The AI buildout is meeting an energy-and-credit test

The market’s latest signal is not that enterprise AI demand has vanished. It is that investors are separating software consumption from the physical infrastructure required to serve it.

On October 13, CoreWeave (CRWV) fell 3.05% in regular trading to $77.81, while Snowflake (SNOW) declined 0.72% and Datadog (DDOG) fell 1.37%. LZB was the strongest name in the supplied scope, rising 2.50%; WSM gained 0.47%, while AN was roughly flat and RH fell 0.75%.^[1] The contrast matters because the day’s geopolitical backdrop centered on the Strait of Hormuz, LNG disruption and renewed sanctions pressure—risks that travel first through energy, shipping, financing conditions and data-center operating costs.^[2][3]

The market tell: demand is holding, but the cost of capacity is under scrutiny

CRWV was the clearest stress signal. Its regular-session decline was the largest among the eight names in scope, and its dollar volume exceeded $2.6 billion.^[1] Same-day coverage linked the move to debt concerns, an insider-sale notice and broader pressure on data-center and AI-infrastructure names. Those reports are market explanations rather than company-confirmed changes in demand, so they should be treated as signals of positioning and risk perception, not proof that backlog has deteriorated.^[4]

The geopolitical channel is unusually direct for an AI-infrastructure trade. Current reporting describes continuing uncertainty around Hormuz traffic, Qatari LNG shipments and the possibility of additional sanctions connected to Iran. Reuters has also reported that more Qatari LNG cargoes were transiting the strait despite ongoing risks, illustrating the market’s central tension: flows can recover while the route remains vulnerable to interruption.^[3] A prolonged disruption would not automatically invalidate AI demand, but it could raise power, cooling, logistics and project-financing costs at the exact point when infrastructure providers are expanding fastest.

The bull case is still visible in company evidence

CoreWeave’s latest available earnings-call evidence argues against a simple “AI demand is breaking” interpretation. In its Q2 FY2026 call, management said demand remained strong across its customer base, that capacity was scarce, and that it had raised prices across SKUs by approximately 25% in July. It also said margin improvement preceded those pricing changes.^[5]

The same call described more than $400 million of annual recurring revenue from storage, CPU, networking and software alongside the GPU business, and said the company was reducing its weighted average cost of capital while securing financing for its growth plan.^[5] That is the constructive version of the story: demand is real, pricing is not yet collapsing, and the platform is broadening beyond rented accelerators.

Snowflake provides a less energy-intensive comparison. In its Q2 FY2027 call, management said AI was adding workloads, accelerating core-platform consumption and expanding the company’s role in the enterprise “agentic” stack. It also reported a 15% non-GAAP operating margin, up more than 400 basis points year over year.^[6] SNOW’s modest decline on October 13 therefore reads less like a direct demand verdict and more like a reminder that even stronger software fundamentals remain exposed to valuation, rates and broad technology positioning.

What the geopolitical shock can and cannot tell us

It can tell us that physical bottlenecks are becoming a larger part of the AI investment debate. A data-center operator can have contracted demand and still face a difficult equity-market path if power availability, construction schedules, equipment delivery, interest expense or refinancing spreads move against it.

It cannot, by itself, tell us that enterprise AI spending is slowing. The available company evidence points in the other direction for CRWV and SNOW: strong customer commitments, expanding use cases and continued platform adoption.^[5][6] The more balanced conclusion is that the market is testing the conversion of demand into durable, financeable earnings.

That distinction also helps explain why the supplied scope is diverging. CRWV is the most exposed to capital intensity and energy logistics. DDOG and SNOW monetize software and data workloads with less direct exposure to electricity and construction risk. AN, RH, WSM and LZB are more useful as demand cross-checks than as clean geopolitical hedges: their performance reflects company-specific and consumer factors as well as the macro backdrop. On October 13, that cross-section did not show a uniform collapse in demand; it showed a market assigning different risk premiums to different business models.^[1]

What to watch next

1. CRWV financing and delivery metrics: Watch interest expense, refinancing terms, data-center commissioning, active capacity and operating-margin progression. The key question is whether backlog converts into cash earnings without a disproportionate rise in leverage or project cost.
2. Energy and shipping normalization: Track verified changes in Hormuz traffic, LNG availability, tanker insurance and freight rates. A recovery in flows would reduce the physical-cost tail risk; renewed attacks or sanctions escalation would increase it.^[3][2]
3. SNOW and DDOG consumption indicators: New workloads, net expansion, usage growth and customer production deployments are cleaner tests of whether AI demand is broadening beyond infrastructure headlines. Snowflake’s recent call specifically tied AI adoption to higher core consumption, making that linkage worth monitoring.^[6]
4. The consumer names in scope: AN, RH, WSM and LZB can help distinguish a genuine broad demand slowdown from a concentrated repricing of AI infrastructure. Their next operating updates matter because geopolitical energy shocks can eventually reach household budgets through fuel, freight and imported-goods costs.

The base-rate reading is mixed but not bearish by default: demand evidence remains resilient, while the cost and financing sensitivity of the AI buildout is becoming impossible for markets to ignore. The next decisive signal will be whether infrastructure companies can preserve pricing and margins as geopolitical risk tests power availability and the cost of capital.