A Split Earnings Deck: AI Software Accelerates While Rates Pressure the Rest

The question this earnings season poses is not whether demand is healthy broadly — it is where demand is healthy and where the pressure lives. The research hypothesis asks whether earnings growth and resilient demand can carry DDOG, SNOW, RH, WSM, ETH, LZB, LESL, and TPX over the next twelve months. The honest read, two weeks into Q3 reporting season, is a split deck: the AI-native software names are reporting accelerating revenue and passing through demand intact, while the consumer and home-furnishings cohort faces a macro backdrop that is genuinely tightening.

The market structure backdrop: more volume, thinner liquidity

2026 has been a year of record issuance sitting inside a market that is structurally thinner. Year-to-date through mid-September, 109 US IPOs raised $146.5 billion — a record, and $71 billion of that excluding SpaceX ^[1]. Yet the third quarter itself stumbled: 31 listings raised $34.9 billion, and that headline figure leans almost entirely on SK hynix’s $26.5 billion offering. Strip the mega deal out and the quarter raised just $8.4 billion across the rest of the calendar ^[2].

That disconnect is the defining market-structure story of 2026. Realized trading volume is at records — average consolidated volume reached 19.1 billion shares year-to-date — but displayed liquidity is thinner, spreads wider, and average trade sizes smaller ^[3]. Concentration, passive flows, and heavy retail participation have transformed how order flow behaves ^[3]. It is a market where the on-ramp for new listings is wide open but pricing power is deal-specific.

Capital return tells the same tale from the corporate side. Dividend reductions hit 19% of announcements in Q3 2026 — the highest proportion of cuts in years — as buybacks slowed alongside mounting macro risk ^[3]. Meanwhile, the Treasury moved to support long-end liquidity, doubling the size of its buyback operations for the 10-to-20 and 20-to-30 year sectors starting September 9 ^[3]. None of that is a growth signal; it is a liquidity-support signal.

The IPO calendar: open, but selective

The fall pickup that looked inevitable in August has proved narrower than hoped. This week (through September 25) delivered one IPO and two SPACs, with five IPOs postponed — highlighted by Bamboo Insurance (BMB), which had been set to raise $665 million, pulling its deal ^[4]. Holtec Nuclear withdrew its $825 million offering outright ^[4]. Anthropic’s listing slid a second time to November ^[1].

The week ahead tests the theme with two sizable names: Oura plans the largest consumer-discretionary IPO in years, and PE-backed Accelevation acts as the read-through for the AI-infrastructure demand that shows up in software earnings ^[1].

The software side: the evidence says acceleration

Against that backdrop, the AI-native software names in this scope are the side of the deck where the evidence runs overwhelmingly in one direction. Snowflake’s most recent quarter (September 2) reported product revenue growth accelerating to 37% year over year — its third straight quarter of acceleration — with net new customer additions up 32% and AI revenue stepping up meaningfully across a broadening portfolio ^[5]. Datadog’s quarter (August 6) was similar: revenue of $1.12 billion, up 36% year over year and above the high end of guidance, with growth accelerating across nearly every cohort ^[6]. Notably, Datadog’s non-AI customer growth also accelerated to the high-20s percent — evidence that rising usage is broad-based, not a single-winner AI story ^[6].

The market has rewarded that. DDOG closed at $268.13 after a 4.4% gain on September 25, up sharply from the mid-$220s a month earlier ^[7]. SNOW closed at $335.94, near the top of its 120-day range ^[7].

The open question on this side is the one Datadog and Snowflake can’t answer for themselves: whether the AI capex that is driving their consumption stays funded. That is precisely the variable the IPO market is now repricing — Renaissance Capital points to AI-spending concerns, a 19-year high in bond yields, and resumed rate hikes as the forces that pulled the fall pickup below expectations ^[2]. For the software names, the tailwind is real; the risk is that it is coupled to an infrastructure buildout that could slow if rates keep rising.

The consumer side: the evidence says pressure

The other half of this deck looks nothing like the first. RH closed at $124.17 after a 2.5% gain on September 25 ^[7], WSM at $231.90 ^[7], LZB at $29.88 ^[7], ETH at $25.72 ^[7], and TPX at $65.81 ^[7]. These are names whose forward earnings depend far more on housing turnover, mortgage rates, and discretionary spend than on AI infrastructure — and the macro backdrop of resumed rate hikes and 19-year-high bond yields is a headwind for that model, not a tailwind.

The sharpest data point in the group is LESL. The pool-care retailer closed at $0.29 after a 13.8% drop on September 25, and it has collapsed over the past four months from roughly the $7–$8 level in June to under a dollar ^[8]. The current price implies a market that has priced in significant restructuring distress, and every subsequent print for this name is binary against a survivable cash position.

The distinction among the five consumer names is worth making explicitly rather than lumping them into one basket: LESL is a distressed-debt-vs-equity question, TPX and LZB are steady compounders sensitive to housing and input costs, WSM is a premium brand with proven pricing power, and RH is a high-leverage, high-beta play on the luxury housing cycle — RH had already run from roughly $124 to $183 by late June before retracing back near $124 now ^[9]. “Basket” is the wrong mental model; the risk profile differs name by name.

The deck at a glance

A few dates carry the “estimated” confidence flag, and ETH and TPX have no confirmed date yet — treat those as unconfirmed rather than fixed (^[10]).

What to watch next

1. DDOG (Nov 5) and SNOW (Dec 2) are the proof points for the acceleration thesis. The question is not whether they beat; it is whether guidance keeps implying AI-driven consumption accelerates through fiscal 2027.
2. The IPO taper is a leading indicator for the software names. If Oura and Accelevation price well and Anthropic finally lands in November, AI infrastructure demand is confirmed. If the fall taper continues, treat it as an early warning on the very capex these software names depend on.
3. LESL’s next print decides the viability question. Above ~$0.30 territory the equity is being priced for restructuring; the December 1 report will determine whether there is a path that keeps the equity in the picture.
4. Rates and housing, not earnings, steer the consumer third. WSM, RH, TPX, LZB, and ETH deliver against whatever the 10-year does; the resumed-hike regime from August is the dominant input.
5. Capital-return trends. With dividend cuts at a multi-year high and buybacks cooling, watch whether the more stable names (WSM, TPX, LZB) defend their return-of-capital programs — that is the clearest signal of management confidence through a rising-rate quarter.

The disciplined frame for this deck is a split thesis, not a single one. Strong evidence supports earnings growth and resilient demand for the AI-native software names, provided their infrastructure driver stays funded. The consumer cohort is a genuinely different question, trading on housing turnover and the rate path, with LESL as the standalone distress case. Weighing the two sides honestly is what separates a research read here from a guess.