While experts continue to debate whether artificial intelligence (AI) is in a bubble or not, whether we are prepared for its transformational impact, or if we should be wary of it or embrace it wholeheartedly, the theme has quietly become one of the top performing themes this year. And the companies leading the charge might surprise you. It is not Nvidia. It is the broader semiconductor ecosystem sitting behind it.

In WisdomTree's monthly thematic update to end of May, semiconductors ranked as the top theme in terms of inflows into European listed thematic exchange-traded funds (ETF)s in May. It was also ranked number two in terms of performance in May across European listed ETFs and mutual funds, falling just slightly behind quantum computing. The companies making memory chips, building AI infrastructure, and securing AI systems are the ones driving returns right now. Nvidia has been getting the headlines for some time, but the picks and shovels of the AI revolution are having their moment.

A recent encounter of my own with AI reminded me of an important lesson for investors in this space. With the UK going through a scorching summer, I noticed our refrigerator wasn't cooling well enough. I asked an AI which setting was colder on my refrigerator's 2-to-8 temperature scale: 2 or 8? AI's answer was 8. Still, no improvement. I then asked a different AI, given nobody has the time to browse through manuals, the same question. The other AI said 2 is colder. Quite literally the opposite answer. I switched the setting and it worked. The second AI was right.

Increasingly, people are mentioning this trend. Asking one AI to verify the answers of another. Or, at the very least, using different AI tools for different purposes. And switching as soon as a superior new solution becomes available. This means that in the world of AI, it's not winner takes all. Not even close. For investors, this directly translates to diversification. At WisdomTree, we have been advocating for the idea of spreading one's bets across the AI value chain for many years. And right now, the data is making that case for us.

This blog outlines three companies leading AI performance this year. What they do, why markets are paying attention to them, and how investors can think about their prospects and role in the AI value chain.

SK Hynix

SK Hynix is a South Korean world-dominant supplier of memory chips, producing three types that between them power virtually every computing device on the planet. DRAM, or dynamic random access memory, is the fast, short-term memory that a processor uses in the moment. We can think of it as the desk where work gets done. NAND is slower storage that holds data even when the power is off. In an office, this is like the filing cabinet. And high-bandwidth memory, or HBM, is a specialised high-performance version of DRAM built specifically for AI. We can think of it as a whiteboard placed right next to the processor, keeping the most critical information instantly accessible at all times. It sits directly alongside GPU processors, such as those developed by Nvidia, inside AI servers. If Nvidia's processors do the thinking, memory chips made by SK Hynix give those processors information to think with. HBM enables this flow of information at extraordinary speed, with the high bandwidth facilitating vast amounts of data flowing through to the processors at exactly the right moment. In the world of AI, where models are processing billions of parameters simultaneously, that speed is of paramount importance.

In the first six months of 2026, the company’s stock was up 305%. But the market excitement has been backed by strong fundamentals. The company delivered a stunning Q1 2026, posting record quarterly revenue of KRW 52.6 trillion, up 60% quarter-on-quarter and 198% year-on-year. Operating profit hit KRW 37.6 trillion, up 405% year-on-year, with an operating margin of 72%. And despite the stock’s strong gains this year, it still trades at a price-to-earnings ratio (P/E) of around 20 and a forward P/E of under 7¹.

Going forward, Bloomberg estimates forecast earnings per share (EPS) to more than double by the fourth quarter of this year. Such forecasts are backed by the belief that global chip wafer shortage is likely to persist until 2030, as stated by SK Group’s Chairman Chey Tae-won, given strong demand growth. Moreover, the company’s foray into bespoke memory chips tailored to specific customer platforms creates a key competitive differentiator.

Micron Technology

Micron Technology is the American equivalent of SK Hynix, and one of only three companies in the world capable of making high-bandwidth memory (HBM) chips at scale, alongside SK Hynix and Samsung. Where SK Hynix dominates the market, Micron is the fast-rising challenger, closing the gap rapidly. Beyond HBM, Micron also makes the broader memory and storage chips that fill AI data centres. We can think of these as the shelves in a warehouse where data sits waiting to be processed. AI servers need roughly six times more of this memory than a standard server, which is why demand for Micron's products has been so explosive.

In the first six months of 2026, the company's stock was up 305%, eerily identical to SK Hynix. Again, the excitement has been well earned. In their latest quarterly earnings, which they refer to as the third quarter of their fiscal year 2026 ending May 2026, Micron delivered one of the most staggering results in the company's history. Revenue came in at $41.5 billion, up 346% year-on-year, with a gross margin of 85% and net income of $28.2 billion. To put that in context, the company earned more in this single quarter than it did in several entire prior years combined. Data centre demand was the engine behind it, with cloud and core data centre business units each posting operating margins above 78%. And looking ahead, Micron guided for their fiscal Q4 revenue of $50 billion, suggesting the trajectory is still pointing upward. The company trades at a P/E of around 22 and a forward P/E of around 13, reflecting strong expected earnings growth ahead².

Going forward, the picture looks compelling. Micron has confirmed that its entire 2026 HBM production is already sold out under long-term contracts, giving investors an unusual degree of revenue visibility. The company is also a confirmed supplier into Nvidia's flagship Blackwell and upcoming Vera Rubin GPU platforms, meaning its fortunes are closely tied to the most important product cycles in AI hardware. The key risk to watch is whether the memory market's current supply tightness eventually gives way to oversupply, a pattern that has caught memory investors off guard before.

Intel

Intel is perhaps one of the most remarkable comeback stories in technology. For decades it was the undisputed king of semiconductors, the company whose chips powered virtually every personal computer on the planet. Then came a long and painful decline: missed product cycles, manufacturing stumbles, and a failure to position itself for the AI era left the stock at multi-year lows as recently as early 2025. But under new CEO Lip-Bu Tan, who took the helm in March 2025, Intel has staged a dramatic revival. The company is now pushing hard into AI through two distinct angles. First, its Gaudi AI accelerator chips offer an alternative to Nvidia's dominant GPUs for running AI workloads. Second, and perhaps more importantly, Intel's traditional CPUs, the central processing units that act as the general brain of any computer system, are proving essential in the AI era for tasks like data preparation, orchestration and the kind of localised AI inference that happens closer to the end user rather than in a giant data centre.

In the first six months of 2026, Intel's stock was up 278%, one of the most striking share price recoveries in the market. The turnaround was given its clearest validation yet in Q1 2026, when Intel reported revenue of $13.6 billion, up 7% year-on-year and well ahead of analyst expectations. Its data centre and AI segment grew 22% year-on-year to $5.1 billion. The results were so far ahead of consensus that the stock posted its best single day since 1987 on the back of the announcement. Intel’s valuation, however, is currently very lofty with a P/E of 448. But markets are pinning their expectations on continued earnings revival with significant contraction in the forward P/E which currently stands at 103³.

Going forward, the investment case is the most nuanced of the three companies here. Intel is not just a chip designer but also a chip manufacturer, and its foundry business, which makes chips on behalf of other companies, could become strategically critical as the US looks to reduce reliance on Asian semiconductor production. The company is pressing ahead with its next generation 14A manufacturing process, with multiple customers actively evaluating the technology. There are, however, still some execution risks. GAAP profitability is still elusive given one-time charges and restructuring efforts⁴, competition from AMD and custom chip designers is fierce, and the current valuation already embeds a significant amount of optimism. But for investors willing to take a longer view, few stories in technology carry as much potential upside if the turnaround continues on its current trajectory.

How does WisdomTree capture the opportunity?

The WisdomTree Artificial Intelligence UCITS ETF was the first thematic ETF launched by WisdomTree in November 2018. The ETF currently stands just shy of $1.5bn in assets under management. Built in partnership with industry experts, the Consumer Technology Association (CTA), the ETF invests across the following categories:

Enablers (50% weight) – companies that develop the building block components for artificial intelligence such as advanced machinery, autonomous systems/self-driving vehicles, semiconductors, or databases used for machine learning.

Engagers (40% weight) – companies that design, create, integrate, or deliver artificial intelligence in the form of products, software, or systems.

Enhancers (10% weight) – companies that provide their own value-added services within the Artificial Intelligence ecosystem, but which are not core to their product or service offering.

It is no coincidence that Enablers carry the highest weight in the ETF at 50%. Semiconductors sit squarely in this category. They are the building blocks without which none of the AI story is possible. No memory chips, no AI servers. No AI servers, no large language models. No large language models, no ChatGPT. SK Hynix and Micron, two of the three companies profiled in this blog, are classified as Enablers, and their performance this year is a direct reflection of that foundational role.

Intel tells a slightly different story. Given that AI is still an evolving part of its broader business, it currently sits in the Enhancer category. But this is precisely where the framework becomes interesting. As AI becomes increasingly core to a company's products and revenues, classifications can and do shift. An Enhancer today can become an Engager or even an Enabler tomorrow. Intel's turnaround, if it continues, may well be a case study in exactly that kind of evolution.

Notably, companies are assessed for inclusion and weighted based on the assessment from the CTA to capture potential success stories from across the value chain rather than weighting by market capitalisation. When performance rotates from one winner to another, which is indeed the case in the world of AI, the merits of this approach become clear.

Closing remark

The three companies profiled here are by no means an exhaustive list. They are simply a reminder that the semiconductor story is broader, more global, and more diverse than the Nvidia narrative alone might suggest. SK Hynix is South Korean. Micron is American, highlighting the global nature of competition in this sector. And Intel is a decades-old giant rewriting its own story. They all, however, have in common eye-watering share price gains this year backed by real earnings, real revenue growth, and in most cases, valuations that are far from frothy. There is space in this ecosystem for multiple winners, and a market-cap approach would miss many of them. That said, one risk deserves an honest mention. The investment case for these companies rests, to a significant degree, on the continued willingness of the world's largest technology companies to spend heavily on AI infrastructure. If that capital expenditure were to slow materially, these businesses would feel it. For now, there is little sign of that happening. If anything, the spending plans being announced suggest the opposite.

¹Bloomberg, SK Hynix, as of 8 July 2026. Returns in KRW.

²Bloomberg, Micron Technology, as of 10 July 2026. Returns in USD.

³Bloomberg, Intel, as of 10 July 2026. Returns in USD.

⁴GAAP, or Generally Accepted Accounting Principles, is the standard set of rules companies must follow when reporting their finances. Non-GAAP figures strip out certain one-off or non-cash items, such as stock-based compensation or restructuring costs, that companies argue do not reflect the true underlying performance of the business. Investors often watch both, but it is worth knowing that non-GAAP numbers will almost always look more flattering than the GAAP equivalent.