闪迪：估值和股价能涨多高？上调估值及目标价至1250美元，乐观情景下可达3000美元

Sandisk: How high can estimates and the stock go? Raisingestimates and TP to $1,250, with blue-sky scenario at $3000 With the backdrop of accelerating NAND price hikes and worries around Turboquant and Mark C. Newman+1 212 845 7822mark.newman@bernsteinsg.com peak earnings, where to from here? SNDK reports FQ3’26 on April 30th. April Li+1 917 344 8339april.li@bernsteinsg.com here), Turboquant fears were overdone. We go into more detail here on potential impactfrom Turboquant, which can have some limited impact on NAND but potentially more onDRAM and zero impact on HDD. Overall though we are believers in Jevons paradox and thatoverall cheaper and more efficient compute will generally drive demand for more compute Phoebe Sun+1 917 344 8481phoebe.sun@bernsteinsg.com We raise SNDK estimates again significantly on better than expected NAND prices. As explained in our recent MEMORY TRACKER (March), memory prices continue to surpriseto the upside with NAND showing the strongest increases and continued acceleration. Webump up ourFY27 non-GAAP EPS to $144 and $224 in our base case and bull case For FQ3’26 (March quarter) we are now at $14.18 vs. consensus at $13.99 andguidance of $12-14 (non-GAAP EPS).The main opportunity for the beat is the ASP whichwe have at +55% and see further upside to this number with our bull case at $17.47, whichassumes NAND ASP up 75% (Micron’s Feb qtr had NAND prices up high 70s%). More importantly for the FQ4’26 guide we are at $25.30 (with a +40% QoQ ASPmove) vs. street at $18.78 (non-GAAP EPS).Plugging in a more aggressive ASP moveof 75% in each FQ3 and FQ4 gets us to $40.53 non-GAAP EPS for FQ4’26 which wouldtranslate into $224 for FY27 showing just how much potential there is for a strong guide.While some data points to ASPs up as much as 100% QoQ, we doubt that would apply to What is the right multiple?Analyzing previous cycles (pre WDC acquisition in 2016),SNDK typically trades 10-13x peak cycle EPS (on 2 year forward estimates). Currently,SNDK is trading at 9.0x consensus, 5.4x and 3.5x on our base and bull cases (respectivelyall on FY27 EPS). This suggests that the market does not believe the higher side of ourestimates are reasonable and/or that there will be an imminent collapse in NAND prices Increasing TP to $1,250, with blue-sky scenario at $3,000.We think the market issignificantly undervaluing earnings power and sustainability of this cycle. Our base-casevaluation applies 11x through-cycle earnings, defined as the average of FY26-29 Bern.base-case EPS of $114, and corresponds to 8.7x FY27 base-case EPS of $144, in-linewith the current multiple on consensus and a level we view as conservative relative tohistorical peak-cycle multiples. Our $1,250 TP is also just 5.5x our bull-case FY27 EPS of BERNSTEIN TICKER TABLE Source: Bloomberg, Bernstein estimates and analysis. INVESTMENT IMPLICATIONS We value SanDisk at 11x our FY26-29 average EPS, or 8.7x FY27 EPS which gives our target price of $1,250. We value Seagate at 21x our FY28 EPS of $29.5, implying $620 per share. We believe Seagate's improving fundamentals andHAMR leadership easily justifies a 21x P/E, and arguably much higher. Our TP for WDC is $340, which is 20x our FY28 EPS. PE of 20x is inline to lower than recent market multiple and arguablyconservative given its 5 year EPS CAGR of 46% (FY25-30E) Table Of Contents Part I: TurboQuant and How does it affect NAND, DRAM and HDD?....................................................................................................................3Part II: Increasing estimates on further NAND price acceleration..........................................................................................................................5Part III: What is the right multiple?.................................................................................................................................................................................... 7Part IV: How high can SNDK go?.................................................................................................................................................................................... 10DETAILS PART I: TURBOQUANT AND HOW DOES IT AFFECT NAND, DRAM AND HDD? As explained in our recent note (see here), Turboquant fears were overdone.We go into more detail here on potentialimpact from Turboquant, which can have some limited impact on NAND but potentially more on DRAM and no impact on HDD.Overall though we are believers in Jevon’s paradox and that overall cheaper and more efficient compute will generally drivedemand for more compute (and thus more memory and storage). Following Google’s unveiling of TurboQuant, a KV-cachecompression algorithm claiming up to a 6x reduction in memory footprint, SNDK shares fell 19%, or roughly 25% from theirYTD peak (STX down 17% from peak; WDC down 21%, Exhibit 1). We continue to view the NAND outlook as constructive, TurboQuant is narrowly applicable to the K