Kioxia trades closer to 2x forward EPS once the cash it will generate is counted, despite eye-popping results. April-June revenue rose 415% and GAAP EPS 4,506% year over year; on sell-side estimates it trades at 3x next-twelve-month EPS, and that year's net income is about a third of its market cap. Even if this is peak earnings, they should stay strong for years, generating enough FCF to buy back 50-100% of the float. Five-year LTAs near record prices are a good business decision. x.com  x.com  … Show more

Kioxia is very shareholder friendly at about 4x earnings, choosing to buy back stock rather than spend on capex. Per @jukan05, it plans to repurchase up to 30 million shares, 5.5% of shares outstanding, for up to ¥800 billion between August 3 and October 30, alongside April-September operating profit guidance of ¥3.16 trillion and a 3-for-1 split. A miss against expectations would matter more if the stock were still at ¥100,000 rather than near ¥49,000. x.com  x.com 

Kioxia's 74% operating margin is a shortage premium with an expiration date set by its own investor deck. AI inference storage demand is real, but management's supply-demand chart shows NAND returning toward balance by late 2027, and its long-term model targets mid-20% operating margins. The stock is about 12x the annualized peak quarter but roughly triple that on the company's normalized plan. The bull case: hyperscaler contracts and the inference-storage tier keep margins structurally higher. … Show more

Kioxia's next-12-month growth is likely to far surpass that of other major memory makers, yet it trades cheaper with no clear reason, possibly even below SK hynix's US ADR. @CeleritasRe notes it is doing a buyback and is probably the only name on the list that could credibly attract a takeover bid. x.com  x.com 

Kioxia's P/E and growth are about as good as investing gets, so anyone worried about holding it here simply owns too much. Fear around Chinese AI models looks misplaced: they cost just as much to serve as US models, which points to an even bigger compute shortage. Short term, momentum has turned and selling across the compute trade is likely to continue, but the plan is to ride it out; there is some risk, as always. x.com 

Kioxia faces eroding pricing power as surging chip production capacity threatens the storage-chip boom. TSMC's signal of a hike in 2026 capex to $64 billion fed fears that new capacity will erode the pricing that made memory a gold mine, amid broader doubts that AI spending turns into stable profits. The stock has fallen 52% from its June highs. After a 600% climb on retail leverage, the descent is rarely gentle, and in cyclical industries supply-side realities catch up to the hype. x.com 

Large passive buying relative to Kioxia's float should support it over the next month. The TPX buying event involves 56 million shares on Thursday (roughly $4-4.4 billion), with another 56 million at November's end; an earnings buyback would add more if announced. Early buyers are stopping out before Thursday. Fundamentals also offer support, so it is not a terrible time to pick a level, though October 30 earnings are the main risk. x.com  x.com  x.com  x.com 

Sandisk's fast-growing data-center revenue strengthens its case among memory names with upside. Data-center revenue rose 61% quarter over quarter; revenue of $11.03B, adjusted EPS of $49.51 and adjusted gross margin of 85.1% all exceeded estimates. For upside among Micron, Sandisk and Kioxia, Micron ranks first, then Sandisk, then Kioxia, until further guidance. x.com  x.com 

Kioxia's need to do a buyback is proof that the flash-memory boom is over. x.com 

Kioxia could double in the next 6-12 months if management delivers on capital returns, since it trades at only 3-4x peak earnings. Alpha Metallurgical Resources, at similar multiples, kept rising on a buyback narrative even after its earnings peaked in 2022. Memory's earnings peak may also be more predictable than coal's, and with analysts seeing it in 2028, Kioxia's stock might not peak until late 2029. x.com  x.com