Leon Shaulov — Intelligence as Infrastructure: How AI Is Rewiring the Economy

Lam Research makes the machines that chip factories use to manufacture memory chips (it's a "semiconductor equipment" or WFE — wafer-fab-equipment — company). It's Shaulov's favorite idea. His logic: for years memory makers barely invested in new capacity, so the market is skeptical that big orders are coming.

He thinks that's wrong — a wave of memory spending is coming around 2027-2028, and because Lam is heavily tied to memory, its sales and profits should jump well past what Wall Street expects (he says estimates are 50-70% too low, with revenue around $55 billion and fatter margins). When the chipmakers finally spend, the toolmaker gets paid first.

Texas Instruments is the largest maker of "analog" chips — the unglamorous components that handle real-world signals like power, sound, and temperature in virtually every electronic device. Shaulov calls it "very good" and thinks the whole analog sector is interesting because supply is tight.

His key idea: analog pricing "could look like memory" — meaning that as shortages bite, prices (and profits) could rise sharply the way they do in a memory up-cycle. He especially likes analog names with an "AI power angle" — those selling into the power-hungry AI data-center buildout.

Renesas is a large Japanese analog and microcontroller chipmaker (this RNECY ticker is its U.S. over-the-counter listing). Shaulov calls it "quite interesting" as part of his bullish view on tight analog supply.

It's the same thesis as Texas Instruments — analog chips are scarce, pricing could firm up like memory, and names plugged into AI power demand have "significant upside."

Micron makes memory chips — both DRAM (the fast working memory in computers and AI servers) and NAND flash (storage). Shaulov groups it with SK Hynix and SanDisk as the memory names riding a cycle he says we "haven't talked about in a decade."

What makes it attractive now: these companies are earning roughly 80% gross margins, and they're locking customers into long-term supply agreements (LTAs). Those contracts give them visibility and make the business less boom-and-bust than memory has historically been — which could earn the stocks a higher valuation.

SK Hynix is the big Korean memory maker (this HXSCL ticker is its U.S. over-the-counter listing) and a leader in the high-bandwidth memory (HBM) that AI chips need. Shaulov names it alongside Micron and SanDisk as the memory complex powering the up-cycle.

The appeal is the same: extraordinary profitability (around 80% margins) and a NAND/DRAM cycle that's finally turning after years of underinvestment, with customer profitability signaling that heavy spending lies ahead.

SanDisk is a pure NAND-flash memory maker (the chips used for storage in phones, laptops, and data centers); it returned to the public market in 2025 as a stand-alone company. Shaulov uses it as the cleanest way to play the NAND cycle he says has been dormant "for a decade."

His view is that NAND is entering a powerful, highly profitable upturn, de-risked by long-term supply agreements — the kind of neglected cycle where the move can be large because so few investors are positioned for it.

TSMC is the world's dominant contract chip manufacturer ("foundry") — it makes the advanced chips designed by Nvidia, Apple, and others. In Shaulov's story it's less a buy than the setup for his real trade. For a decade it was essentially the only big spender on new capacity, and he argues even it "underspent significantly" — "made a mistake" it will have to fix.

That under-investment, plus new foundry competition from Intel and Samsung, is exactly what creates his thesis: multiple chipmakers all needing to spend at once, which is great for the equipment suppliers. TSMC's ~70% margins show how much room there is to spend.