LRCX

Lam Research is the odd one out this week — it's the company that beat expectations and actually went up. It reported on July 29: earnings of $1.82 per share against the $1.69 analysts expected, and $6.72 billion in revenue, up about 30% from a year ago. The stock rose roughly 6% right after and has climbed around 18% since. For the full fiscal year, revenue was $23.2 billion and profit was $7.3 billion, up 36%. The guide for next quarter is roughly $8.1 billion, which is a big step up from the $6.7 billion it just did. After watching Celestica, Credo, Coherent, and NVIDIA all get punished for good news, this one is worth understanding. What Lam does is simple enough: it makes the machines that make the chips. It doesn't design chips or sell them — it sells the etching and deposition equipment that companies like TSMC, Samsung, and Micron need to build a factory. That's a different place in the food chain than everything else we've looked at. Lam gets paid when someone builds a fab, and fabs take years to plan and build, so its revenue arrives on a slower, steadier schedule than a company selling into this quarter's data center order. It also has a large service business — about $2 billion a quarter just maintaining machines already installed — which keeps coming in whether or not anyone is buying new ones. Management raised its estimate for total industry equipment spending to around $140 billion for 2026, and a big chunk of the current demand is memory, driven by the high-bandwidth memory that AI chips need. So why did this one get rewarded when the others didn't? Our guess is timing. The market spent late July worrying about whether AI spending is sustainable, and Lam's business is tied to factory construction that's already committed and underway — harder to cancel, easier to believe in. That's a real advantage, but we'd be careful not to oversell it. Chip equipment is famously cyclical: when a building cycle ends, orders don't slow down gradually, they fall off a cliff, and this stock has more than tripled in about a year. One thing we noticed and don't love — insiders have sold Lam stock 54 times in the past six months and bought exactly zero times, including about $47 million from one board member. That's the same pattern we flagged at Celestica and Credo, and we read it the same way: not a reason to avoid the stock, but not a vote of confidence either. If you own this, the number to watch is that industry equipment spending forecast. As long as it's going up, Lam is fine. When it flattens, get ready for a rough ride. (written with the help of Claude)