The Demand Thesis Is Working—Just Not Everywhere

The opening snapshot is a useful warning against treating “resilient demand” as a single market factor. Software-linked names are being rewarded, while AI infrastructure and consumer-sensitive companies are being repriced more skeptically. The result is a selective thesis: earnings growth can support parts of this group, but the market is demanding proof that demand converts into durable margins and cash flow.

The tape is separating scalable software from the physical and consumer edges

At the October 23 regular-session close, SPY fell 0.52%, QQQ was essentially flat at -0.01%, and DIA declined 1.24%. Within the specified basket, DDOG rose 4.62% and SNOW 3.65%, while CRWV fell 2.46%, WSM 3.19%, RH 1.94%, and LZB 1.38%. AN slipped 0.49%; ETH, the Ethan Allen ticker, gained 0.40%.^[1][2]

That dispersion matters more than the index move. It says investors are still willing to pay for software growth and AI-adjacent operating leverage, but they are less willing to assume that every dollar of AI or discretionary demand will arrive on schedule.

The six-month comparison makes the contrast sharper. From April 28 through October 23, SPY returned 6.78% and QQQ 7.12%. DDOG gained 104.98% and SNOW 128.28%, but CRWV lost 48.60% and AN lost 19.86%. The winners also carried much higher realized volatility: 74.07% for DDOG and 68.04% for SNOW versus 12.05% for SPY.^[3]

What the news is saying about software

Current reporting points to a change in the market’s interpretation of AI risk. Reuters reported that U.S. software stocks reached fresh 2026 highs as earnings expectations rose and fears of AI-led disruption faded. CNBC separately described a rotation from semiconductors toward software, while its cybersecurity coverage highlighted the way AI is increasing the perceived need for security spending.^[4]

That is a constructive backdrop for DDOG and SNOW, but it is not the same as proof of a twelve-month earnings outcome. The market can move from “AI will destroy software” to “AI expands software demand” faster than customer budgets, contract sizes, or margins can validate the shift. The relevant evidence to watch is therefore operational: durable consumption, renewal behavior, expansion within existing accounts, and whether AI-related workloads carry attractive economics.

Why CRWV is the important counterexample

CRWV is the cleanest stress test of the thesis. It is exposed to the same broad AI-investment narrative that benefits software, yet it has materially underperformed: down 48.60% over the six-month window and down 60.69% at its maximum drawdown in that period. Its annualized volatility was 95.12%, versus 12.05% for SPY.^[3]

That gap does not by itself prove that demand is weakening. It does show that demand is only one part of the infrastructure equation. Investors are also pricing power availability, build-out timing, capital intensity, financing conditions, customer concentration, and the possibility that capacity arrives before utilization does. A resilient-demand thesis must explain those constraints, not wave them away.

The consumer side is asking a different question

The consumer names in the scope—AN, RH, WSM, ETH, and LZB—are not a clean read-through on software demand. They are a test of whether households will commit to cars, furniture, and home-related purchases while financing costs remain meaningful and sentiment is weak.

The latest macro snapshot is mixed. Unemployment was 4.2%, real GDP growth was 2.1% year over year, and industrial production was growing 2.25%. But CPI inflation was 3.72%, the 10-year Treasury yield was 4.96%, consumer sentiment was 48.1—down 12.7% year over year—and the high-yield credit spread was 3.4%.^[5] This is not a recessionary snapshot in the supplied data, but it is also not a frictionless environment for discretionary purchases or highly levered expansion.

![A contemporary living-room setting connects weak sentiment and financing costs to the consumer names in the thesis.]

The practical implication is that “resilient demand” should be segmented. A software subscription that becomes embedded in a workflow may hold up better than a large-ticket purchase that can be postponed. Even within consumer, brand strength and exposure to housing turnover can produce very different outcomes.

A balanced test of the hypothesis

Evidence for the thesis:

• DDOG and SNOW have strongly outperformed both SPY and QQQ over six months.^[3]
• Current reporting describes improved sentiment toward software’s ability to benefit from, rather than be displaced by, AI.^[4]
• Employment and GDP remain positive in the latest macro snapshot.^[5]

Evidence against a broad version of the thesis:

• CRWV’s decline and extreme volatility show that AI demand does not eliminate infrastructure, financing, or utilization risk.^[3]
• AN, RH, WSM, and LZB are exposed to more cyclical, rate-sensitive spending, where demand can be deferred.^[2]
• High long-term yields and very weak consumer sentiment are meaningful headwinds to the most discretionary parts of the basket.^[5]
• The software winners have already experienced large gains and large drawdowns, so the market’s bar for continued execution is high.^[3]

The base case is therefore conditional: earnings growth and resilient demand can support selected names, especially where recurring software usage is translating into durable economics. The weaker version of the thesis is that all eight symbols benefit together. The current tape does not support that conclusion.

What to watch next

1. Software quality of growth: Look for evidence that AI-related usage expands retention, consumption, and gross profit rather than merely increasing activity.
2. Infrastructure utilization: For CRWV and the broader AI build-out, track customer commitments, power and capacity delivery, financing needs, and the pace at which new capacity becomes productive.
3. Consumer conversion: For AN, RH, WSM, ETH, and LZB, watch traffic, order trends, ticket size, promotional intensity, and cancellation behavior—not just headline revenue.
4. Rates and sentiment: A durable improvement in consumer sentiment alongside lower long yields would reduce pressure on discretionary demand; renewed yield or credit-spread widening would do the opposite.^[5]
5. Breadth of the next move: If leadership remains concentrated in a few software names while infrastructure and consumer exposure continue to lag, the market is validating selectivity, not a broad demand reacceleration.

This is a research framework, not a forecast or trading recommendation. The next decisive evidence should come from company-reported operating metrics and guidance, not from extrapolating one strong software session across the full basket.