Bernstein raises SNDK target price to $1,250, with a blue-sky case at $3,000
AI summary card
Bernstein raises SNDK target price to $1,250, with a blue-sky case at $3,000
The report argues that TurboQuant has only a limited impact on NAND demand, while NAND ASP continues to rise more than expected, leaving substantial room for further upside in SanDisk earnings and valuation.
- FQ3’26 non-GAAP EPS is forecast at $14.18, above consensus of $13.99 and the company guidance range of $12-14.
- FQ4’26 non-GAAP EPS is forecast at $25.30, above market expectations of $18.78; if NAND ASP rises 75% for two consecutive quarters, FQ4’26 EPS could reach $40.53.
- FY27 non-GAAP EPS is raised to $144 in the base case and $224 in the bull case, driven mainly by stronger-than-expected NAND ASP.
- The target price is raised from $1,000 to $1,250, based on average FY26-29 EPS of about $113.93 and an 11x valuation; the blue-sky case approaches $3,000 using 13x bull-case FY27 EPS.
- The report believes TurboQuant mainly affects KV cache and the HBM/DRAM layer, with limited impact on long-term model storage, training data, checkpoints, and static data demand for NAND.
Report interpretation
Overview
Bernstein's SanDisk research centers on the view that the market is underestimating the NAND price cycle and SanDisk's earnings leverage. After TurboQuant triggered a pullback in the share price, the report reiterated its bullish stance, arguing that KV-cache compression has at most a marginal impact on NAND demand, while long-term data storage, training data, checkpoints, and general static data in AI infrastructure will continue to drive storage demand expansion. Based on stronger NAND ASP, the report substantially raises earnings estimates and lifts the target price to $1,250.
Core views
The report's core views are: first, TurboQuant concerns are overstated, with the impact concentrated on inference KV cache and the HBM/DRAM layer, no impact on HDD, and only a possible limit on some future incremental penetration for NAND; second, after years of underinvestment, the NAND industry remains in a structural supply shortage, and AI-driven data center exabyte demand remains strong; third, the magnitude and persistence of NAND ASP increases may significantly exceed market expectations, driving FY27 EPS well above consensus; fourth, SNDK currently trades at only about 9.0x consensus FY27 EPS, 5.4x Bernstein base-case EPS, and 3.5x bull-case EPS, below the historical peak-cycle range of 10-13x, implying both earnings upgrades and valuation re-rating room.
Analysis framework
The report uses a three-layer analytical framework: first, it assesses TurboQuant's actual impact across different storage tiers; second, it rebuilds FQ3’26, FQ4’26, and FY27 EPS under NAND ASP scenarios; and third, it uses SanDisk's historical peak-cycle P/FE multiples and relative valuation versus the SOX index to determine the target price and blue-sky case.
Methodology notes
Assessing the impact of KV-cache compression across the HBM/DRAM, NAND, and HDD layers.
The report argues that TurboQuant mainly affects inference KV cache and cannot influence model weights, training workloads, or static data storage, so it does not fundamentally alter the medium- to long-term supply-demand trajectory for NAND.
Deriving EPS leverage under different quarterly NAND ASP growth assumptions.
The base case assumes FQ3’26 NAND ASP rises 55% quarter over quarter and FQ4’26 rises 40%; the blue-sky case assumes both quarters rise 75%, which materially lifts FQ4’26 and FY27 EPS.
Referencing the two-year forward P/E multiples from multiple earnings peak cycles before SanDisk was acquired by WDC.
The report believes SanDisk typically traded at 10-13x two-year forward EPS during more sustained historical peak cycles, while the current implied multiples for Bernstein's base and bull-case EPS are lower.
Asset mapping & comparison
Structured mapping from thesis to named assets (strengths, weaknesses, peers, risks).
- SNDK.USCore coverage name
- Strengths
- Rising NAND ASP, strong AI storage demand, improving supply discipline, and valuation below historical peak-cycle levels.
- Weaknesses
- Highly cyclical business, with disclosure and investor communication viewed by the report as confusing.
- Comparison
- Valuation remains below historical SanDisk peak-cycle levels and at a discount to the SOX index.
- Risks
- A decline in NAND prices, earnings already near peak-cycle levels, and the market not accepting the durability of high EPS.
- NANDPrimary earnings driver
- Strengths
- The strongest and still accelerating price increases, supported by AI data storage demand and supply shortages.
- Weaknesses
- If ASP peaks quickly or only affects part of shipments, EPS upside will be limited.
- Comparison
- The report believes TurboQuant's direct impact on NAND is weaker than on the DRAM/HBM layer.
- Risks
- A reversal in the price cycle and changes in supply or competition due to technological improvements.
- DRAM/HBMThe storage layer more directly affected by TurboQuant
- Strengths
- Still supported by AI compute and inference demand.
- Weaknesses
- KV-cache compression may have a more visible effect on DRAM/HBM utilization efficiency.
- Comparison
- Compared with NAND, DRAM/HBM is closer to the layer targeted by TurboQuant.
- Risks
- Inference efficiency gains weaken memory demand per unit of compute.
- HDDRelated storage asset class
- Strengths
- The report says TurboQuant has no impact on HDD, and long-term AI and cloud storage demand may still support demand.
- Weaknesses
- Highly exposed to the hyperscale cloud capex cycle.
- Comparison
- Compared with NAND, HDD plays a different role in long-term mass storage; the report also mentions STX and WDC valuation.
- Risks
- Absorption of cloud capex, changes in procurement patterns, and NAND technology improvements taking share from HDD.
- STXPeer comparison name
- Strengths
- The report notes Seagate's HAMR leadership and improving fundamentals support a higher valuation.
- Weaknesses
- Highly affected by the HDD market and cloud capex.
- Comparison
- Seagate's target price is derived as $620 using FY28 EPS of $29.5 and a 21x P/E.
- Risks
- WDC catches up faster in HAMR, cloud demand slows, and NAND substitutes for HDD.
- WDCPeer comparison name
- Strengths
- The report says WDC has a FY25-30E EPS CAGR of 46%, with a $340 target price implying 20x FY28 EPS.
- Weaknesses
- The HAMR technology transition may weigh on gross margin and EPS.
- Comparison
- Shares exposure to NAND and HDD cycles with SNDK, but the business mix differs.
- Risks
- HDD demand volatility, NAND technology improvements, and HAMR transition risk.
Key data
- Target price$1,250Raised from the prior $1,000, based on average FY26-29 EPS of about $113.93 and an 11x valuation.
- Blue-sky valuationAbout $3,000Derived using 13x bull-case FY27 EPS; the text also notes a figure approaching about $2,950 based on peak EPS.
- FQ3’26 EPS forecast$14.18Above consensus of $13.99 and company guidance of $12-14; blue-sky case is $17.47.
- FQ4’26 EPS forecast$25.30Above market expectations of $18.78; if ASP rises more aggressively, the blue-sky case is $40.53.
- FY27 EPS forecastBase case $144, bull case $224Driven mainly by stronger NAND ASP.
- Average EPS for 2026-29Base case $113.93, blue-sky case $167.93, consensus $69.92The base case is 63% above market, while the blue-sky case is 47% above the base case and 140% above consensus.
- Historical peak-cycle valuation10-13xThe two-year forward P/FE range SanDisk has historically traded at during sustained peak cycles.
- Current valuation comparison9.0x consensus, 5.4x base case, 3.5x bull caseAll based on FY27 EPS, showing the market has not fully recognized the upside scenario.
- AI data center demandExpected 2026 exabyte growth in the high-60% rangeManagement's latest guide has been raised materially from the low-20% range two report cycles ago, and does not include the additional 75-100 EB potential demand in 2027 previously associated with KV cache.
Impact & implications
If the report's view proves correct, SNDK's investment thesis will not be just a short-term NAND pricing trade, but a combination of earnings upgrades, a repricing of cycle durability, and narrowing valuation discount versus SOX. For investors, the key debate is the peak magnitude of NAND ASP, how long it lasts, and whether the market will value SanDisk using historical peak-cycle multiples.
Risks
- Near-term earnings estimates for NAND may be too high, and a downturn in the price cycle would pressure SanDisk's results and target price.
- SNDK's disclosure and investor communication are relatively confusing, which may hinder participation from quality-oriented investors.
- If NAND weakness shifts from cyclical to structural, DCF value and asset impairment value could be materially below replacement cost.
- The market may continue to believe current EPS is already at or near peak-cycle levels, and therefore may not assign historical peak-cycle multiples.
- If TurboQuant or other efficiency-improvement technologies expand to more workloads, some of the incremental AI storage demand could weaken.
What to watch
- SNDK will report FQ3’26 results on April 30, with focus on EPS, NAND ASP, and FQ4’26 guidance.
- Whether NAND ASP can reach the base-case assumptions of +55% quarter over quarter in FQ3’26 and +40% quarter over quarter in FQ4’26.
- Whether the two consecutive quarters of +75% quarter-over-quarter ASP growth required for the blue-sky case are supported by real pricing and shipment volumes.
- Whether management confirms the expectation for 2026 data center exabyte growth in the high-60% range, and whether KV-cache-related demand is reintroduced.
- Whether SNDK's discount to the SOX index and to historical peak-cycle P/FE multiples narrows.
- Whether NAND pricing and inventory data from peers such as Micron continue to confirm the upcycle.