From roughly 2010 to 2024, betting on the Nasdaq-100 and betting on software were the same trade. Overlay a chart of NDX against IGV (the iShares Expanded Tech-Software Sector ETF) and the two lines practically sit on top of each other. The correlation hovered above 0.95 for most of that stretch—near-perfect.

That relationship is now broken.

Over the past 18 months, the Nasdaq-100 has ripped higher while IGV has flatlined or, depending on where you draw the line, actively declined in relative terms. The spread between the two is the widest it has ever been. This isn't a minor wobble. It's a structural crack in the idea that "tech" is one trade.

Understanding why these two cousins are no longer speaking tells you more about where markets are heading than almost any other signal right now.

What NDX and IGV Actually Own

Before dissecting the break, it helps to know what you're comparing.

The Nasdaq-100 (NDX) tracks the 100 largest non-financial companies on the Nasdaq exchange. In theory, that's "tech." In practice, it's dominated by a handful of mega-caps: Apple, Microsoft, Nvidia, Amazon, Meta, Alphabet, Broadcom. These seven names alone account for roughly half the index's weight.

IGV is narrower. It holds about 120 pure-play software companies—think Salesforce, Adobe, ServiceNow, Palantir, Intuit, Cadence Design Systems. It excludes the semiconductor names, the hardware makers, and the advertising platforms that dominate NDX.

For most of a decade and a half, this distinction didn't matter. The rising tide of cloud adoption, SaaS subscription revenue, and low interest rates lifted everything. Software multiples expanded alongside hardware. When Microsoft rallied, it pulled both indices because it lives in both.

But something changed. And the change has a three-letter name.

The AI Wedge

The divergence began accelerating in late 2024 and has widened ever since. The catalyst is straightforward: AI spending flows to infrastructure first, and software second—if software gets any of it at all.

Think of the AI economy as a gold rush. The Nasdaq-100, with its heavy Nvidia, Broadcom, and Apple weighting, owns the people selling pickaxes, wagons, and denim jeans. IGV owns the people standing in the river panning for gold. The pickaxe sellers are getting rich regardless of who finds gold. The panners? They need to prove there's actually nuggets in their pan.

Here's how the money flows in practice:

• Phase 1 (2023–2025): The infrastructure build-out. Hyperscalers spent over $200 billion on GPUs, data centers, and networking equipment. Nvidia's revenue quintupled. Broadcom's custom ASIC business exploded. This money shows up directly in NDX.
• Phase 2 (2025–ongoing): The application layer. This is where software is supposed to benefit. AI copilots embedded in every SaaS product, higher pricing, new product categories. But so far, the revenue lift for traditional software companies has been incremental, not transformational.

The market is pricing Phase 1 as a certainty (because it already happened) and Phase 2 as a question mark. That question mark is the entire divergence.

The Software Revenue Problem

Let's get specific. When Salesforce added Einstein AI to its platform, did customers suddenly pay 50% more? No. Most got a modest upsell or saw it bundled in. When Adobe rolled out Firefly into Creative Cloud, did new subscriptions surge? Marginally. The problem is that AI features in existing SaaS products feel like expected upgrades, not revolutionary new value. Customers aren't opening their wallets wider—they're expecting more for the same price.

Meanwhile, a new class of AI-native startups is threatening to eat the lunch of legacy software entirely. Why pay $300/month for a traditional CRM when an AI agent can manage your pipeline for $50? The incumbents in IGV aren't just failing to capture the AI upside—some of them are on the wrong side of it.

The Multiple Compression Nobody Talks About

There's a second, less discussed force widening the gap: interest rate sensitivity.

Software companies—especially the high-growth, not-yet-profitable ones—are long-duration assets. Their value depends heavily on cash flows years into the future. When rates were near zero, the market was happy to pay 15x, 20x, even 30x revenue for a SaaS company growing at 25%.

Rates are no longer near zero. With the 10-year Treasury hovering around 4.5%, those future cash flows get discounted harder. Software multiples compressed from an average of ~15x forward revenue in 2021 to roughly 7–8x today. That's nearly a 50% haircut on valuation before the business even does anything wrong.

The mega-caps in NDX feel this less. Apple generates $100 billion in annual free cash flow. Nvidia is printing money today, not promising to print it in 2030. The Nasdaq-100 has shifted from a growth index to a profitable growth index, while IGV still contains plenty of companies burning cash to build market share.

In other words: NDX adapted to the rate environment. IGV didn't.

Is Software Dead? Not Exactly.

Here's the counter-argument, and it deserves serious consideration:

1. AI Revenue Is Coming—It's Just Late

Enterprise software adoption cycles are slow. Budgets are set annually. Procurement takes 6-12 months. If AI copilots genuinely increase worker productivity by 20-30% (and early data from Microsoft's Copilot and GitHub Copilot suggests this is in the ballpark), companies will pay for it. The revenue just hasn't shown up in quarterly earnings yet because we're still in the early adoption phase.

2. Valuation Matters Eventually

If NDX trades at historically high multiples (driven by Nvidia at 35x forward earnings and Apple at 30x) while IGV trades at multi-year lows, mean-reversion math favors software. You don't need software to "win" the AI race. You just need the gap to narrow. A 10% re-rating of IGV combined with a 5% pullback in NDX-heavy names would close a meaningful chunk of the spread.

3. The Acquisition Cycle

When large-cap tech companies are flush with cash (and they are), they go shopping. The most likely targets? Mid-cap software companies with AI-enhanced products and sticky customer bases—exactly the kind of company sitting inside IGV. M&A premiums could act as a floor under software valuations.

What This Means for Your Portfolio

The NDX-IGV divergence isn't just an interesting chart pattern. It's a real-time map of where the market thinks value is being created—and where it isn't.

If you hold a broad "tech" ETF like QQQ and think you're diversified within technology, you're not. You own a semiconductor and mega-cap advertising bet with some software on the side. Your exposure is heavily tilted toward the AI infrastructure winners.

That might be exactly what you want. But you should know it's what you own.

Three ways to think about positioning:

1. If you believe AI software monetization is coming: IGV at current valuations might be one of the better risk/reward setups in the market. You're buying proven businesses at compressed multiples, with a free option on AI revenue acceleration. The key is patience—this trade may take 12-18 months to work.

2. If you think traditional software is being disrupted: Stay overweight NDX and consider trimming pure software exposure. The companies building AI models and the infrastructure to run them may capture value that previously went to SaaS vendors. In this scenario, IGV doesn't recover its correlation to NDX—it structurally de-rates.

3. If you're unsure (the honest answer for most people): Own both, but understand you're making two distinct bets, not one. Rebalancing between them when the spread gets extreme in either direction is a disciplined approach.

The Bigger Picture

The NDX-IGV split is really a story about the maturation of the technology sector. For 15 years, "tech" was a monolith. Cloud, SaaS, mobile, social—everything went up together. Correlations were high because all these businesses rode the same macro wave: the digitization of everything.

AI has broken that wave into two. There's the physical layer (chips, servers, power, networking) and the application layer (software, services, interfaces). The physical layer has clear winners and is being valued accordingly. The application layer is still figuring out its business model.

History tells us that in every technology revolution, the infrastructure phase always leads. Railroads came before the businesses that used them. Fiber optic cable was laid before anyone figured out how to monetize the internet. AWS made money before the apps running on it did.

If that pattern holds, software's turn is coming. The question is whether the same software companies—the ones sitting inside IGV today—will be the ones to capture it, or whether a new generation of AI-native applications will take their place.

That's the trillion-dollar question this divergence is asking. And the answer will determine whether you want to buy the dip in IGV or let it keep sliding.