1. Why This Industry Exists
Every business runs on software it can't build itself: databases, CRM, security, workflow. Rather than buy a program once, companies now rent it monthly — like leasing a delivery fleet instead of buying trucks. Customers keep paying because switching breaks everything downstream. Portfolios hold it for durable, compounding, high-margin recurring revenue that grows faster than GDP.
2. What's Happening Right Now
What happened: IGV sits at 104.35, up +2.39% (1m), +19.52% (3m), +25.76% (6m) — trouncing SPY by +2.26 / +16.94 / +7.72 points. The 3-month spread is the story: software beat a nearly flat SPY (+2.58%) by ~17 points. Oracle posted record FY2026 results with cloud infrastructure revenue up 77% and remaining performance obligations growing $85 billion to $638 billion.
AltIndex flags Nebius (~$209) and Palo Alto Networks (~$375) as top-scored cloud names.
Why it happened:AI infrastructure spend is now flowing through the income statement, and September earnings and capex commentary will determine whether that translates into a stronger second half or a valuation reset. Backlogs are converting to revenue; multiples re-rated as rate-cut expectations lifted long-duration cash flows.
What it sets up: Next 4–8 weeks hinge on Q3 prints proving backlog converts to margin, not just bookings.
3. How the Money Works
Revenue is subscription: booked once, billed forever, with net revenue retention above 100% meaning existing customers spend more each year — a landlord who raises rent while tenants never leave. The two costs that decide profitability: sales & marketing (customer acquisition) and cloud hosting/compute. Scale helps enormously — code written once serves millions at near-zero marginal cost, so gross margins hit 75–85%. Great businesses (ServiceNow) show falling S&M as a share of revenue while retention rises; average ones buy growth. Net revenue retention is the gold standard.
4. The 4 Macro Drivers
Driver 1: Interest Rates / Discount Rates
Mechanism: Software profits arrive years out, so valuation is long-duration. Higher rates raise the discount rate, shrinking the present value of distant cash flows — multiples compress hardest here.
Now: Rate-cut expectations underpinned the 3-month +16.94pt outperformance vs SPY; falling discount rates re-inflated the priciest names.
2nd-order effect: Juniors watch the headline cut; the real move is the terminal-rate re-rating — where rates settle, not the next 25bp. Growth software leads when the market prices the plateau.
Threshold: 10-year Treasury breaking back above ~4.75% reverses the multiple tailwind.
Driver 2: Hyperscaler Capex Cycle
Mechanism: Cloud infrastructure demand is set by Amazon, Microsoft, Google and Oracle capex budgets. Their spend is downstream software's revenue.
Now:The sector entered an aggressive multi-year expansion driven by a projected $500–650 billion hyperscaler spending wave. Oracle's OCI is a direct beneficiary. 2nd-order effect: The trap is assuming capex = profit. Backlogs are growing faster than revenue — depreciation from GPU buildouts hits margins before revenue catches up. Watch the gap.
Threshold: A single hyperscaler trimming FY guidance signals cycle peak.
Driver 3: Enterprise IT Budget Growth
Mechanism: Software is sold from corporate IT budgets, which flex with GDP and CFO confidence. Weak demand → longer sales cycles, smaller seats.
Now: Budgets are prioritizing AI and security over general SaaS; consolidation onto platforms squeezes point solutions.
2nd-order effect: Most miss that budget reallocation matters more than budget size — dollars shift from legacy tools to AI copilots, killing mid-tier vendors even in an up market.
Threshold: Net new ARR growth decelerating two straight quarters across bellwethers.
Driver 4: AI Monetization
Mechanism: AI enters both as a cost (compute/inference bills) and a demand driver (new products, usage-based upsell).
Now:Oracle hosts AI workloads for OpenAI, Meta, Nvidia and Microsoft simultaneously — AI compute is monetizing fastest at the infrastructure layer. 2nd-order effect: The obvious trade is buying AI winners; the subtle one is spotting whom inference costs crush — SaaS firms bundling AI free while paying rising compute bills see gross margins quietly erode.
Threshold: A marquee SaaS name cutting gross-margin guidance on AI costs.
5. Industry Map
| Sub-Industry | What It Does | Key Driver | Main Risk |
|---|---|---|---|
| Cloud Infrastructure (IaaS) | Rents compute, storage | Hyperscaler capex | Depreciation, price war |
| Application SaaS | Workflow, CRM tools | IT budgets | Seat saturation |
| Cybersecurity | Protects cloud, endpoints | Threat volume | Platform consolidation |
| Data/Analytics | Warehousing, observability | Data growth, AI | Usage-based volatility |
| Neoclouds (GPU rental) | AI compute capacity | AI training demand | Customer concentration |
The read: Infrastructure and security capture durable spend; usage-based and neocloud names carry the sharpest boom-bust risk.
6. Company Case Studies
Case Study 1: Oracle (ORCL) — Legacy database reborn as hyperscale AI landlord
Business: Sells databases and enterprise apps, now pivoting to OCI cloud compute rented to AI labs. Revenue is subscription plus consumption; key cost is data-center capex/depreciation. At scale, IaaS gross margins improve as capacity fills.
Moat: Decades of database lock-in plus enterprise trust; widening as it converts existing customers to OCI. Oracle shed its legacy reputation to become a hyper-scale AI infrastructure play.
Macro Linkage: Driver 2 (hyperscaler capex) hits hardest — but Oracle is the capex spender here, so its margins face the depreciation drag before revenue fully lands.
Watch: RPO backlog and OCI revenue growth. RPO grew to $638 billion; Q4 IaaS revenue rose 93%. Signals demand pipeline conversion.
Risk: Massive capex funded by debt; if AI demand stalls, stranded capacity crushes returns. Early sign: RPO growth decelerating while capex climbs.
Valuation: Forward EV/Sales and P/E; re-rated sharply on OCI. Fair-to-expensive — priced for flawless backlog conversion.
Case Study 2: CrowdStrike (CRWD) — Security platform selling into a non-discretionary budget
Business: Cloud-native endpoint and cloud security via subscription modules. Revenue recurs with high NRR; key cost is R&D and S&M. Single-agent architecture means each new module is near-pure margin at scale.
Moat: Data network effect — more endpoints means better threat detection, widening the lead. Consolidation tailwind as customers drop point tools.
Macro Linkage: Driver 3 (IT budgets) — security is the last line CFOs cut, so CRWD is defensive within software; also Driver 4, AI-driven threat detection.
Watch: Net new ARR and module-adoption rate (customers with 5+ modules). Signals land-and-expand health.
Risk: A major outage or breach denting trust; premium multiple leaves no room for error. Early sign: rising customer churn or gross-retention slippage.
Valuation: EV/Sales and EV/FCF; among the richest in software. Expensive — justified only if NRR and margins hold.
Case Study 3: ServiceNow (NOW) — The workflow operating system for the enterprise
Business:Managing large-scale cloud infrastructure enables continuous streamlining and automation of operations — that's ServiceNow's specialty. Subscription revenue, very high NRR; key cost S&M. Scale drives strong FCF margins. Moat: Deeply embedded workflow platform — ripping it out breaks IT operations. Widening via AI agents layered onto existing seats.
Macro Linkage: Driver 4 (AI monetization) — AI copilots drive premium-tier upsell; Driver 3, as budgets consolidate onto platforms benefits NOW.
Watch: Subscription revenue growth and Pro-Plus (AI) tier attach rate. Signals AI upsell traction.
Risk: Seat-based model saturates in a weak hiring market; growth decelerates. Early sign: cRPO growth slowing below 20%.
Valuation: Forward P/E and EV/FCF; premium to software peers. Fair — durable growth supports the multiple.
7. How to Value These Companies
Use EV/Sales and EV/FCF, not P/E — many reinvest profits away, so earnings understate value; recurring revenue and cash generation are the true engine. Rule of 40 (growth + margin ≥ 40%) screens quality. Typical EV/Sales: 6–20x depending on growth. The most common junior mistake: paying a high-growth multiple on a decelerating name — the multiple compresses faster than revenue grows, and you lose both ways.
8. KPIs That Actually Matter
| KPI | What It Signals | Why It Beats EPS | Benchmark |
|---|---|---|---|
| Net Revenue Retention | Existing-customer expansion | Captures stickiness EPS hides | >110% strong |
| Remaining Performance Obligations | Contracted future revenue | Forward-looking vs backward EPS | Growth >20% |
| Rule of 40 | Growth-profit balance | Profit alone misses growth | ≥40% |
| Gross Margin | Pricing power, compute cost | Reveals AI cost drag | 75–85% |
| Net New ARR | Fresh demand momentum | EPS lags bookings | Accelerating |
| FCF Margin | Real cash economics | Beats non-cash-distorted EPS | >25% |
The read: Retention and RPO tell you tomorrow's revenue; EPS only tells you yesterday's accounting.
9. Risk Map
Risk 1: Multiple Compression from Rising Rates
Long-duration cash flows get discounted harder when rates rise, hammering the priciest names first. Transmission: discount rate up → present value down → multiple contracts even with flat fundamentals. Precedent: 2022's rate shock cut software multiples in half despite growing revenue. Early warning: 10-year yield breaking higher while forward EV/Sales stays elevated — the setup for a violent de-rating.
Risk 2: AI Compute Cost Squeeze
SaaS firms bundling AI features pay rising inference bills while charging customers little extra. Transmission: compute cost up → gross margin down → Rule of 40 breaks → multiple compresses. No clean precedent yet — this is the new risk. Early warning: a bellwether cutting gross-margin guidance and blaming AI/hosting costs. Watch the gap between AI feature launches and actual monetization.
Risk 3: Platform Consolidation Killing Point Solutions
Customers standardize on fewer vendors, starving single-product firms. Transmission: budget reallocation → mid-tier ARR shrinks → growth stalls → valuation collapses. Precedent: numerous 2023–24 point-solution SaaS names crushed as buyers consolidated onto Microsoft and platform vendors. Early warning: decelerating net new logos at smaller vendors while platforms report accelerating module attach — the money migrating in real time.
Risk 4: Neocloud / Customer-Concentration Bust
GPU renters live on a few mega-contracts and heavy debt-funded capex. Transmission: one customer leaves or AI demand cools → utilization drops → depreciation strands returns → equity impairs. Precedent: telecom/dot-com overbuild busts follow the same capex-ahead-of-demand pattern. Early warning: contracted backlog growth stalling while capex and debt keep climbing, or a top-three customer non-renewing.
10. Cycle Playbook
| Phase | Sector Behaviour | Why | What to Own |
|---|---|---|---|
| Early Expansion | Outperforms | Rates fall, budgets thaw | High-growth SaaS |
| Mid Cycle | Leads | Budgets flush, capex peaks | Infrastructure, platforms |
| Late Cycle | Volatile | Multiples stretched | Security, profitable names |
| Recession | Lags then defensive | Budgets cut | Mission-critical, high-NRR |
| Recovery | Rebounds first | Multiples re-rate early | Beaten-down growth |
Now: Early-to-mid cycle — falling discount rates and the AI capex wave are driving software's ~17-point 3-month lead over SPY. Favor infrastructure and platforms while backlog still converts.
11. Structural Themes
Theme 1: AI Compute as the New Infrastructure Layer
AI has shifted from a software feature to a physical buildout. Artificial intelligence has outgrown its training phase to trigger the largest physical computing buildout in modern history. Accelerating because inference demand compounds as adoption spreads. Winners: infrastructure owners (Oracle, hyperscalers) and neoclouds with locked contracts; losers: SaaS firms eating compute costs. Position before consensus: own the toll-collectors, avoid the toll-payers with weak pricing.
Theme 2: Sovereign & Vertical AI Stacks
Governments and regulated industries demand data-resident, controlled AI. Palantir and Nebius partnered to deliver a complete sovereign AI stack to Palantir customers. Accelerating on data-sovereignty regulation and geopolitics. Winners: vertically integrated stacks and neoclouds offering isolation; losers: generic multi-tenant vendors. Position early: own the integrators bundling compute, security and compliance before the premium is priced in.
12. Portfolio Reference
| Factor | Value |
|---|---|
| S&P 500 weight | ~11–13% (software within Tech ~33%) |
| Typical dividend yield | ~0.5% (mostly reinvest) |
| Beta vs S&P 500 | ~1.2–1.4 |
| Overweight when | Rates falling, budgets expanding |
| Underweight when | Rates rising, multiples stretched |
| ETF | Focus | Expense Ratio |
|---|---|---|
| IGV | Software & cloud | 0.41% |
| WCLD | Pure-play SaaS | 0.45% |
| SKYY | Cloud computing | 0.60% |
13. Three Questions You Should Be Able to Answer
Q1: Why can a software company grow revenue yet destroy value?
A: Because growth bought with heavy S&M or debt-funded capex doesn't compound if retention is weak. If NRR sits below 100%, the company loses more from churn than it gains from upsell, so each new customer just replaces a lost one. Value comes from durable, expanding relationships — not gross bookings. A neocloud growing revenue 80% while burning cash and depending on one customer is fragile, not valuable.
Q2: How do falling rates and rising AI costs pull software in opposite directions?
A: Falling rates lower the discount rate, inflating long-duration multiples — the tailwind behind software's 17-point lead over SPY this quarter. But rising AI inference costs compress gross margins, breaking the Rule of 40. So a name can see its multiple expand on rates while its fundamentals quietly deteriorate on compute bills. The two forces net out per-name: infrastructure toll-collectors win both; margin-squeezed SaaS wins one, loses the other.
Q3: Bull vs bear on software today, given the macro?
A: Bull: AI spend is flowing through the income statement, rates are easing, and backlogs like Oracle's $638B RPO signal years of revenue. Bear: backlogs are growing faster than revenue, meaning depreciation hits margins before cash arrives, and valuations already price perfection. What flips it: Q3 prints showing backlog converting to expanding margins confirms the bull; any capex-guidance cut confirms the bear.
Research via live web search | Saturday, September 19, 2026 | Industry Rotation Series
⚠️ Disclaimer: This report is AI-generated and is intended solely for self-educational and informational purposes. Nothing in this report constitutes investment advice, a solicitation to buy or sell any security, or a recommendation of any kind. All market data, analysis, and investment ideas presented here are for learning purposes only. Past performance is not indicative of future results. Always conduct your own research and consult a qualified financial advisor before making any investment decisions.