← All reports THE JEFF REPORT CASE FILE 015 · US / Taiwan · 04 Jun 2026
Three-way test
Jeff · Independent Equity Research Interactive Briefing
Three-Way Ranking 4 June 2026 SNDK · SIMO · 8299 TT
SanDisk vs Silicon Motion vs Phison

Three Ways to Own the 2026 NAND Supercycle, Made Interactive

The maker, the merchant, the hybrid. Drag the assumptions, reweight the scenarios, and watch the targets move. Every chart and calculator below is live.

Rank 1 · The investment
NASDAQ: SIMO
Silicon Motion
HOLD conv 3 / 5
PWPT $289 (-4%) · Fwd P/E 34.5x
Asset-light. Survives a reversal.
Rank 2 · The value tilt
TPEx: 8299 TT
Phison
HOLD buy tilt · conv 3 / 5
PWPT flat · Fwd P/E 15.0x
Cheaper. Watch the inventory.
Rank 3 · The trade
NASDAQ: SNDK
SanDisk
HOLD high-beta · conv 2 / 5
PWPT $1,690 (-8%) · Fwd P/E ~10x
Most torque. It is the commodity.
How to read this document. None of the three is a screaming buy today: all sit near highs and at or above their street targets. Ranked on risk-adjusted basis, SIMO is the better investment (asset-light, net cash, a growth engine partly decoupled from the NAND price), SanDisk is the better trade (cheapest multiple, most torque, but it is the commodity itself). The interactive tools below let you stress-test that call yourself.
01

Scenario Lab: Reweight the Probabilities

Each name has a bear, base and bull case. Move the probability sliders (they auto-normalize to 100%) or edit the scenario prices, and the probability-weighted target recomputes live against the current spot. Switch the company with the toggle.

Probability-weighted target
$289
-4% vs spot $301
FIG. 015-A

House case probabilities: SIMO 30/45/25, Phison 30/40/30, SanDisk 30/40/30 (bull/base/bear). The dashed marker on the chart is the current spot price; the solid claret marker is your weighted target.

02

DCF Lab: Build Your Own Fair Value

A forward-earnings DCF. Fair value = projected EPS, grown at your CAGR for N years, capitalized at your exit multiple, discounted back at your WACC, plus net cash per share. Drag the inputs; the implied price, the upside gauge, the EPS projection and the sensitivity grid all update together.

Implied fair value
$290
-4% vs spot $301
Exit-year EPS
$12.3
Undisc. target
$319
Net cash / sh
$6
FIG. 015-B
Sensitivity: implied price by exit P/E (rows) and EPS CAGR (columns). Green = upside vs spot, red = downside.

House assumptions, tuned so each default lands near its base case. SIMO: $8.73 FY26E EPS, 14% CAGR, 3 yrs, 30x exit, 11.5% WACC, $6 net cash. Phison: NT$188, 12% CAGR, 3 yrs, 16x, 12.5%, NT$60. SanDisk: $180 FY27E EPS (peak), 0% CAGR, 2 yrs, 13x, 13%, $0. Dial the exit multiple to where each name actually trades to see how much quality premium is embedded. These are model outputs, not price targets.

03

Segment Economics: Which Parts Make Money

The three earn money in different ways, governed by one question: is NAND flash in your cost of goods? That decides whether a segment has stable, high-quality margin or is a leveraged bet on the NAND price. The mix charts below show where each company's revenue comes from today.

SanDisk: by end-market

Cal Q1 2026 · the NAND maker

FIG. 015-C

Silicon Motion: by product line

Q1 2026 · the fabless controller

FIG. 015-D

Phison: by application

Q1 2026 · controller + module (est.)

FIG. 015-E

SanDisk segment revenue, by quarter

$M · Datacenter is the AI-driven leg, exploding from a small base

FIG. 015-F

Revenue, indexed to 100 at Q1 2025

All three on one scale. SanDisk and Phison ramp hardest

FIG. 015-G
The synthesis. The highest-quality money is the bare controller IC (SIMO's chips, Phison's chips): no NAND in COGS, so ~50-70% margins that do not collapse when the cycle turns. The biggest money right now is the enterprise module (Phison 61%, inside SanDisk's 78%), but that is a NAND-spread profit rented from peak ASPs. The worst money is consumer / retail, which is why Phison is shrinking it. SIMO earns on chips (stable), Phison on chips plus a levered NAND-inventory spread (hybrid), SanDisk entirely on the NAND price (pure cyclical).
Full segment tables and descriptions

SanDisk (cal Q1 2026): $5,950M revenue, 78.4% gross margin

SegmentRev $M% mixYoYWhat it is
Edge3,66361.6%+295%SSDs / embedded NAND for PC & phone OEMs
Datacenter1,46724.7%+645%Enterprise SSDs for hyperscalers / AI
Consumer82013.8%+44%Retail cards, USB, branded SSDs

Bit shipments were flat YoY and down high-teens QoQ. Revenue nearly doubled on ASP alone (price per gigabyte up ~130% QoQ). Fixed-cost fabs mean a rising NAND price drops almost entirely to gross profit, and reverses just as hard.

Silicon Motion (Q1 2026): $342.1M, 47.2% GM

SegmentEst. mixYoYNAND cost?What it is
eMMC + UFS controllers~47%+140%NoChip for phone / IoT / auto storage
Client SSD controllers~38%+43%NoNVMe controller chip in PC SSDs
SSD solutions (Ferri, boot, MonTitan)~13%+755%YesFinished drives SIMO assembles

Phison (Q1 2026): NT$40.97B, 61.3% GM

SegmentEst. mixGrowthNAND cost?Margin
AI / enterprise modules (Pascari)~38%10 to 38%Yes (cheap LTA NAND)Highest now
Client SSD controllers~50% share+490% shipNo60-70%
Mobile / eMMC / UFS controllerslow single %+12xNoHigh
Retail / consumer modules<10%, shrinkingdecliningYes (spot NAND)Low ~20-30%
04

The Margin Divergence

One chart tells the whole quality story. Over the last five quarters, SanDisk's gross margin rocketed from a 22% trough to 78% as NAND ASPs exploded; Phison's climbed from 29% to 61% on its enterprise mix shift; SIMO's stayed roughly flat in the high-40s, because its controller-IP margins do not swing with the NAND price. Flat is the feature, not the bug.

Gross margin by quarter, calendar-aligned

SanDisk is the cyclical (it is the NAND price). SIMO is the stable IP house. Phison sits between.

FIG. 015-H

SanDisk figures are calendar-mapped from its fiscal quarters. Q1 2025 SanDisk 22.5% was the cycle trough; Q4 2026 guidance is 79-81%.

Margin by revenue type

Where the durable money is versus the rented money

FIG. 015-I

Controller IC margin is stable across the cycle. Enterprise-module margin is high now but cyclical. Consumer-module margin is structurally low.

05

Market Structure & Share

Two different competitive maps matter here. In raw NAND, SanDisk is the smallest of five scale makers. In merchant controllers, Phison and Silicon Motion are a duopoly that the NAND makers increasingly outsource to. And the NAND ASP line below is the single variable that drives every earnings model in this report.

NAND maker share

Q4 2025 · SanDisk plays here, as the smallest of five

FIG. 015-J

Client SSD controller share

SIMO and Phison play here, as a duopoly-plus-one

FIG. 015-K

NAND ASP: 1Tb TLC contract price ($)

The whole thesis in one line. Doubled in 6 months, with steep further increases guided

FIG. 015-L

Jul 2025 $4.80 to Nov 2025 $10.70 (+123%). Q1 2026 contract +50-60% QoQ, Q2 2026 guided +70-75% QoQ (TrendForce). The Q1/Q2 2026 points are projections from those guided step-ups.

06

Valuation: Cheap on Peak Earnings

Plot forward P/E against expected revenue growth and the trap is visible: the fastest-growing, cheapest-looking names (SanDisk, Phison) are the most cyclical, and the low multiple is the market discounting peak earnings, not a free lunch. Bubble size is market cap.

Forward P/E vs revenue growth

Lower-left looks cheapest. It is also where the cyclical risk concentrates. Bubble = market cap

FIG. 015-M
Peer multiples, early June 2026
CompanyMkt capFwd P/ERev growthGross mgnModel
Silicon Motion$10.1B34.5x+83%47%Fabless controller
Phison$19.2B15.0x+219%61%Controller + module
SanDisk$267B~10x+162%78%Integrated NAND maker
Micron (ref)$1.20T11.3x+67%58%Integrated DRAM + NAND
07

The Call, in Words

Pick SIMO (rank 1)

  • The only one with a company-specific growth engine (BlueField / MonTitan) partly decoupled from the NAND spot price.
  • Net cash, no NAND inventory, controller-IP margins that survive a downturn.
  • The cost: you pay 34.5x and it trades above its street target. Limited near-term upside.

Pick SanDisk (rank 3, the trade)

  • Cheapest multiple (~10x) and the most direct torque to rising NAND ASPs.
  • But it is the commodity: 78% margin at the top, 22% at the bottom, GAAP losses as recently as FY2025.
  • Highest bear-case downside of the three. Size it as a satellite, not a core.
Phison sits between: the cheapest controller exposure and the best probability-weighted number, but carrying a NT$72B NAND-inventory tail that the other two do not.
If you must own one. For a core, hold-it-and-sleep position today, SIMO. For the highest expected return conditional on the supercycle extending, SanDisk, sized small. The cleanest institutional expression is a long-SIMO core, a SanDisk torque sleeve, and Phison only as a valuation-gap pair against SIMO.
08

Risk Matrix

RiskTypeProbImpactApplies
NAND ASP reversal in H2'26/27 collapses EPSCyclicalMHAll
NAND wafer COGS inflation crushes module marginFinancialMHPhison, SNDK
NT$72.2B inventory mark-down if NAND reversesFinancialMHPhison
Operating deleverage swings SanDisk back to lossesCyclicalMV.HIGHSanDisk
NAND makers insource controllers (Samsung RISC-V)ExecutionL-MHSIMO, Phison
Valuation: all near highs, at/above targetsMarketHMAll
Taiwan Strait conflict; TSMC dependencyGeopoliticalLV.HIGHSIMO, Phison
Hedge-fund semi exposure at ATH; insider sellingSentimentMHAll
09

Catalysts & Monitoring

Near-term (0-6m)

  • Phison monthly revenue prints (next early July), the highest-frequency tell. Binary.
  • Phison Q2'26 earnings 7 Aug 2026: gross-margin and inventory line. Binary.
  • SIMO Q2'26 vs the $393-411M guide. Positive-leaning.
  • SanDisk Q4 FY2026 print: GM guided 79-81%, and the $42B backlog. Positive.
  • TrendForce Q3'26 NAND contract-price guidance. Binary, the master variable.

Medium & long term (6m+)

  • SIMO MonTitan ramp to 5 Tier-1 CSPs; PCIe Gen6 tape-out Q3'26. Positive.
  • Phison enterprise mix crossing >50% of revenue. Positive.
  • Consumer demand elasticity breaking (PC units down 10%+). Negative.
  • New NAND fab capacity arriving late 2027/28, the cycle's endpoint. Negative when it lands.