IGV vs. IGPT: The AI Infrastructure Layer Is Eating Software

The stark divergence in year-to-date returns between these two funds is more than a market quirk; it reads as a signal of a technological inflection point. While the iShares Global Clean Energy ETF (IGV) has fallen 8.279%, the iShares Artificial Intelligence & Big Data ETF (IGPT) has rallied 7.046%. This performance gap is a concrete manifestation of capital flowing from a maturing software application layer toward the foundational infrastructure of the next paradigm: artificial intelligence.

This pattern mirrors earlier technology S-curves. Consider the shift from personal computers to the internet, or from basic mobile phones to smartphones. In each case, the initial wave of adoption focused on end-user applications. But the steepest part of the growth curve-and the most durable returns-came when investment poured into the underlying infrastructure: the web protocols, the app stores, the cellular networks. AI appears to be following that same arc. The market is now pricing in the exponential adoption of AI, which requires massive compute power, specialized hardware, and data platforms. IGPT's gain signals that investors see this infrastructure layer as entering its steep, accelerating phase.

IGV's decline, by contrast, reflects a sector where growth is plateauing. Clean energy, while critical, faces a more linear adoption curve compared to the transformative potential of AI. The data supports this view: IGPT's net assets are a fraction of IGV's, but its higher expense ratio and turnover suggest a fund actively chasing a faster-moving, more volatile theme. The bottom line is that the performance gap is a leading indicator. It shows capital is shifting from established, slower-growth sectors to the fundamental rails of a new technological paradigm.

Mapping the Adoption Trajectory: Infrastructure vs. Application

The performance gap isn't just about sentiment; it's a direct reflection of where the exponential growth curves are pointing. For AI infrastructure, the fundamental driver is a projected 14% annual growth in data center power demand through 2030. This isn't a speculative forecast. It's a physical necessity. As AI workloads shift from training to real-time inference, the demand for distributed, high-performance compute creates a compounding need for power. This trajectory effectively means the worldwide data center power load will double over the next five years. That's the kind of scaling math that defines an infrastructure S-curve.

Compare that to the clean energy build-out. Here, the evidence shows the transition is moving from planning to execution at a blistering pace. In the first three quarters of 2025, the U.S. alone saw 20.4 gigawatts of new clean energy capacity announced. This surge is being fueled by a hard deadline: a law forcing companies to begin construction by July 2026 to claim tax credits. The result is a short-term manufacturing and construction race that has already powered the sector's recent rally.

The key difference lies in the nature of the growth. The clean energy surge is a powerful, concentrated wave of project development hitting a policy deadline. It's a linear acceleration in capacity additions. The AI infrastructure growth, however, is a non-linear, exponential demand for a fundamental resource-power-that must be built out in parallel. One is a sprint to meet a regulatory finish line; the other is a marathon of continuous, scaling investment in the physical substrate of the next computing paradigm.

This frames the investment case. The clean energy rally is real and driven by immediate execution. Yet its growth, while impressive, is currently bounded by a finite set of projects and policy timelines. The AI infrastructure story, by contrast, is about a foundational layer that must scale continuously to support an entire new software economy. The data center power metric shows that market is still in its early, steep phase. For an investor mapping the adoption trajectory, the exponential curve of AI infrastructure suggests a longer runway for capital deployment and value creation, even as clean energy executes its own powerful build-out.

Financial Metrics and Fund Structure: Assessing the Investment Vehicles

The stark performance divergence between these two funds is mirrored in their fundamental structures. IGV, the iShares Global Clean Energy ETF, is a much larger vehicle, with net assets of $6.36 billion compared to IGPT's $674 million. This scale difference is significant. IGV's size suggests it is a core holding for many institutional portfolios, while IGPT operates as a smaller, more specialized thematic fund. The cost of ownership also reflects this contrast. IGPT carries a higher annual expense ratio of 0.56% versus IGV's 0.39%. For a fund chasing a high-growth, volatile theme like AI infrastructure, this premium is a tangible drag on returns.

More telling is the index focus that defines each fund's holdings. IGV's underlying index is explicitly built around software industry and select companies from interactive home entertainment and interactive media and services. This is a pure-play on the application layer-the end-user software and services that run on top of infrastructure. In contrast, IGPT's index measures U.S.-traded stocks from the software industry but also includes companies with significant exposure to technologies or products that contribute to future software development through direct revenue. This broader mandate is the key differentiator. It allows IGPT to capture companies providing the fundamental tools and platforms that enable software, which includes the specialized hardware, data platforms, and cloud services that form the AI infrastructure layer.

This structural difference is the investment thesis in a nutshell. IGV is a bet on the software that uses the infrastructure. IGPT is a bet on the infrastructure itself. The higher expense ratio and smaller scale of IGPT are the costs of accessing this more concentrated, forward-looking segment. The fund's structure is designed to capture the economic characteristics of companies building the fundamental rails for the next computing paradigm, not just the applications that run on them. For an investor, this means IGPT offers a more direct, albeit costlier, lever on the exponential growth curve of AI infrastructure.

Catalysts, Risks, and What to Watch

The investment thesis here hinges on a clear technological S-curve: capital is shifting from the application layer to the foundational infrastructure. For this thesis to validate, we need to watch for concrete signs that the AI infrastructure build-out is accelerating. The primary catalyst is the pace of physical construction. Specifically, monitor the rate at which new data centers are being announced and brought online, alongside the speed of power grid upgrades to support them. The evidence points to a 14% annual growth in data center power demand through 2030, a physical necessity that will require massive, coordinated investment. When announcements of new facilities and power partnerships become more frequent and larger in scale, it will signal that the market is pricing in the exponential adoption of AI, not just the software.

A key risk to this thesis is that the clean energy rally, which has been spectacular, could be overextended. The sector has already seen a 47% return in 2025, a powerful move fueled by a hard policy deadline. The risk is that this creates a valuation disconnect. If the sector's momentum stalls or policy support faces a reversal, a correction could occur. For now, the catalyst is clear: watch for signs that the clean energy surge is hitting a plateau or that the policy-driven construction wave is beginning to slow. That would free up capital and attention for the longer-term AI infrastructure story.

The primary catalyst for a reversion in this narrative would be a major breakthrough in AI software that drives exponential productivity gains. If a new model or application dramatically lowers the cost of AI computation or unlocks entirely new use cases, it could accelerate the application layer's S-curve. This would shift the investment focus back toward the software companies that build the end-user tools, potentially challenging the infrastructure premium. The evidence suggests AI is about building intelligence itself, a paradigm shift that could consolidate the industry for a decade. But a true software breakthrough could reset the adoption trajectory.

For an investor, the forward view is about monitoring these competing adoption curves. The AI infrastructure thesis depends on sustained, physical investment in power and compute. The clean energy thesis depends on policy execution and project delivery. The key metrics to watch are the pace of data center announcements and power grid investments for the AI story, and the trajectory of new clean energy project approvals and construction starts for the energy story. The bottom line is that the market is currently pricing the exponential growth of AI infrastructure. The catalysts and risks outlined here will determine whether that pricing is justified or if the paradigm shift is still in its early, steep phase.