Korea just announced $1.3T into semis. I went back and redid my equipment names

Last week Korea, Samsung and SK basically committed almost 2,000 trillion won (\~$1.3T) into chips, AI data centers and the whole stack over the next 10 yrs. Biggest industrial plan in their history.

So I sat down and reworked every semi equipment name I follow. My take:

If HBM and DRAM stay the AI bottleneck, the first dollar a fab spends in a capex cycle goes to equipment. Not packaging, not wafers. Tools. Equipment orders hit the P&L way faster than fab output ramps.

$AMAT is the one I keep coming back to. Not because they win any single step the hardest, but because you literally can't build a fab without them showing up. PVD, CVD, CMP, epi. HBM, DRAM, NAND. Samsung, SK, Micron, TSMC, Intel glass substrate, advanced packaging. They touch all of it. Pulled the chart on moomoo, it's been riding the 20 EMA since April and hasn't even tested it. Tells you where the money is parked.

$LRCX is the higher beta HBM bet. HBM needs way more etch, TSV and wafer thinning vs vanilla DRAM. AMAT mgmt themselves said HBM has 3-4x the process steps. If HBM stays tight LRCX prints the most.

Quick rest of the bench:

$KLAC. Inspection. Boring grower, yields matter more every node.

$ONTO. Advanced packaging metrology. Underrated if CoWoS keeps choking.

$TER. Test. Indirect, longer test times = good for them.

$AMKR. OSAT, rides TSMC capex.

$ACMR. Different story, China domestic substitution play, not the global capex trade.

Ranking on the storage capex trade: AMAT, LRCX, then KLAC, then ONTO/AMKR, then TER, ACMR is its own thing.

If I could only hold one for the AI decade I'd still pick AMAT. Capex itself is the trend.

Am I sleeping on anything here? Anyone bigger on LRCX over AMAT?