. CoreWeave was identified as one of the two worst performers in the Nasdaq-100 that day, and its shares were reported to have fallen more than 25% over the prior month.^[1]

The implication is not that every software company has the same exposure. DDOG and SNOW monetize data, observability, governance and application workloads; CRWV is more directly tied to the cost, utilization and financing of compute infrastructure. A durable demand story can therefore coexist with a falling multiple if investors decide the required capital is too expensive or the payback period is too long.

For the consumer names in scope, the geopolitical transmission is different. Higher fuel prices, tariffs and household-budget pressure can challenge discretionary demand even when headline retail spending remains positive. September U.S. retail sales reportedly rose 0.8% against a 0.3% forecast, but contemporary analysis described the gain as occurring alongside pressure from gasoline, tariffs, food costs and benefit reductions.^[6] RH’s latest reported quarter also illustrates why margin—not only revenue—is the key question: the company reported $55.1 million, or 600 basis points, of tariff benefit in adjusted EBITDA.^[6]

What would confirm or weaken the growth thesis

The bullish case requires more than continued AI enthusiasm. It requires evidence that customers are converting experimentation into production workloads, that usage growth is translating into durable revenue, and that platform companies can expand margins while infrastructure and energy costs rise.

The opposing case would show up first in the operating details: slower consumption, longer sales cycles, rising customer concentration, weaker utilization of compute capacity, more debt-funded expansion, or guidance that assumes cheaper money and calmer energy markets than the geopolitical backdrop can provide.

For the consumer group, the key distinction is between affluent or necessity-supported demand and broad-based discretionary resilience. A strong aggregate retail-sales number can coexist with weaker furniture, home-improvement and auto demand if fuel and food absorb more of household income.

What to watch next

1. Hormuz shipping and Russian refining: Further attacks, a prolonged closure, or evidence that alternative supply cannot replace disrupted barrels would keep the energy and inflation channel active. A credible de-escalation would test how much of the technology selloff was geopolitical risk premium versus a deeper AI-financing reset.
2. Rates and inflation expectations: The market needs to see whether higher oil feeds into broader inflation and changes the expected path of policy. The key risk to long-duration growth is not one expensive week of fuel; it is a higher-for-longer discount rate.
3. CRWV and data-center economics: Watch utilization, customer commitments, power availability, debt costs and cash generation. These metrics will distinguish a temporary valuation shock from a structural capital-cycle problem.
4. DDOG and SNOW consumption: Management’s claims about AI adoption are meaningful only if they continue to produce expanding platform usage, durable retention and operating leverage in subsequent reports.
5. RH, WSM, AN and LZB demand quality: Look for traffic, order trends, cancellations, promotions and tariff pass-through. The market will care less about nominal sales than about whether margins and household affordability are holding together.

The base-rate conclusion is balanced: the evidence still supports real AI demand, but the geopolitical shock has changed the hurdle. Growth can survive expensive energy and capital if workload adoption is strong and infrastructure is disciplined. It cannot rely indefinitely on the assumption that funding, power and logistics will remain cheap and available.