JPMorgan Raises Kioxia Target Price to ¥155,000, Shifting Valuation Logic from Cyclical Discount to Cash Flow Capitalization
Alina Collins
JPMorgan nearly doubled its Kioxia target to ¥155,000 from ¥80,000, with almost all the upside coming from expanding the P/E multiple from ~7× to ~11× rather than higher earnings forecasts — signaling that the sell side is starting to price part of Kioxia's profit as sustainable cash flow, not just peak-cycle earnings.
The target nearly doubled — where did the upside come from?
JPMorgan applies FY2027 EPS of ¥9,896.79 × ~11× P/E to reach a ¥155,000 year-end 2026 target.
EPS was revised up only 8–11%, yet the target nearly doubled. This means → the bulk of the increase is multiple expansion, not a linear lift in profit estimates.
In plain terms = the market previously gave Kioxia only ~7× P/E, embedding a heavy cycle discount. Moving to ~11× — roughly half a standard deviation above the global memory sector's historical average — says JPMorgan now believes part of the profit won't vanish with the cycle.
Why is "cheap P/E" often a trap in NAND stocks?
Traditional NAND cycle stocks carry a classic valuation trap: P/E is lowest when profit is highest — the denominator sits at its peak, and once prices fall, the seemingly cheap multiple evaporates.
JPMorgan's real question this time is not "how much can Kioxia earn in FY2027?" but "how much of that profit can be treated as more durable income?"
This reflects a pricing-framework shift from "discount the cycle peak" toward "capitalize the cash flow" — though they have not yet gone all the way to a growth-stock multiple.
What are the five variables behind this call?
Demand mix tilting to data centers: NAND bit demand CAGR is ~22% for 2025–2028; data-center share rises from 30% to 50%, with absolute volume growing from 295 EB to 909 EB. The driver has shifted from smartphone restocking to data-center pull.
Product-mix upgrade: enterprise SSD (eSSD) revenue share rises from 34% in FY2025 to 67% in FY2028. This means → higher ASPs and more sustainable margins.
Front-end cost decline: BiCS8 is already in volume production, targeting >80% of GB output by FY2026-end; the 10th-gen 1Tb TLC product is in final reliability testing, with sample shipments targeted for summer — prices rising, mix upgrading, and costs falling all at once.
Why do long-term agreements and shareholder returns matter?
Kioxia is shifting its LTAs — long-term supply agreements — from single-year to multi-year terms. In plain terms = this is not just about locking volume; it lets the market treat a portion of profit as "visible cash flow," directly changing the basis for the cycle discount.
Kioxia expects to reach net cash by the end of FY2026 Q1; management is discussing cash-flow-based shareholder returns starting in FY2027 and has not ruled out studying a dividend as early as FY2026 H2.
This reflects a fundamental shift: for a NAND company long weighed down by debt and cycle-driven valuation, the ability to return cash is itself a reason for repricing.
What is AI SSD actually doing — beyond "more data"?
Kioxia's AI-inference product lineup goes well beyond the vague logic of "more data means more storage." The CM series handles high-bandwidth KV-cache reads/writes — the rapid-fire buffer used during large-model inference. The GP series, built on XL-FLASH, targets ultra-low-latency, ultra-high-IOPS workloads. The LC series uses QLC to deliver 245 TB of raw capacity.
AiSAQ, combining Milvus and NVIDIA cuVS, tackles DRAM and CPU bottlenecks in vector databases — the search-and-recommendation layer underneath AI applications.
This means → Kioxia's product logic is to embed SSDs into the efficiency constraints of GPU systems, not merely to expand cold storage — and that is the key to commanding higher pricing.
Where does the real investment debate sit now?
JPMorgan's conclusion: Kioxia can no longer be treated purely as a traditional NAND cycle stock, but it has also not fully proved it can escape the cycle discount. ¥155,000 represents "a higher cycle midpoint," not "a cycle-free growth multiple."
The real debate has moved from "will 2026 profits be strong?" to "how much repeatable profit survives into 2027–2029 and beyond?"
The answer will emerge over the next four quarters, tracked through LTA coverage rates, execution of shareholder returns, the BiCS cost curve, and AI SSD customer qualification progress.
Content is for reference only, not financial advice.