Somewhere deep in the AI supply chain sits a component roughly the size of a grain of rice, and there are now three exchange-traded funds competing to sell it to you.
An MLCC is a multilayer ceramic capacitor. It stores a small amount of charge, smooths out voltage, and costs a fraction of a cent. A modern AI server contains thousands of them. Until recently, the number of people outside the passive-components industry who could define one was, I would estimate, in the low hundreds.
There are now three ETFs on them. Defiance got to market first with a pure-play capacitor fund. Roundhill followed with a broader version taking in circuit boards and packaging. Global X has now arrived with this one, covering ceramic, aluminium electrolytic, film, tantalum, polymer and silicon capacitors, plus chip resistors, inductors, signal transformers and circuit protection.
That is a remarkably specific list. It is also, I want to be clear, a defensible thesis. Capacitor content per server has genuinely risen, the manufacturing base is genuinely concentrated, and picks-and-shovels exposure to AI infrastructure is a real argument. What is funny is not the idea. What is funny is that three product teams independently reached the ceramic capacitor in the same short window, having presumably worked their way down from chips to memory to power to cooling to the passive-components, like archaeologists reaching bedrock.
There is nothing below the capacitor. I have checked. The next fund down would be dielectric powder.
An autocallable is a structured product that pays you an attractive coupon, caps your upside, redeems early if the reference index behaves, and hands you the losses if it does not. For roughly two decades these were sold through private banks, accompanied by a forty-page term sheet and a conversation about suitability.
They now have tickers.
Pacer has filed a ladder of these, which is the part that gives me pause. Not one product, a range. The construction is sound enough: pooling autocallables into a fund removes the single-issuer credit risk and the single-observation-date cliff edge that made individual notes so unforgiving. Diversification genuinely helps here. This is a better mousetrap than the note it replaces.
I have no notes on the strategy. Currency-hedged US technology exposure for Australian investors is an entirely reasonable thing to want, and hedging the currency is arguably the responsible version of the trade.
I simply want to acknowledge the ticker. Somebody at that firm sat in a room, worked through the available three- and four-letter codes, and landed on the hypertext transfer protocol. It is the best ticker in this batch by a distance, and possibly the most searchable ETF ever listed, in the sense that it is completely unsearchable.
Whoever did that deserves a small trophy.
Seven Pictet lines arrived at once, covering US, European, World and World ex-US equity, in accumulating and distributing flavours. The approach is systematic stock selection that strips out style, size, sector and region effects to isolate what is left, with a stated ambition of beating the benchmark by roughly a percentage point after fees, with low tracking error. Notably, it is neither a factor model nor a language model, which is a refreshing thing for a firm to say out loud in the current climate.
Goldman's arrival in emerging markets extends its existing range using the same grammar with one word changed. Pictet says AI-Enhanced. Goldman says Data Enhanced.
Both are describing quantitative equity, which has existed for forty years and once went by the name “enhanced indexation” before that phrase acquired a faint smell. The word doing the work in both names is “enhanced”, which is among the great load-bearing adjectives of asset management. It promises improvement without ever specifying the baseline. The interesting question is not whether these strategies work, because some of them demonstrably do. The interesting question is what we will be calling them in five years, once “AI” has aged into the same gentle embarrassment as “smart” and “quant” and “enhanced” itself.