Dilikoglu ResearchIDEAS. ANALYSIS. PERSPECTIVE.
AI MEMORY · SKHY

SK hynix stock research & valuation

Model Preview / Research in Progress · Memory manufacturer · HBM / DRAM / NAND

Overview

DILIKOGLU RESEARCH VIEW · MODEL 2026-10-09

Investment summary

Model Preview / Research in Progress · Model-based fair values; no formal recommendation or published price target.

SK hynix research tests HBM leadership and memory pricing against capital reinvestment, cycle normalization and shareholder claims. Values use the Nasdaq ADS basis.

Current model Base fair value / ADS
$244.37
Current model probability-weighted fair value / ADS
$230.16
Key catalyst
HBM4 ramp, Yongin commissioning and LTA cash conversion
Principal risk
Customer qualification and packaging ramps may fall behind demand.

USD per ADS. Valuations are DR model estimates, separate from dated report targets. Scenario probabilities and supporting assumptions appear below.

Latest available price
Unavailable
Base return vs latest price
—

Evidence review →

Primary operating KPI
Memory bit growth, selling prices and cash returns
Last recorded model update
2026-10-09
Explore the analytical model → · Portfolio Analytics →

Investment Thesis

HBM leadership and supply discipline versus cycle normalization

  1. HBM is modeled within DRAM, with NAND and Solidigm consolidated.
  2. Q2 results and management capacity commentary anchor a separate forecast of volume, pricing and cash returns.
  3. Buyback cash, share claims, noncore investments and currency translation reconcile to USD per ADS.

Variant perception

Capacity leadership creates value when volume and pricing generate returns above the capital required.

What could make us wrong

Qualification delays, sharp ASP erosion or capital spending without returns

Key catalysts

HBM4 ramp, Yongin commissioning and LTA cash conversion

Scenario conclusions are Dilikoglu Research estimates. Company evidence, management guidance and analyst assumptions are distinguished in the analytical model.

Scenario Valuation

ScenarioProbabilityFair value / ADSReturn vs price
Failure10%$0.00—
Bear25%$76.25—
Base45%$244.37—
Bull20%$505.67—

Probability-weighted fair value: $230.16

FY2027–2036 FCFF DCF; normalized earnings cross-check

Monte Carlo median: $211.47. Distributional cross-check, separate from the scenario valuation.

Key Base-case drivers · model 2026-10-09

These selected forecast inputs are Dilikoglu Research estimates. Units are shown explicitly; debt funding is not a lender commitment. Reported anchors and management guidance are distinguished in the full model source register.

InputBase estimateUnits
WACC10.00%
Terminal growth3.00%
Terminal ROIC16.00%
Scenario contributions and downside sensitivity

Derived from the current model above. These are scenario-weighted estimates, not calibrated probabilities of future returns.

ScenarioWeightFair valueContribution to PW
Failure10%$0.00$0.00
Bear25%$76.25$19.06
Base45%$244.37$109.97
Bull20%$505.67$101.13

Bull contributes 43.9% of PW. Material upside-tail dependence: the weighted value should not be read as a typical outcome.

Excluding Bull and renormalizing the other weights to 100%: $161.29. Formula: sum of non-Bull weighted values ÷ (1 − Bull weight). This is a conditional diagnostic, not a replacement target.

Company-specific numerical probability calibration is not established. See the operating scenarios and falsifiers in the full analytical model.

Model

Model scope: Scenario-based independent research. Interactive controls are diagnostics; published valuation outputs reflect the published research baseline.

Open full analytical model →

Explore the interactive model here

Research

Model Preview / Research in Progress. No initiation report or formal recommendation has been published.

Revision history

This register begins with the documented baseline below. It is not a complete historical track record. Snapshot values remain fixed; the investment summary reflects the current published model.

· Model refinement

Quarterly cash-flow and source refinement

What changed: Quarterly reconciliation, financing friction, source units and October 8 price reference.

Why: Improve operating and equity-claim transparency without market-price calibration.

Valuation impact: Derived scenario values in USD per ADS; no report recommendation assigned.

Recorded Base fair value
$244.37
Recorded PW fair value
$230.16
Model date 2026-10-09 · USD per ADS · DR estimates

Change from recorded 2026-09-21 entry: Base +0.00; PW +0.00 USD per share. These are differences between recorded endpoints, not isolated causal contributions.

Model provenance

Model SHA-256: 96ac7d9414f7d1d03c60a175ad543bdff511f2d07386807bbf97d780b1b9a6f1

This fingerprint identifies the model file used to record these values.

View current model · Review source disclosures

· Baseline recorded

Dated model baseline

What changed: Recorded scenario valuation baseline.

Why: Historical starting point; the investment summary shows the current model.

Valuation impact: DR estimates, USD per ADS; no recommendation assigned.

Recorded Base fair value
$244.37
Recorded PW fair value
$230.16
Model date 2026-09-21 · USD per ADS · DR estimates
Model provenance

Model SHA-256: 863260dd015b1c34f18df3ca0094896d271631752d6ffdaf6a83c2d9f9f5177e

This fingerprint identifies the model file used to record these values.

View current model · Review source disclosures

Sources

Key Evidence

Filings and contractual documents take priority, followed by company materials and then third-party context. Forecasts are DR estimates.

Evidence review ·

Full Evidence Register

The original model ledger below records its verification status at the model date. The dated review above identifies subsequent checks and remaining limitations.

June consolidated balances and income

Reported · June 30 consolidated statements, not parent-only accounts. KRW millions converted to trillions. Q2 D&A rounded from the earnings deck. Investment gains and dividends are excluded from forecast operating FCFF. Exchangeable bonds fully exercised by June; weighted-average EPS dilution is not a remaining convert claim.

Listing, FX and July offering

Reported · One ADS represents 0.1 ordinary share. June outstanding ordinary shares plus 17.79m newly issued shares. Net offering proceeds before other expenses translated at the July 14 closing FX. September 18 ECB-derived KRW/USD is a dated translation convention, not a live quote. Forecast FX and IPO other costs are estimates.

Buyback and share awards

Dilikoglu Research estimate · Assume full KRW40trn authorized buyback completes by opening 2027 at September 18 Korean ordinary close. Deduct the cash and the corresponding ordinary shares together. This is an estimated completion bridge, not a claim of purchases already executed. Add 1m ordinary award-equivalent claims, stress 0.5–2m. Cash costs retain future SBC; no perpetual buyback benefit.

Q3 volumes and FY26 investment

Management guidance-commentary · Q3 DRAM approximately 10% sequential bit growth, NAND low-single-digit growth. The 3% NAND midpoint and KRW47–49trn interpretation of high-40trn capex are DR arithmetic. No annual 2027 revenue or margin guidance is claimed.

HBM allocation, supply and pricing

Dilikoglu Research estimate · HBM included within DRAM. Q2 DRAM/NAND shares of 73/27% are rounded presentation mix proxies; HBM KRW14trn allocation is not disclosed and stressed at KRW10–18trn. FY27 demand and ASP indexes reference annualized Q2. Shared wafer capacity, node productivity, packaging and NAND constrain shipment demand. MU price-normalization rates, priors and terminal return conventions are shared; SK hynix volume, pricing levels, taxes and capex are separately underwritten.

Opening cash and non-operating investments

Dilikoglu Research estimate · Cash plus short-term financial instruments/investments less borrowings, current tax net of tax receivable, customer advances and claims. Long-term financial instruments added separately; long-term investments and associates receive scenario recovery haircuts, with an explicit KRW15trn realization-tax reserve at Base rather than adding investment gains to earnings. Reported deferred-tax liabilities are covered within that reserve, not deducted twice. Operating lease payments remain in operating costs; leases are not subtracted again as financial debt. Noncontrolling interests and subsidiary award reserve deducted.

H2 bridge and capital commitments

Dilikoglu Research estimate · H2 revenue of KRW200trn, 83% gross margin and KRW14trn opex are research forecasts anchored to Q2 and Q3 bit-growth outlook. H2 PPE cash capex of KRW48trn minus reported H1 capex plus KRW0.7trn intangibles. Scenario H2 revenue factors vary. June KRW61.485trn purchase commitments are scheduled within annual cash-capex floors, not deducted again as debt. No value assigned to unconfirmed Japan, packaging disposals or Solidigm fundraising.

Operating costs, assets and working capital

Dilikoglu Research estimate · Cash product costs allocated from reported Q2 COGS excluding estimated 90% manufacturing D&A. Annual asset cohorts charge D&A once; 7/20-year lives, land/CIP/equipment allocations and commissioning lags are assumptions. Other non-trade payables are not all treated as operating: KRW8trn opening operating-accrual proxy, rest represented in cash capex and claim reserves. A 25% tax rate, future margins and capital floors are forecasts, not reported results. Lease D&A is removed from the FCFF add-back using half of H1 ROU depreciation as a Q2 proxy; recurring lease expense remains in costs. H2 and later D&A are research asset-schedule estimates.

DCF, Monte Carlo and portfolio evidence

Dilikoglu Research estimate · Five operating years plus five normalization years; terminal cash reinvestment NOPAT×growth/ROIC. Correlated cycle, execution, costs, capex, working capital, WACC, FX, HBM allocation, noncore recovery and award uncertainty. Judgment-based probabilities, no calibration to market. USD ADS short history must not be filled with Korean ordinary returns.

Quarterly forecast and financing friction

Dilikoglu Research estimate · FY27 quarterly revenue shares are Failure 27/26/24/23%, Bear 25/25/25/25%, Base 23/24/26/27% and Bull 22/24/26/28%. Cash costs and operating expenses follow revenue shares, D&A is even and cash capex follows 22/24/26/28%. Quarter-end working capital uses annualized activity; Q4 reconciles to the annual closing convention. These are allocation estimates, not quarterly company guidance. Funding discounts of 20/10/5/3% plus 2% issuance costs apply only to incremental liquidity shortfalls. The shortfall itself is already in FCFF; only issuance friction is deducted from equity value.

Solidigm financing

Management guidance-commentary · October 1 clarification states that no financing structure has been determined. Solidigm operating cash flows stay within consolidated NAND; no separate listing proceeds, additional SOTP uplift or speculative dilution are assumed.

Comparison with Micron cycle assumptions

Dilikoglu Research estimate · Scenario probabilities, WACC and long-run ASP decline rates match MU. SK hynix Base/Bull normalization begins at half-rate in FY28; the refined MU model holds FY28 prices before half-rate FY29 normalization. This difference is retained because SK hynix has not disclosed aggregate LTA price floors or annual revenue guarantees comparable to the durability assumed in MU. Company-specific contracts, margins and capacity are not interchangeable.

Company Filings

See the company model’s evidence register for applicable disclosures.

Company Materials

Financing / Counterparty Sources

See the company model’s evidence register for applicable disclosures.

Regulatory / Infrastructure Sources

See the company model’s evidence register for applicable disclosures.

Industry / Market Data

Research updates

Free email updates when we publish new reports or substantive research changes. No fixed sending schedule; unsubscribe at any time.

If the form does not load, use the signup link below.

Subscribe free →

Sign up through Beehiiv and confirm your email address to join. Beehiiv manages your subscription and unsubscribe options. Questions? research@dilikogluresearch.com.