Direct:
Check my UBS report on Kioxia from a few days ago as well.
Goldman upgraded Kioxia (TYO: 285A) to Buy and nearly doubled its target, from ¥48,000 to ¥93,000, on the view that this NAND up-cycle runs longer and peaks higher than its prior numbers allowed. The argument from analyst Shuhei Nakamura is that the industry structure has held steady while the cycle itself has shifted under it.
Six players still make NAND, and the Herfindahl-Hirschman index, the standard concentration measure, has barely moved in a decade. That is Goldman's basis for treating NAND as the same cyclical industry it has always been, but one whose current up-cycle lasts far longer than past ones. Peak profits should keep climbing through the fiscal year ending March 2029 instead of topping out early the way they have in past cycles, and the steep price collapses that historically dragged makers into losses look unlikely for the next two to three years.
That longer runway rests on Goldman's revised supply/demand balance, which now shows NAND in deficit (excess demand) through 2028: roughly -4.4% in 2026, -4.6% in 2027, and -3.0% in 2028. The prior model ran only through 2027 and put that year at -2.1%, so both the depth and the length of the shortage moved further against supply. Samsung (KRX: 005930), SK Hynix (KRX: 000660), and Micron (NASDAQ: MU) are directing capital toward DRAM and HBM instead, which limits the NAND capacity that would normally close a shortage.
On Kioxia specifically, Goldman sees the DRAM procurement risk for enterprise SSDs as largely resolved, and the BiCS8 transition lowering cost and lifting margin over time. It lists IR Day, quarterly results, and continued NAND price increases as near-term catalysts.
Goldman is not raising the multiple at all. It still values peak-cycle earnings at a mid-to-high single-digit P/E, so the entire upgrade comes from those peak profits being higher and lasting longer, which puts the weight on the FY3/29 earnings power actually showing up.
For anyone watching consumer SSD prices, a deficit running through 2028 means the upward pressure on NAND contract prices has years left to it, and a quick return to bargain-bin drive pricing is hard to underwrite.
Drafted with AI assistance against parallel reporting.
Sources:
- Goldman Sachs, Kioxia Holdings rating change (Buy from Neutral, analyst Shuhei Nakamura, price target ¥93,000 from ¥48,000), late May 2026: NAND supply tightness through CY28, peak profits sustained through FY3/29, supply/demand balance estimates, HHI/industry-structure view, mid-to-high single-digit P/E on peak earnings, DRAM/HBM capex prioritization, enterprise-SSD DRAM procurement risk resolved, BiCS8 cost/margin, IR Day catalysts
- TipRanks / The Fly (May 2026): Goldman upgrade of Kioxia to Buy, ¥93,000 target, AI-server NAND demand thesis
- Futu / market coverage (May 2026): NAND entering new profitability up-cycle, Samsung NAND ASP +70% QoQ and SK Hynix +45% in Q1 2026
- Tom's Hardware (2026): Kioxia 2026 capacity sold out and likely through 2027, end of the cheap 1TB SSD
- Tom's Hardware (2026): Samsung and SK Hynix warn AI-driven memory shortages persist through 2027 and beyond
- Astute Group / Everstream (2026): memory makers shifting cleanroom and capex toward DRAM and HBM, limiting NAND supply growth
Reddit: https://ift.tt/tdJprGz