Gold Rush
AI is a gold rush. OpenAI struck the motherload when it released ChatGPT. Anthropic, Google, Deepseek, others stampeded in after.
Gold mining is risky. So is building frontier models. Meta's flagship Llama model turned out a duster last year. Apple swapped its own efforts to hire Google to work its AI chats. OpenAI is in the poll position to win the frontier race. It's also losing mountains of money.
Picks
More risk-sensitive prospectors rush to sell stuff the gold miners need. The "picks and shovels" trade. This way you don't tie yourself to any single claim where you may strike it rich or you may strike out. It's a popular investing framework too.
The AI gold rush has spurred loads of pick and shovel winners. Demand quickly jumped for accelerators — GPUs, ones made by Nvidia. Investors noticed. Nvidia's business and share price launched. Then, last year, the bonanza trickled into memory — HBM and DRAM. Even energy had a turn.
AI fever is still going strong. Usage is still ramping. Fabs — the factories that make all the chips — already struggle to keep up. They'll need more capacity. Who’s bringing the picks and shovels for that?
And Shovels
A thing you do if you're already selling picks to gold miners is you go ahead and sell them shovels too.
Lamplighter introduced ACMR as a cleaning service. It delivers precision cleaning for fabs making circuits and memory. That's 70% of its business. Investors don't expect much from it.
The other 30% of its business comes from machines that go into making chips. It makes packaging machines. It makes electrochemical plating machines. It makes deposition kit. That business grew 40% last year.
Despite running into capacity constraints in 2025, fabs have been gun shy about committing more capital to new plants. They only make money with massive volumes. Those massive volumes, however, are piling up. Logic and memory makers need to make more stuff. The fabs need more machines to do that.
With capacity sold out for 2026, Fabs are starting to dip their toes in the water. They're starting to order. This book of ACMR's business looks likely to pick-up even more in the second half of 2026. Investors haven't ignored this completely. Prices for companies delivering wafer fabrication equipment have been on a tear.
File my claim
Enabling miners to take risks can be a lucrative business. It can also be a lot less risky than the business of mining or building frontier LLMs. As demand from the big AI labs piles up, Fabs are starting to do something. Whatever they do, they'll need more machines. That opens the door for ACMR's next step up in its business.
Investors noticed. ACMR doubled in 2025. But they're playing catch-up. ACMR's non-cleaning order book looks to expand even more in 2026. It has the chance to be another bonanza and deliver value to shareholders.
Disclaimer: None of this is investment advice. It's meant to illustrate ways LCM thinks about investing. Things that LCM decides are good investments for LCM and its clients are based on many criteria, not all of which are covered here. Some or all of LCM's ideas may not be suitable for other investors. LCM does not recommend investing either long or short any position mentioned. LCM may own positions in some of the companies mentioned. Some of its ideas will lose money — investing entails risk. See full disclaimer here.