Yet more SMH into the Trading 212 SIPP today. The VanEck Semiconductor ETF has been one of the best performers in the portfolio — and for good reason. Semiconductors are the picks and shovels of the digital gold rush, and right now we're in the middle of the biggest technology buildout in history. AI, cloud computing, EVs, advanced manufacturing — every single one of these trends runs on chips. And SMH owns the companies that design, manufacture, and supply them.
── What's Inside SMH ──
For anyone new to this ETF, SMH tracks the MVIS US Listed Semiconductor 25 Index — 25 of the largest semiconductor companies listed on US exchanges. The top holdings read like a who's-who of the chip industry. Nvidia — the AI chip king, the company that everyone wants exposure to. TSMC — the world's most advanced chip manufacturer, the factory the entire tech industry depends on. ASML — the Dutch company that makes the machines that make the chips, a genuine monopoly in extreme ultraviolet lithography with no competitor anywhere close. Broadcom — networking chips, wireless semiconductors, and a growing software business. AMD — CPUs, GPUs, data centre chips, giving Nvidia some competition. Qualcomm — the wireless chip giant behind most smartphones. Applied Materials and Lam Research — the equipment makers that supply the fabs. And a supporting cast of analog chip makers, memory manufacturers, and specialised semiconductor companies that fill out the ecosystem.
It's a concentrated bet — 25 stocks, heavily weighted toward the largest names, with Nvidia and TSMC often making up 20-25% of the fund between them. That concentration is a feature, not a bug: when semiconductors do well, SMH does very well. When they don't — and they won't always — SMH will feel it more than a broader tech ETF. I'm comfortable with that. I have VUAG and VWRP for broad-market ballast. SMH is a higher-conviction satellite position that earns its place through exposure to a secular trend I believe in strongly.
── Why Semiconductors Keep Winning ──
The chip industry has always been cyclical — boom and bust, capacity shortages followed by gluts, feast and famine. I've been around long enough to remember multiple semiconductor cycles, and they're not going away. But something has changed in the last few years that I think makes this cycle different from the ones I've watched in the past.
Semiconductors used to be cyclical because demand was largely driven by the PC and smartphone upgrade cycles — consumer electronics with predictable replacement patterns. Today, chip demand is driven by structural, multi-decade buildouts that don't follow consumer replacement cycles. AI data centres are being built at a pace that's straining the power grid. Cloud computing is still growing as enterprises migrate workloads. Electric vehicles use 2-3x the semiconductor content of a combustion engine car. Industrial automation, 5G infrastructure, defence systems, medical devices — all of them are becoming more chip-intensive every year.
The AI buildout alone is staggering. The hyperscale cloud providers — Microsoft, Amazon, Google — are spending hundreds of billions of dollars collectively on AI infrastructure, and the vast majority of that spend flows through to semiconductor companies. Nvidia's data centre revenue has gone from single-digit billions to a run rate that would have seemed absurd three years ago. TSMC is building new fabs in Arizona, Japan, and Germany to meet demand and diversify geographically. ASML's order backlog stretches years into the future, and each of their EUV machines costs more than a private jet and takes multiple 747s to ship. These are not companies that are worried about next quarter's consumer spending. They're building the infrastructure for the next decade of computing.
── Start Small, Never Give Up ──
This buy isn't a market call. I'm not saying semiconductors are cheap right now — they're not. I'm not saying there won't be pullbacks — there will be, probably sharp ones, because that's what this sector does. I'm saying that I believe in the long-term structural demand for chips, I believe SMH is a well-constructed vehicle for that exposure, and I believe that consistently adding to positions I have conviction in — small amounts, regularly, through thick and thin — will produce better results than trying to time the perfect entry point.
Start small is better than not starting at all. That's not just a slogan — it's the founding philosophy of this entire site. £5 a day invested in SMH is better than waiting for the perfect dip that never comes. £20 a week is better than sitting in cash because 'things feel expensive.' The maths of compounding doesn't care about your feelings. It only cares about time in the market, not timing the market.
And never give up. Markets will fall. SMH will have bad months, bad quarters, maybe bad years. The chip cycle hasn't been abolished — it's just been joined by secular demand that I think will make the troughs shallower and the peaks higher over time. If I'd sold every time the portfolio dipped, I'd have nothing. The investors who win are the ones who keep buying through the fear.
Compounding is the way to go. Always has been, always will be. Everything else — stock tips, market timing, get-rich-quick schemes — is noise. The signal is simple: buy good assets, buy them regularly, hold them for a long time, and let the maths do the heavy lifting. SMH is a good asset. I'm buying it regularly. I'm holding it for a long time. That's the whole strategy.
As always, nothing on this site is financial advice. I'm a 66-year-old UK investor sharing what I do with my own money in the Trading 212 SIPP. SMH can go down as well as up — semiconductor ETFs are volatile and concentrated. Past performance doesn't guarantee future results. Do your own research. Start small, stay consistent, and never invest money you can't afford to lose.