The AI and demand thesis is being tested on two clocks

The Nov. 4 midterm result created an unusual split in the market signal: investors welcomed greater political clarity broadly, but debated whether the next Congress could make data-center development more difficult even as AI usage keeps growing. In the same session, Datadog’s earnings beat could not offset a customer-specific Q4 usage assumption. These are related risk factors, but not the same one: one concerns demand conversion, the other the ability and cost to supply compute.

That distinction matters for the hypothesis that earnings growth and resilient demand can support DDOG, SNOW, CRWV, AN, RH, WSM, ETH and LZB over the next year. The evidence so far is uneven rather than uniformly confirming or rejecting it.

A policy win for clarity, not necessarily for new capacity

After Democrats won control of both chambers, Reuters reported a broad global-equity rally alongside a sharp repricing in technology shares. Investors saw less immediate political uncertainty, but they also weighed whether opponents of new data centers would gain influence. Reuters coverage described the counterintuitive possibility that tighter constraints on new facilities could benefit established technology platforms by making scarce compute more valuable, while also raising a real execution risk for projects that need new power and permitting.^[1]

This is the geopolitical-policy channel for the group: local opposition, energy and grid constraints, and federal oversight may affect the timeline and economics of capacity additions. It is not proof that AI demand is weakening. Nor does a political shift alone establish that any specific bill or permitting restriction will pass. The path from election outcome to enacted policy remains uncertain.

Datadog’s print: strong quarter, narrow customer warning

Datadog reported Q3 revenue of $1.20 billion, up 35% year over year, and highlighted growth in larger customers. Yet coverage of its outlook said Q4 revenue guidance assumes its largest customer—an undisclosed AI company—reduces usage to a minimum commitment. The company’s shares closed at $188.72, down 14.4% on Nov. 4; the after-hours snapshot was $189.40 at 16:27 ET, about 0.36% above the regular close. The immediate market tell was not that AI usage broadly fell, but that a single large customer could change near-term consumption math.^[2][3][4]

That is a useful stress test for consumption-linked software: a headline beat and overall growth can coexist with customer-level volatility. The next evidence should be whether the assumed usage reduction is isolated, whether other AI customers continue expanding, and whether management’s following-quarter outlook recovers from the conservative baseline. One customer does not define the whole AI market; it does show how quickly concentrated usage can affect expectations.

Snowflake and CoreWeave: workloads are growing, but capital and capacity remain the check

Snowflake’s latest reported quarter showed product revenue growth accelerating to 37% year over year in Q2 FY2027, with management citing strength in core data-platform usage and AI revenue. That is constructive evidence for real workloads, not just announced AI pilots. Its next test is whether consumption and expansion persist as customers scale applications into production.^[5]

CoreWeave’s Q2 FY2026 update reported revenue of $2.6 billion, up 112% year over year, and a $104.2 billion revenue backlog, up 246%. Management said demand remained strong across its customer base and described ongoing capacity deployment and financing work. Backlog supports visibility, but does not eliminate buildout, delivery, customer concentration, power, or financing risks. The election-linked data-center debate adds a further timing variable to an already capital-intensive model.^[6]

The policy debate therefore cuts both ways. A harder-to-expand supply base could reinforce scarcity and pricing for existing operators; delays or higher energy and construction costs could also slow the conversion of commitments into revenue and cash flow. Those are scenarios to monitor, not an established outcome.

Consumer names give a different kind of demand evidence

The current basket also includes home furnishings and auto retail, where the relevant signals are orders, margins, financing and consumer budgets—not AI compute. La-Z-Boy’s fiscal Q1 FY2027 release showed retail written same-store sales up 3% and delivered retail sales up 10%, while consolidated sales fell 3% (or 1% excluding the divested casegoods business) and management characterized the consumer backdrop as challenging. Its retail strength and consolidated weakness can both be true.^[7]

RH’s Q2 FY2026 result also needs normalization: revenue rose 2.6% to $922.2 million, but reported adjusted EBITDA margin included $55.1 million of tariff-refund benefit, worth 600 basis points. RH said it expected $50 million of unplanned supply-chain cost increases tied to higher oil prices amid the Middle East conflict, with tariff proceeds partly offsetting them. That is a direct example of geopolitics and trade policy affecting margins even when the demand story remains in motion.^[8]

AutoNation’s latest Q3 report offers a more resilient after-sales and financing picture alongside mixed headline performance: adjusted EPS was $5.11, while revenue was $7.01 billion and slightly below the year-ago level; the company cited record after-sales gross profit and growth in its finance portfolio. This is not evidence that discretionary demand is uniformly strong—it suggests businesses can offset softer vehicle volumes through other revenue streams.^[9]

Williams-Sonoma’s latest available results in this research pass were Q2 FY2026: comparable brand revenue grew 6.2%, and the company raised its full-year outlook. That earlier report is a positive but lagged data point, not a read-through on demand after the latest election developments.^[10]

ETH: policy sensitivity is visible, direction is not settled

Reuters reported that cryptocurrencies recovered from overnight losses as traders weighed the implications of the election outcome for digital-asset legislation. The short-lived swings underscore that ETH is exposed to the regulatory interpretation of political change as well as broader risk appetite. The evidence supports volatility and policy sensitivity; it does not establish a durable trend in either direction.^[11]

What would confirm or challenge the hypothesis?

The supportive case is that usage and revenue are growing in enterprise AI and cloud infrastructure, some consumer businesses have pockets of resilient demand, and operators with existing capacity could benefit if new supply faces more friction. The countercase is that a customer-level pullback can overwhelm a strong quarter, infrastructure commitments must still be delivered and financed, and household demand or margins may depend on temporary supports such as tariff refunds.

What to watch next

• AI usage quality: Datadog’s customer concentration and subsequent consumption trends; Snowflake’s product-revenue growth and expansion metrics.
• Capacity and policy: Data-center permitting, local opposition, power access, project schedules and any concrete congressional or state policy proposals. Separate enacted rules from political rhetoric.
• CoreWeave conversion: Backlog recognition, deployment pace, customer mix, cash generation and financing terms—not backlog size alone.
• Consumer confirmation: Upcoming company reports and order trends for WSM, RH and LZB; for AN, vehicle volumes alongside after-sales and finance performance.
• Geopolitical cost pass-through: Energy and freight costs, tariff changes, and whether companies can absorb or pass through those costs without damaging demand.
• ETH policy channel: Legislative proposals and regulatory actions, while distinguishing them from daily price swings and sentiment-driven reversals.

A constructive reading of the thesis requires sustained conversion of AI workloads into revenue and cash flow, plus consumer order resilience that survives cost and financing pressure. A more cautious reading gains force if reported usage slows, project delivery slips, or margins rely increasingly on temporary offsets. The Nov. 4 session has not resolved that balance; it has made the monitoring points more specific.