vgt is barely budging despite software stocks struggling—a contradiction that forces a harder look at what investors are actually buying when they think they are buying “tech” or “AI. ” The S& P 500 Software Index is down 22% year-to-date, yet Vanguard Information Technology ETF (VGT) is down 8% year-to-date, a gap that highlights how concentration, index design, and exposure along the AI value chain can matter more than the headline category “software. ”
What is Vgt really tracking—and what does that exclude?
VGT tracks the MSCI US Investable Market Information Technology 25/50 Index. In practice, that makes VGT a pure-play U. S. information technology sector fund: no geographic diversification, no fixed income sleeve, and no defensive hedging. Practically 100% of the portfolio sits in information technology.
That narrow definition is the first hidden truth. When software sells off, many investors assume a “tech ETF” should move in lockstep. But VGT’s mandate is broader than software alone. Its structure spans semiconductors, cloud infrastructure, and emerging AI technologies, with 400-plus holdings designed to capture the breadth of the sector without requiring investors to pick individual names.
Mechanically, the fund is built to be a buy-and-hold vehicle rather than a tactical trading product. The return engine is the underlying businesses themselves: no options overlays, no leverage, and no synthetic instruments. The expense ratio is described as $9 per $10, 000, and portfolio turnover is 0. 08, signaling low churn.
Why Vgt can diverge from software: the weight of three companies
The most consequential driver of VGT’s resilience is concentration at the top. NVIDIA represents 18% of the fund and Apple another 15. 8%. With Microsoft at 10. 4%, these three positions total 44% of the portfolio. In a year when software stocks are described as being in a “quiet crisis, ” that composition matters because neither NVIDIA nor Apple is a pure software company.
NVIDIA’s role is framed as central to AI infrastructure. The valuation “picture has shifted, ” with the stock trading at approximately 22 times forward earnings—presented as a level reflecting scale without the speculative premium that previously increased vulnerability. Operationally, NVIDIA’s Q4 FY2026 revenue reached $68. 13 billion, up 73. 2% year-over-year, with Data Center revenue at $62. 31 billion, a 75% year-over-year gain. On the earnings call, CEO Jensen Huang said: “The agentic AI inflection point has arrived. Grace Blackwell with NVLink is the king of inference today, delivering an order-of-magnitude lower cost per token. ” For Q1 FY2027, the company guided to approximately $78 billion in revenue.
Apple, by contrast, is described as the steadying weight. Its installed base surpassed 2. 5 billion active devices, characterized as a recurring revenue platform that can insulate it from quarter-to-quarter volatility that can punish pure software names.
The implication is structural: even if large parts of software weaken, a fund dominated by an AI hardware leader and a massive device ecosystem can behave differently than an index of software publishers. This is not a judgment about safety; it is an explanation for why category labels can mislead.
Where the “AI value chain” shows up—and what it suggests investors may be missing
The diversification inside VGT is not just about number of holdings; it is also about where those holdings sit in the AI economy. Beyond the top names, VGT is described as holding semiconductor equipment makers, cloud infrastructure providers, cybersecurity platforms, and emerging quantum computing positions—each in a small, diversified slice. That matters because AI’s growth can create bottlenecks outside the most visible software brands.
Separately, an “AI value chain” framing argues that markets have focused on a small set of mega caps viewed as the biggest winners of the AI revolution, contributing to extremely high valuations and an increased risk of disappointment and correction. It notes performance dispersion since February and says annual reports brought attention to the scale of AI-related investments and the time required before companies can expect a positive return on investment.
In that framework, “hidden winners” can sit in overlooked segments that remain essential to the continuation of the AI buildout. It points to technological bottlenecks linked to high-performance computing, particularly fibre optics, connectivity, and memory. These tensions can reinforce the dominant position of leading companies by giving them greater pricing power and influence in shaping industry standards. It also lays out an “active investment approach” that would replace very expensive tech giants—described as fundamentally sound, such as Nvidia—with other stocks upstream or downstream whose valuations may be more reasonable while still offering predictability in sales and earnings per share growth.
Examples cited include Taiwan-founded TSMC, described as the world’s leading dedicated semiconductor foundry, producing its latest 2-nanometre chips with a stated 30% gain in power efficiency, positioning it to supply GPU designers such as Nvidia and ASIC designers such as Google. It also cites US-based Micron Technology (founded 1978) as one of three global-scaled dynamic RAM producers alongside Samsung and SK Hynix, in a consolidated industry with high barriers to entry. High bandwidth memory is described as a critical bottleneck in AI computing, and Micron is characterized as a qualified and increasingly important supplier with multi-year visibility in demand growth from hyperscalers including Amazon Web Services, Google Cloud, and Microsoft Azure.
Taken together, this creates a practical question for investors watching vgt hold up: are they benefiting from broad exposure across the AI architecture—or are they unintentionally concentrated in a few dominant platforms that happen to be behaving better than software right now?
Accountability questions investors should demand answers to
Verified facts from the provided material show a clear divergence: software stocks are down sharply year-to-date while VGT is down less; VGT is effectively all-in on U. S. information technology; and 44% of the fund is in three companies, led by NVIDIA at 18% and Apple at 15. 8%. It also describes a low-turnover, no-derivatives structure and emphasizes AI infrastructure exposure through semiconductors and related segments.
Informed analysis grounded in those facts: the “resilience” story is inseparable from concentration and from where AI profits accrue in the stack. If the market narrative is “AI, ” the underlying reality can be closer to “who controls compute, devices, and infrastructure. ” That can help explain why a fund can resist a software drawdown—but it also raises transparency issues for investors who believe they are buying broad tech exposure while a small set of names dominates outcomes.
For public accountability, the essential demand is clarity: investors deserve plain-English disclosure of what is driving performance at any given moment—sector breadth, index rules, or the weight of a few holdings. Until that becomes standard, vgt will continue to look like a paradox on the surface while the real story sits in its concentration and its place in the AI value chain.
