Maplelane Capital · long/short technology & semiconductor hedge-fund manager — running synthesis of his appearances, with per-transcript breakdowns and a stock index.
His favorite — Lam Research's high memory exposure into a memory-capex boom (a mid-27/28 story); he thinks street estimates are ~50-70% too low, with ~$55B of revenue and margins going significantly higher.
Left-for-dead foundry now "picking up customers" — evidence for the multi-spender thesis (more foundry spenders → semi-equipment boom) more than a single-name buy.
The lone big foundry spender that "underspent significantly" (~70% margins) and "made a mistake" it must rectify — the setup that drives the multi-spender semi-equipment capex boom.
In one line: A semiconductor-cycle bull — AI is long-term deflationary but near-term inflationary (skyrocketing CPU/memory/infrastructure input costs). After a decade of capex discipline the chip complex is in a powerful multi-year up-cycle: multiple foundry/memory spenders, all of which underspent, must now spend heavily — so own the semi-equipment makers (Lam Research), the memory names (Micron, SK Hynix, SanDisk — ~80% margins, a NAND cycle "we haven't talked about in a decade"), and the analog names (Texas Instruments, Renesas), where he thinks street estimates are 50-70% too low.
AI: deflationary long-term, inflationary first. Long-run, LLMs mean "a lot more for a lot less" (healthcare ~18% of GDP a prime target). But short-term it's inflationary — CPU/memory/infrastructure input costs are skyrocketing and the labor market is robust (software-engineer hiring +18% in a month as the old economy implements LLMs). Sticky inflation first, then deflation as robotics layers on — "I do not envy" the Fed.
The semi up-cycle — from one spender to many. For a decade only Taiwan Semi spent, and even it "underspent significantly." Now Intel's foundry is "picking up customers" and Samsung is announcing — multiple spenders, all under-invested, with forward-capex indicators (customer profitability) flashing. WFE goes from ~$120-130B toward ~$300B over 3-4 years.
Own the suppliers with pricing power. Customers run 70-80% margins (memory ~80%, Taiwan Semi ~70%, semi-equipment ~50%), so the equipment makers have pricing power on top of unit growth. Long-term agreements (LTAs) make the memory/equipment names less cyclical — which should re-rate the multiples. Pristine balance sheets fund M&A and buybacks.
Favorites. Lam Research (his #1 — high memory exposure, a mid-27/28 story, ~$55B revenue, street ~50-70% too low); the memory complex (Micron, SK Hynix, SanDisk); and analog (Texas Instruments, Renesas), which "could look like memory" on pricing. Find the names with an "AI power angle" for the biggest upside.
Discipline. The group has run hard ("100 miles an hour in a 60 zone") so expect accidents, but "the next 50 to 100 is up."
Transcripts
One dated page per appearance — each has its full stock table, talking points, and the saved transcript. Newest first.