Written by Cassian Vance
The artificial intelligence investment narrative is undergoing a violent structural shift. For the past three years, capital has overwhelmingly favored the foundation model layer and semiconductor manufacturers, operating under the assumption that the creators of the intelligence engine would capture the lion’s share of its economic value. Meanwhile, the software sector was left for dead. In the first quarter of 2026, the iShares Expanded Tech-Software Sector ETF (IGV) plunged 24%—its worst quarter since 2008—as investors panicked that AI agents would destroy traditional SaaS seat-based pricing models in what was dubbed the “SaaSpocalypse.”
However, a starkly different reality is emerging: model companies are structurally disadvantaged in capturing the full efficiency and profit gains unlocked by their own technologies. Instead, it is the application and infrastructure software layer that is exhibiting the most durable, high-margin value capture.
As macro strategist Aelia Capitolina (@Areskapitalon) recently noted on X: “Model companies cannot actually capture all the efficiency and profit gains that AI unlocks by replacing labor. They can only capture the frontier model leadership premium, and achieving that premium requires piling on massive additional compute.”
This dynamic has created a historic mispricing in the software sector. While frontier model developers burn billions in a perpetual compute arms race, enterprise software platforms—the core constituents of the IGV ETF—are quietly monopolizing the margins and breaking out into semiconductor-like price action.
The Profit Paradox of Frontier Models
To understand the bull case for the IGV ETF, one must first examine the structural headwinds facing the model layer. The fundamental issue is that foundational AI models are incredibly capital intensive to train and maintain. Because the technological moat at the model layer requires constant, exponential capital expenditure, companies like OpenAI and Anthropic are effectively trapped in a cycle where they must immediately reinvest any revenue they generate back into compute just to remain relevant.
Consequently, model providers can only charge a “frontier premium” for access to the absolute bleeding edge of intelligence. The massive efficiency gains created when an enterprise replaces labor with an AI agent do not flow back to the model provider; they are retained by the enterprise or captured by the software platforms that facilitate, monitor, and secure those AI workflows. The intelligence layer is becoming commoditized, but the workflow layer is becoming indispensable.
The Software Sector Rotation: IGV Holdings Flex Pricing Power
The market is finally waking up to this reality. Over the past month, the IGV ETF has rebounded nearly 14%, snapping a brutal three-month decline. This recovery is not a dead-cat bounce; it is driven by undeniable fundamental strength from the ETF’s heaviest weightings, which are proving that AI is a massive tailwind for software, not a disruptor.
Consider the recent performance of IGV’s top constituents:
Oracle (ORCL – 9.6% Weight): The largest holding in the IGV ETF has surged over 35% in the past month. Oracle’s Q3 fiscal 2026 earnings revealed a 44% year-over-year surge in cloud revenue, driven directly by massive demand for its OCI AI infrastructure.
Palantir Technologies (PLTR – 7.3% Weight): Palantir recently reported Q1 2026 U.S. commercial revenue growth of 133% year-over-year, driven by insatiable demand for its Artificial Intelligence Platform (AIP). Overall revenue grew 85%, crushing consensus expectations.
Salesforce (CRM – 6.5% Weight): Despite fears that AI would destroy CRM seat licenses, Salesforce reported $41.5 billion in FY26 revenue (+10% YoY). More importantly, its Agentforce AI product reached $800 million in ARR, up 169% year-over-year, proving that enterprises will pay a premium for embedded, agentic AI workflows.
Datadog (DDOG – 2.3% Weight): On May 7, Datadog reported its first $1 billion revenue quarter (+32% YoY) and saw its stock surge 31% in a single day. The company noted that usage of its AI monitoring agents more than doubled, proving that as AI workloads scale, the need for software observability scales exponentially alongside them.
Why “Not All Software is Born Equal”
The initial “SaaSpocalypse” selloff treated all software as equally vulnerable to AI disruption. This was a critical error. As analysts at Patria Investments recently noted, “Application software… is the layer most directly in the line of fire. Developer tools and data infrastructure are not being disrupted by AI, they are being made more essential by it.”
The IGV ETF is heavily weighted toward these essential infrastructure and enterprise platforms. Companies like Microsoft (8.4%), Palo Alto Networks (5.9%), and CrowdStrike (4.8%) are not vulnerable point solutions; they possess deep proprietary data moats and high switching costs. You cannot simply replace CrowdStrike’s endpoint security network or ServiceNow’s IT service management workflows with a generic LLM wrapper.
Valuation and Risks
Despite the recent 14% bounce, the IGV ETF remains down roughly 11% year-to-date, trading at $91.15, well below its 52-week high of $117.99. This YTD underperformance presents a compelling entry point.
The primary risk to the IGV thesis is the reassertion of “duration math.” Software companies typically generate cash flows weighted heavily towards future periods, making their valuations exceptionally sensitive to changes in market discount rates and macroeconomic tightening. Furthermore, while enterprise and infrastructure software are protected, the ETF does hold some exposure to lower-tier application software that may genuinely suffer seat-compression from AI agents.
However, the aggregate financial profile of the top IGV holdings—characterized by accelerating revenue, expanding operating margins, and massive free cash flow generation—provides a strong buffer against these risks.
The Verdict: BUY
The IGV ETF represents the optimal vehicle for capturing the economic value of the AI revolution without taking on the severe capital expenditure risks of the model layer. The Q1 2026 “SaaSpocalypse” created a generational mispricing by conflating vulnerable point solutions with mission-critical enterprise infrastructure. As top holdings like Oracle, Palantir, and Datadog continue to post blowout earnings driven by AI adoption, the market is realizing that software is the true beneficiary of the AI supercycle. With the ETF still down double-digits year-to-date, IGV is a high-conviction BUY for institutional investors looking to position for the next phase of the AI trade.
***
Disclaimer: This article is for informational purposes only and does not constitute financial advice. Investors should conduct their own due diligence before making any investment decisions.
