TurboQuant panic is seen as overdone; Bernstein upgrades WDC but still prefers STX and SNDK

Overview

This is a Bernstein rating review on U.S. IT hardware, HDD, and memory companies. The core backdrop is that after Google Research released TurboQuant on March 24, 2026, the market worried that the compression algorithm could reduce storage demand, causing WDC, STX, and SNDK to pull back sharply. The report argues that this concern is exaggerated: TurboQuant mainly affects inference-stage KV cache and does not involve training, model weights, or static data storage, so it will not affect HDD demand and will only have a limited indirect impact on NAND.

Core views

The main message of the report is: 'short-term panic creates a buying opportunity, while long-term AI storage demand remains strong.' Bernstein upgrades WDC to Outperform because valuation and the share-price pullback now offer a better risk-reward profile, and raises the target price to $340. Relative to peers, however, the report still prefers STX, because STX's leadership on the HAMR roadmap is expected to translate into faster cost declines and higher margins starting around FY27-FY28. SNDK also remains Outperform, supported by the NAND cycle and the valuation framework.

Analysis framework

The report combines event-driven analysis, industry supply-demand forecasting, company technology roadmaps, earnings-model upgrades, and P/E-based valuation multiples. At the event level, it assesses TurboQuant's real impact on HDD and NAND demand; at the industry level, it updates assumptions for data center storage, HDD bits shipped, and ASPs; at the company level, it compares WDC's dual-track ePMR/HAMR strategy with STX's HAMR lead; and at the valuation level, it derives target prices from FY28 or FY27 EPS and target multiples.

Methodology notes

• Event impact assessmentTurboQuant demand impact decomposition

  Limit the compression algorithm's impact to KV cache, GPU HBM, and system DRAM, and judge how that flows through to storage-media demand.

  The report believes TurboQuant does not compress static data, training data, or model weights, so HDD demand is unaffected; NAND may only see limited indirect impact from reduced cold-cache offload.

• Industry supply-demand modelData center storage CAGR and ASP assumptions

  Update revenue forecasts using assumptions for total data center storage, HDD bits shipped, NAND pricing, and HDD ASPs.

  The report estimates 2024-2030 data center total storage demand CAGR at 31%, data center HDD demand CAGR at about 25%, and adjusts HDD ASP assumptions from declines to stability and even slight increases.

• Technology roadmap comparisonHAMR vs ePMR cost curve

  Compare the long-term unit-cost, gross-margin, and operating-margin implications of STX's HAMR and WDC's ePMR/HAMR dual-track strategy.

  The report thinks WDC's short-term ePMR extension is helpful, but it may also imply that HAMR scaling is delayed; STX's HAMR lead is expected to bring stronger areal-density gains and long-term cost advantages.

• Valuation methodP/E target multiple valuation

  Use future-year EPS and a target P/E multiple to derive target price.

  WDC's $340 target price is based on 20x FY28 EPS; STX's $620 target price is based on 21x FY28 EPS; SNDK's $1,000 target price is based on 11x FY27 EPS and an FCF/EV framework.

Asset mapping & comparison

Structured mapping from thesis to named assets (strengths, weaknesses, peers, risks).

• WDC

  Core upgraded name

  Strengths

  The recent share-price pullback makes valuation more attractive; the ePMR route is stronger in the near term, and revenue, ASP, and EPS estimates have all been raised.

  Weaknesses

  HAMR commercialization may be delayed, and long-term cost and margin improvement may lag STX.

  Comparison

  Compared with STX, WDC's near-term ePMR path is steadier, but its long-term HAMR cost advantage is less clear.

  Risks

  Cloud provider capex digestion, NAND substitution for HDD, and HAMR transition pressure on gross margin and EPS.

• STX

  Report's preferred HDD name

  Strengths

  HAMR leadership, stronger long-term areal-density and cost-reduction potential, and a more obvious margin advantage after FY27.

  Weaknesses

  The investment case takes longer to play out and still depends on successful HAMR scaling and continued cloud demand.

  Comparison

  The report believes STX is better positioned than WDC to benefit from the HAMR cost curve, and FY28 EPS is expected to be materially higher than WDC's.

  Risks

  Cloud procurement slowdown, WDC catching up with HAMR more quickly, and NAND technology progress eroding HDD share.

• SNDK

  NAND beneficiary and maintained Outperform

  Strengths

  An improving NAND price cycle and a higher valuation framework support the target price.

  Weaknesses

  Company disclosure and investor communication are viewed as somewhat messy, and near-term earnings are highly sensitive to the cycle.

  Comparison

  Compared with the HDD names, SNDK has more direct exposure to the NAND pricing cycle.

  Risks

  A downcycle in NAND, overly aggressive near-term forecasts, and NAND weakness shifting from cyclical to structural.

• HDD industry

  Core beneficiary of AI data growth

  Strengths

  A clear unit-storage-cost advantage versus NAND, data center and AI demand driving bits growth, and higher industry concentration supporting ASPs.

  Weaknesses

  Demand is highly dependent on hyperscale cloud providers, and technology roadmaps and customer procurement timing have a large impact.

  Comparison

  Compared with NAND, HDD has a strong cost advantage for large-scale cold and warm data storage.

  Risks

  NAND technology improvements, cloud capex volatility, and changes in customer procurement patterns.

Key data

• TurboQuant technology impactKV cache memory reduction 6x; inference performance gains up to 8x on Nvidia H100The report emphasizes that this technology does not compress static storage data, so it should not weaken HDD demand.
• Stock pullbackWDC -21%; STX -17%; SNDK -26% from recent highsMost of the decline occurred after the TurboQuant blog post was released on March 24, 2026.
• Data center HDD demand2024-2030E CAGR about 25%Driven by AI workloads, content generation, data retention, and data sovereignty requirements.
• Total HDD bits shipped2024-2030E CAGR about 23%Device-related HDD growth is slower, but data center demand is becoming the main driver.
• WDC rating and target priceOutperform; TP $340Upgraded from Market-Perform; target price is based on 20x FY28 EPS, with FY25-FY30E EPS CAGR of about 46%.
• STX rating and target priceOutperform; TP $620Maintained Outperform; target price is based on 21x FY28 EPS, with FY25-FY30E EPS CAGR of about 51%.
• WDC EPS upward revisionsFY27 EPS $12.68; FY28 EPS $17.19Adjusted FY27 and FY28 EPS were raised by 45.8% and 80.7%, respectively.
• STX EPS upward revisionsFY27 EPS $19.52; FY28 EPS $29.50Adjusted FY27 and FY28 EPS were raised by 10.7% and 25.7%, respectively.
• Long-term margin comparisonFY30E STX gross margin 65% vs WDC 60%; operating margin 57% vs WDC 47%The report attributes the gap to STX's cost advantage from its HAMR lead.

Impact & implications

If the report's judgment is right, the sell-off in storage stocks triggered by TurboQuant looks more like a sentiment shock than a deterioration in fundamentals, improving the risk-reward profile for WDC, STX, and SNDK. For portfolios, the report suggests using the pullback to add exposure to the storage hardware chain, but with a preference ranking that still favors STX because its HAMR lead is more likely to create medium- to long-term cost and margin advantages; although WDC is upgraded, its long-term HAMR transition pace still needs to be validated.

Risks

• If hyperscale cloud providers enter a digestion phase for capex, or if procurement patterns change, HDD demand may come in below expectations.
• Technological progress in NAND or a larger-than-expected price decline could take additional storage share away from HDD.
• If WDC encounters major technical or production ramp issues in its HAMR transition, gross margin and EPS could be pressured.
• If STX's HAMR lead is caught more quickly by WDC, market-share gains and margin upside could narrow.
• SNDK's near-term earnings estimates may be hurt by a downcycle in NAND, and disclosure and communication issues may weigh on valuation.

What to watch

• Whether WDC's HAMR qualification, mass production, and commercialization progress lags its roadmap.
• Whether STX's HAMR shipment mix, unit cost, and nearline HDD gross margin improve as expected in FY27-FY28.
• Whether data center HDD exabyte growth continues at a mid-20s pace.
• Whether HDD ASPs remain stable, or rise 1%-2% quarter over quarter in CY26.
• Whether the NAND price cycle continues to improve and changes the substitution relationship with HDD.
• Whether AI capex and storage procurement by cloud providers shows signs of digestion or delay.