Memory Heat Check: MU, WDC, 005930.KS, 000660.KS, 2408.TW

Published on: Apr 22, 2026
Author: Brandon Kwan

The hottest stocks going are somehow still priced like the party ends by midnight. Memory makers are the spine of the AI build-out, but they keep trading on recession math and hangover multiples. In the past eight hours across Asia and Europe into US hours, the most active sector on real money and fast money screens was semiconductor memory, even as retail in the UK chased a big airline win at Rolls-Royce and a clean earnings beat at Reckitt, and Indian momentum funds kept leaning into PB Fintech, Tata Power, and United Breweries. Nice stories. The main stage is still the chips that feed the chips.

AI memory stocks lead as multiples stay stubbornly cheap

1. Micron Technology (MU): What drove attention today is the same thing that has been driving it all year: the market is finally connecting the dots between AI server orders and the memory content that makes those racks useful. HBM3E shipments are scaling, DRAM pricing has turned, and Micron keeps showing up as the liquid US pure-play every time the tape wants AI exposure without paying GPU-god valuations. Trading profile: large cap, deep liquidity, options-friendly, and classic cycle beta. Margins and cash flow whipsaw with ASPs, which is why the stock still gets valued like a boom-bust even as mix shifts to higher value bits. Key takeaway: if you believe AI is a multi-year capex cycle and not a 2024 fad, MU is still the blunt instrument the market reaches for when it wants DRAM torque with US governance. The bear case is simple too: if AI slows or supply floods, memory gets paid last and hits the floor first.

2. SK Hynix (000660.KS): Attention today is about positioning and leadership. Hynix remains the HBM supplier of record for the AI duopoly, and supply chain chatter keeps pointing to tightness extending into next year as capacity crawls up the learning curve. Every time Nvidia talks about more powerful accelerators, Hynix gets bid on the read-through for HBM layers and yields. Trading profile: KOSPI heavyweight with heavy foreign flow sensitivity to the won, capex-gulping investment cycle, and earnings that swing far wider than foundry peers. The stock offers the purest HBM leverage but lives under the usual export-control and customer-concentration clouds. Key takeaway: the market is paying for the crown jewel of AI memory, but it is still applying a cyclical discount. If HBM stays supply constrained and pricing rational, operating leverage does the heavy lifting; just know that any misstep on ramp yields or China exposure will get punished in real time.

3. Samsung Electronics (005930.KS): Today’s bid is the low-drama, high-cash pick. Samsung is pressing hard to close any HBM performance gap while enjoying the same DRAM and NAND tailwinds that are lifting the group. Investors who missed the high-beta Hynix move rotate here for balance sheet comfort, vertical scale, and a buyback story. Trading profile: mega-cap with bottomless liquidity, diversified across handsets, displays, logic, and foundry. That diversification blunts memory torque and helps when the cycle turns against you. Valuation still reflects a conglomerate discount and the habit of investing through the downcycle. Key takeaway: Samsung is the safer memory upcycle ride. You won’t get the wild upside of a pure-play, but you also won’t get carried out on a stretcher if ASPs wobble for a quarter. For institutions that need to be in the theme without sleeping under the desk, this is the button they press.

4. Western Digital (WDC): The attention driver today is the double-barrel AI narrative: NAND price repair on one side and nearline HDD demand from data centers on the other. WDC keeps catching flows on any whisper of better flash pricing discipline and every time hyperscalers talk about capacity adds. Optionality around portfolio structure and strategic moves lurks in the background and keeps the rumor premium alive. Trading profile: mid-to-large cap volatility machine with a history of sharp moves around earnings, commentary on ASPs, and capital structure decisions. It is less an AI purity play and more a storage beta proxy that gets a seat at the table as the data glut grows. Key takeaway: if you want a storage name that straddles NAND and spinning rust, WDC is your ticket. You are betting on management to squeeze margins out of two very different cycles at once while keeping balance sheet risk in check.

5. Nanya Technology (2408.TW): What lit it up today is the classic late-in-cycle rotation to the higher-beta satellites. DRAM ASPs are creeping higher, Taiwan flows are risk-on, and technicians love a clean breakout when the big caps have already run. Nanya’s operating leverage cuts both ways, which is exactly why it becomes a crowd toy when the tide turns. Trading profile: mid-cap DRAM specialist with thinner liquidity than the giants, high sensitivity to spot price moves, and earnings that can swing from famine to feast in a hurry. It is not for widows and orphans. Key takeaway: Nanya is a leverage play on the DRAM cycle tightening. If you believe the supply side keeps its discipline and AI keeps slurping bits, the torque here is real. If not, this is the one you trade rather than marry.

The irony under all this activity is that the market still prices memory like a short-term sugar high, not a structural demand rerate. That is why the multiples look like clearance-rack specials even as units per server and high-bandwidth content rise quarter after quarter. Investors are choosing their poison: pure-play torque with headline risk, or conglomerate insulation with diluted upside. Meanwhile, other corners of the tape prove the market still rewards clean narratives. UK buyers piled into Rolls-Royce on a real contract win and into Reckitt on an earnings beat. In India, breakout hunters chased PB Fintech, Tata Power, and United Breweries on technicals and incremental fundamentals. Those are solid trades. The weight of global liquidity, though, keeps gravitating back to the silicon that feeds the AI machines. The cheapest hot stocks are staying cheap because the crowd still thinks the music stops soon. If it doesn’t, this sector’s next leg is going to squeeze a lot of bravely underweight portfolios.

AI Copper Lithium