Silicon Motion Technology Corporation (NASDAQ: SIMO) — The Indispensable Middleman of the Memory Super-Cycle, Priced as if the Cycle Were Repealed

An independent equity research note · 2026-06-27

⚡ Claude’s Take

This is the author’s own independent opinion and general information, not investment advice. The detailed analysis that follows is deliberately position-free and carries no price target; this opening section is the single place a view is expressed.

Verdict: HOLD / AVOID-here / accumulate only on a cycle reset. Great execution, full-to-rich price. Not a short. Conviction: medium.

Silicon Motion is, right now, the best-positioned independent NAND flash controller designer in the world, and it is riding a genuine, once-a-decade memory super-cycle in which NAND prices rose 55–60% in a single quarter and the flash makers are so busy chasing HBM/DRAM/enterprise AI margins that they are handing SIMO ever more of the merchant-controller market on a plate. The Q1-2026 print — record revenue of $342M, +105% year-over-year, with a Q2 guide to grow another 15–20% sequentially — is the real thing, and the new enterprise/cloud-AI vectors (the MonTitan eSSD controller, boot drives qualified into NVIDIA’s DPU/NVLink/Ethernet-switch platforms, automotive Ferri) give it, for the first time in its history, a credible non-cyclical TAM. This is not a story stock; it is a real business firing on every cylinder.

But the share price has done something the income statement has not yet earned. At ~$305 the stock trades at ~9.5x trailing sales, ~60x trailing GAAP earnings and ~12x tangible book — every one of them the richest multiple in SIMO’s decade as a public company — for a fabless price-taker that controls neither the supply nor the price of its single most important input, whose five largest customers ARE the same NAND makers that could re-insource controllers, and whose last down-cycle (2023) saw EPS collapse 69% to $1.59. The 8x move off the April-2025 low ($37.89 → a $336.90 all-time high) is roughly half earnings-recovery, half multiple-expansion; the multiple half is now underwriting a permanent super-cycle and a flawless enterprise pivot simultaneously. That is two bold “must-be-trues” stacked on a business that, four years out of the last ten, earned less than half its peak. Framing: a late-cycle momentum compounder, not a value name — the FactorsToday tape confirms it (beta ~1.5, lifetime max drawdown −93%, +363% trailing-year, near-zero Value loading). My directional zone: fair value ~$150–230 on a normalized ~$6–9 mid-cycle EPS at 18–26x; I’d want to be accumulating in the ~$120–170 band that a cycle wobble would hand you, not chasing $300+. Tag: “the toll-collector on a flood — wonderful while it rains.”

What flips me bullish: durable evidence that MonTitan + boot-drive enterprise revenue is structurally scaling toward and past the 20% diversification target with the cycle flat (i.e., growth that survives the next NAND down-leg). What flips me bearish: the classic NAND turn — a bit-supply normalization in 2027-28 that craters ASPs and controller volumes while SIMO sits on a $420M+ inventory book, or a major NAND maker re-insourcing client/UFS controllers.

📈 Stock Price Action — Five-Year Event Map

Over five years SIMO has traced a full cyclical wreck-and-resurrection: from ~$59 (mid-2021) up to the high-$80s in the 2021 memory boom, into a multi-year grind down to a $37.89 low in April 2025, and then a near-vertical ~8x melt-up to a $336.90 all-time high on 22-Jun-2026. It now sits at $305.28 (26-Jun-2026), ~9% off the high, with a 52-week range of roughly $70 → $337. The stock is at, or within a hair of, the most expensive point in its public life — on a tape that has gone almost straight up for twelve months. (Price moves are FACT; attributed causes are INTERPRETATION.)

Cycle narrative. SIMO entered the period as a profitable but unglamorous controller vendor (1). MediaTek’s May-2022 bid (2) should have been a clean exit at ~$114/ADS, but the stock never believed it — China’s antitrust regulator (SAMR) never cleared the deal in the worsening US–China climate, and MediaTek walked in July 2023 (4), paying a $160M break fee. Through that whole window the NAND market was sliding into one of its periodic gluts (3), and SIMO’s revenue fell from $946M (2022) to $639M (2023) with EPS down 69%. The capitulation low — $37.89 in April 2025 (5) — coincided with a broad tariff/recession scare layered on the memory trough. From there, the AI build-out flipped NAND from glut to acute shortage; prices spiked, the flash makers pivoted capacity to high-margin HBM/DRAM/enterprise, and SIMO’s revenue went vertical (6–8). The +105% YoY Q1-2026 print and a Q2 guide for another 15–20% sequential gain, plus the MonTitan/boot-drive enterprise narrative and a string of sell-side target hikes, carried the stock to a $336.90 all-time high (8) before a modest pullback to $305.

1. Executive Summary

Silicon Motion Technology Corporation is the largest independent (merchant) supplier of NAND flash controllers — the processor-and-firmware brains that sit between a NAND flash array and the host system in an SSD, an embedded mobile storage chip (eMMC/UFS), or a specialized industrial/automotive drive. It is a Cayman-incorporated, Taiwan-headquartered fabless designer that files as a U.S. foreign private issuer (Form 20-F), with ADSs listed on Nasdaq (1 ADS = 4 ordinary shares). Founded around its controller IP and led since inception by founder-CEO Wallace Kou, it has roughly $886M of 2025 revenue across three lines: SSD controllers (45–50% of 2025 sales), eMMC/UFS embedded controllers (40–45%), and SSD solutions / modules (0–5%).

The investment question is not whether SIMO is a good business in this moment — it plainly is. Revenue hit a record $342M in Q1-2026 (+105% YoY); gross margin is a healthy 47–49%; the balance sheet is net cash with essentially no debt; ROE and ROIC were both ~27% in 2025 and are climbing. The question is whether the price — ~$10.2B of market value, ~9.5x trailing sales, ~60x trailing GAAP EPS, ~12x tangible book, all the richest in the company’s public history — is paying for a normalized business or for the peak of a cycle that has, historically, always turned.

The bull case is unusually credible for a cyclical: a real, AI-driven memory super-cycle that management expects to persist into 2027; genuine, durable market-share gains as cash-constrained NAND makers exit the low end and outsource more controllers to SIMO; and a new secular leg — enterprise/cloud-AI storage via the MonTitan controller and boot drives designed into NVIDIA’s data-center platforms — that for the first time gives SIMO a structural, non-consumer TAM. The bear case is equally clear: SIMO is a price-taker on NAND, with no control over the input that drives its own demand and ASPs; its largest customers are the very NAND makers (Micron, Kioxia, SK Hynix) who could re-insource controllers; its earnings are flattered by ~$90M/year of below-operating-line income (FX, government subsidies, interest); its 2025 free cash flow collapsed to ~$6M as inventory ballooned $215M; and the last down-cycle cut EPS 69%. At the current multiple, the cycle has effectively been priced out of existence.

Bottom line: a high-quality execution story trading at a low-quality-of-margin-of-safety price. The structural improvements are real; the valuation already capitalizes them as permanent. The institutional sections below quantify both sides and assign no recommendation.

2. Business Overview

2.1 What SIMO sells

A NAND flash controller is the SoC plus firmware that manages a flash storage device: it handles error correction (LDPC), wear-leveling, garbage collection, the flash translation layer, host interface protocol (PCIe/NVMe, UFS, eMMC, SATA), security, and increasingly performance-shaping for AI workloads. NAND flash itself is a brutal commodity; the controller is where the differentiation, the firmware IP, and the margin live. SIMO designs the controller, licenses/sells it (often with firmware) to NAND makers and module makers, and in some cases sells a more complete “SSD solution.” It is fabless — wafers are fabricated by foundries (TSMC and others) and packaged/tested by OSAT partners.

The three reportable product families (2025 revenue mix in parentheses):

• SSD controllers (45–50%). Two sub-segments. Client/edge SSD controllers go into PCs, gaming consoles, and workstations; SIMO’s eight-channel PCIe Gen5 controller leads on performance, and a four-channel DRAM-less PCIe Gen5 controller (launched Dec-2025) is targeted to become the volume PCIe-5 leader as the standard moves mainstream. Enterprise SSD controllers — the MonTitan family (PCIe Gen5 now, Gen6 4nm taping out Q3-2026) — target data-center and AI-infrastructure storage. SIMO claims >50% share of merchant PCIe-5 client controllers.
• eMMC and UFS controllers (40–45%, up from 25–30% in 2023). Embedded mobile storage controllers for smartphones (mostly UFS) and a large, growing array of “smart device” markets (eMMC): smart TVs, set-top boxes, smart glasses, smartwatches, IoT, smart door locks, automotive. This is the fastest-growing legacy line, driven almost entirely by share gains as flash makers exit.
• SSD solutions (0–5%, shrinking). Module-level and specialized solutions, including the Ferri industrial/automotive embedded line and the emerging enterprise boot-drive business (SM8000-series controllers).

2.2 How it makes money — and the cyclicality embedded in that

SIMO sells controllers (and firmware) priced per unit; revenue is volume × ASP. Both inputs are cyclically sensitive: in a NAND up-cycle, device demand and SSD content rise, mix shifts to higher-ASP high-end controllers (PCIe-5 vs PCIe-4, UFS vs eMMC), and revenue compounds; in a glut, the reverse. Crucially, SIMO does not primarily make money on the NAND itself — it is not a memory maker — so rising NAND prices are not directly a revenue tailwind. The benefit is second-order: AI demand → NAND shortage → flash makers reallocate to high-margin enterprise/HBM and exit low-end consumer and merchant controller development → SIMO wins outsourced and module-maker business. That is the mechanism management leans on, and it is real, but it is a consequence of the cycle, not a hedge against it.

Revenue is recurring only in the loose sense that design wins persist across a product generation; there is no subscription, no contracted backlog of the SaaS kind, and visibility is a quarter or two. Revenue concentration is high: the five largest customers were 66% of 2025 sales, and the >10% customers in 2025 were PHISEMI, Kioxia, AFASTOR, and Micron. Geographically, 99% of 2025 revenue was ex-US and 79% came from just three jurisdictions (China, Japan, Singapore), with operating expenses largely in New Taiwan dollars — a structural FX exposure.

Verdict. A genuinely differentiated product (firmware/controller IP, margin-rich) sold into a brutally cyclical commodity end-market (NAND), with high customer concentration and a customer base that doubles as its most dangerous potential competitor. Good business, structurally exposed business model.

3. Industry Dynamics

3.1 Structure of the NAND controller market

NAND flash is supplied by an oligopoly of six players — Samsung, SK Hynix (incl. Solidigm), Kioxia, Micron, plus the smaller Chinese entrant YMTC and Western Digital/SanDisk’s flash JV lineage. Each of these makers needs controllers for its finished products (SSDs, eMMC/UFS). They can build controllers in-house (captive) or buy from merchant suppliers. SIMO is the dominant independent merchant controller vendor; its principal merchant competitors are Phison (Taiwan, the other big merchant) and, in pockets, Marvell (enterprise SSD controllers), Realtek, and a handful of smaller players. The captive controllers of Samsung, SK Hynix, and Micron are the real structural ceiling — they represent demand that, by definition, SIMO cannot address.

The size of the merchant opportunity is therefore a function of the make-vs-buy decision across the NAND makers, and that decision is currently swinging hard in SIMO’s favor. The reason is capital allocation inside the memory industry: the AI build-out has made HBM and high-margin enterprise NAND/DRAM so lucrative, and supply so scarce, that the flash makers are concentrating their finite R&D and wafer capacity on those segments and withdrawing from low-margin consumer controller development — outsourcing it to SIMO and Phison, and even monetizing raw wafers to module makers who then rely on SIMO firmware. This is the single most important industry dynamic for the thesis: it is a genuine, supply-side-driven secular share shift, and it is happening on top of a cyclical price spike.

3.2 The capital cycle (Marathon lens)

NAND is a textbook capital-cycle commodity: high returns attract capacity, capacity gluts the market, ASPs collapse, capacity is cut, shortage returns, repeat. SIMO is one cycle-derivative removed from this — it doesn’t own fabs — but its fortunes track the same wave. We are currently at the shortage phase: NAND prices +55–60% QoQ in Q1-2026, allocation finalized by the makers before mid-2025, and 2027 expected (by management) to be tighter than 2026, with relief not arriving until new mega-fabs ramp in late 2027 (DRAM) / 2028 (NAND). Marathon’s framework says exactly this kind of euphoric, high-return phase is when capital floods in and sows the next glut. The bull retort — that AI demand is a structural step-change, not a normal cyclical wiggle — is plausible but unproven; every prior memory boom was also narrated as “different this time.” The prudent base case treats the current super-cycle as a cycle, elongated perhaps, but not abolished.

3.3 Regulation and geopolitics

SIMO is a Taiwan-centric supply chain (controller R&D in Taiwan; SSD-solutions R&D in Taiwan and China) selling overwhelmingly into Asia. The structural risks: US–China export controls (which already helped sink the MediaTek deal via China’s SAMR), tariffs (the April-2025 panic was tariff-driven), and Taiwan-Strait geopolitical tail risk. None is acute today, but all are live, and SIMO is more exposed than a US-domestic peer.

Verdict: structurally attractive at this moment in the cycle, structurally treacherous across the cycle. The merchant-controller share shift is a genuine secular positive; the underlying NAND market is one of the most violently cyclical in technology, and SIMO has no control over it.

4. Competitive Position

4.1 The moat — narrow, real, and of a specific type

Naming the mechanism in Greenwald’s taxonomy: SIMO’s advantage is a blend of (a) firmware/controller intangibles (two decades of accumulated IP across NAND generations, qualification track records, automotive-grade reliability) and (b) a relationship-and-scale advantage with the NAND makers that is closest to a cost/scale edge with a sliver of customer captivity. It is not a network effect, and it is not a wide scale-with-captivity moat of the kind that protects a dominant platform.

The most defensible piece is the one management emphasized repeatedly on the Q1 call: the ability to secure NAND allocation in a shortage. Because SIMO has been a strategic partner to the flash makers for 20+ years and engages their projects with large downstream customers, it can source NAND from three different makers even when supply is rationed — which lets it keep its module-maker and AI-storage customers supplied when rivals cannot. In an allocation-constrained world, being the one who can get the NAND is a genuine, monetizable edge. The firmware IP is the second piece: a NAND maker exiting consumer controller development cannot trivially replace SIMO’s qualified, multi-generation firmware, especially for QLC management, automotive reliability, and AI performance-shaping (the MonTitan “performance shaping” / multi-token KV-cache architecture is a concrete technical differentiator management cites for AI inference).

4.2 The pressure tests

• Switching costs: moderate. A NAND maker or module maker that has qualified SIMO’s controller and firmware for a product generation faces real re-qualification cost to switch — but the switch up (in-housing) is exactly what the captive ceiling represents, and the makers can and do choose make-vs-buy generation by generation.
• Does the advantage show in the numbers? Partially. Gross margin is a stable 45–49% across the cycle — a controller business, not a commodity — which is evidence the IP has value. But operating margin and ROIC swing from ~27% (2021) to ~6%/12% (2023), which tells you the moat protects gross economics, not through-cycle returns; the cyclicality overwhelms the moat at the operating line.
• The captive ceiling: the single biggest competitive risk. The makers outsource when it suits their capital priorities (today: chase AI margins) and could re-insource when consumer controllers become strategically interesting again. SIMO’s share gains are partly a gift of the current capex priorities of its customer-competitors — a gift that can be withdrawn.
• Phison: the direct merchant rival, of comparable scale, keeps the market competitive and caps pricing power. SIMO’s claimed >50% PCIe-5 client share is real but contested.

Verdict: a durable but narrow advantage — enough to earn a premium gross margin and to win the outsourcing wave, not enough to abolish cyclicality or to defend against a customer that decides to compete. The moat is real; it is not wide, and it is partly contingent on its customers’ choices.

5. Growth History and Forward Opportunities

5.1 The historical record

Revenue (US$M): 457 (2019) → 540 (2020) → 922 (2021) → 946 (2022, prior peak) → 639 (2023, −32%) → 804 (2024, +26%) → 886 (2025, +10%) → Q1-2026 $342M (+105% YoY). EPS/ADS tracked the cycle even more violently: $5.74 (2021) → $5.19 (2022) → $1.59 (2023, −69%) → $2.65 (2024) → $3.65 (2025) → quarterly run-rate now ~$1.6–2.0 and climbing. This is the defining fact about SIMO’s growth: it is not a smooth compounder. The 2021-22 peak was followed by a near-halving of earnings, and the current surge is off a deeply depressed 2024-25 base. Anyone extrapolating the Q1-2026 +105% is extrapolating from a trough.

That said, the secular trend underneath the cycle is up: trough-to-trough and peak-to-peak, revenue and the embedded/eMMC-UFS franchise have grown, driven by SSD adoption, rising storage content per device, and merchant share gains. The eMMC/UFS line going from 25–30% of sales (2023) to 40–45% (2025) is the clearest evidence of structural share capture independent of price.

5.2 Forward vectors

1. The current super-cycle ramp (cyclical + share). Management guides Q2-2026 to $393–411M (+15–20% QoQ) and “sequential growth across the portfolio” for the rest of 2026, a “record revenue year,” with full-year operating margin up vs 2025. The drivers: mobile eMMC/UFS up 30–35% QoQ and 140%+ YoY on share gains; client SSD benefiting from PCIe-5 mix-up and share as makers exit retail; eMMC into a 900M+ unit/year smart-device market where “major flash makers are essentially gone.”
2. MonTitan enterprise/cloud-AI SSD controllers (the secular leg). The most important forward story. Ramping in production with 2 customers now; +5 Tier-1 CSPs (3 Asian, 2 US) expected to ramp later in 2026; target 5–10% of (expanded) 2026 revenue, growing into 2027+. TLC-based compute/KV-cache SSDs (near-CPU/near-GPU, NVIDIA “CMX”) are ramping first; high-capacity QLC (the larger long-term TAM) ramps as 2-terabit QLC die become broadly available. PCIe Gen6 4nm controller tapes out Q3-2026 with design wins already secured at multiple Tier-1s (incl. 2 NAND makers + several CSPs) for a 2027-28 ramp.
3. Enterprise boot drives. SM8000-series controllers designed into a “leading AI GPU manufacturer’s” DPU (NVIDIA BlueField) since Q4-2025, now qualifying into next-gen DPU + Ethernet + NVLink switches for a 2H-2026 platform launch, with content density up 2–4x per generation. Plus a telecom-infrastructure design win and sampling with a “leading search-engine company” for its TPU architecture. Security-differentiated firmware is the claimed edge.
4. Automotive Ferri. 10+ years of automotive-grade investment now monetizing as makers vacate low-density storage; demand accelerating across US/Europe/China/Japan.

Management’s diversification target: ~20% of revenue from MonTitan + boot drives + automotive combined. If hit durably, it materially changes the through-cycle earnings profile.

Verdict: high-quality forward growth potential (genuine new TAM, real design wins) layered on low-quality trailing growth (cyclical recovery off a trough). The secular vectors are credible and, if they scale, are the single best reason to own SIMO. They are also early, unproven at scale, and not yet large enough to offset a NAND down-leg.

6. Financial Quality

6.1 The income statement — cyclical with a margin-quality wrinkle

Gross margin is the bright spot: stable at 45–49% across boom and bust (FY25 48.3%, Q1-26 47.2%, guided to 48.5–49.5% in Q2-26, with management “confident” of 50% this year and a typical 45–55% range). That stability is the clearest financial proof of the controller IP’s value — a true commodity would not hold ~48% gross margin through a 32% revenue collapse.

Below the gross line, operating margin is violently cyclical: 26.7% (2021) → 22.7% (2022) → 6.4% (2023) → 11.5% (2024) → 10.5% (2025) → 18.2% (Q1-26) → guided 21–22% (Q2-26). The swing is operating leverage: R&D is the dominant cost (FY25 R&D $263M, ~30% of sales) and is largely fixed/committed, so revenue declines flow straight to the operating line. R&D intensity rises in absolute terms as SIMO funds MonTitan/Gen6 — a deliberate growth investment, but one that will keep near-term operating margin below the gross-margin ceiling.

The quality-of-earnings wrinkle: GAAP net income exceeds operating income in every recent period because of a large, volatile non-operating line — interest income, government subsidies, FX gains/losses, unrealized investment holding gains, and disposal gains. FY25: operating income $93.0M, but net income $122.6M; the gap is ~$47M of below-the-line income (and pretax income of $140M actually exceeds operating income). Q1-26: operating income $52.2M, net income $66.8M — again ~$23M of non-operating help. Roughly $90M/year — a quarter-plus of pretax income — sits below the operating line, and a meaningful chunk of it (FX, unrealized investment gains, disposal gains) is non-recurring and non-controllable. The cleaner read of run-rate earnings power is the operating number; GAAP EPS overstates the durable franchise. Note too that Q1-26 GAAP EPS ($1.98) exceeded non-GAAP EPS ($1.58) — the reverse of the usual pattern — precisely because non-operating gains flatter GAAP while non-GAAP strips them out alongside SBC. Use operating earnings, or non-GAAP, for valuation; do not anchor on the GAAP $5.06 TTM.

6.2 The cash-flow and inventory red flag

This is the most important QoE item. FY25 earnings did not convert to cash. Operating cash flow was only $61.4M against net income of $122.6M (cash conversion 0.50); free cash flow collapsed to ~$6.3M (from $98.8M in 2023 and $32.7M in 2024). The cause: inventory ballooned +$215M to $421.8M (from $199M at end-2024) — a deliberate build ahead of the expected ramp and to pre-secure scarce NAND. Cash fell from $277M (end-2025) to $211M (Q1-26) on the build plus the dividend.

This cuts both ways, and the read on it is the bull/bear crux:

• Bull read: building inventory and locking NAND allocation ahead of a known shortage is exactly the right move; it is working capital deployed into a visible, sold-forward ramp, and it will unwind into cash as revenue catches up.
• Bear read: a $215M inventory build in a violently cyclical commodity-adjacent business is precisely how cyclicals get caught at the top. If the NAND cycle turns before the inventory ships, SIMO is sitting on a swollen, potentially write-down-prone book (cash conversion cycle already stretched to 325 days). The 2023 down-cycle saw a $72M inventory release generate cash — the mirror image of today’s build.

The honest position: the build is defensible given current visibility, but it converts SIMO’s “net cash, no risk” balance-sheet story into a “net cash, but with a large, cycle-sensitive working-capital bet embedded” story. Watch inventory and OCF every quarter.

6.3 Returns and the balance sheet

ROE 27.1% and ROIC ~27% in FY25 (up from 12.6%/12.5% in the 2023 trough; peak 57%/57% in 2021). These are genuinely high at the top of the cycle and genuinely mediocre at the bottom — the through-cycle average ROIC is perhaps high-teens to low-20s, respectable but not a fortress-moat signature. The balance sheet is a real strength: ~$211M cash, only ~$2.6M of capital leases, net cash ~$200M+, no financial debt, current ratio 2.8x. SBC is modest for a semi (~$26M FY25, ~3% of revenue; only $8.4M in Q1-26), so dilution is mild and the share count is roughly flat-to-down (134.2M ordinary, ~33.6M ADS).

Verdict: economics that are excellent at the top of the cycle and ordinary at the bottom; a fortress balance sheet partly committed to a cyclical inventory bet; and GAAP earnings that flatter the durable franchise via below-the-line income. Quality is real at the gross-margin line and in the balance sheet; it is cyclical everywhere else, and the cash-conversion break in 2025 is a genuine yellow flag.

7. Capital Allocation

Management’s capital-allocation record is competent and conservative, but not value-creating in the per-share-compounding sense, and it carries one large historical asterisk.

The MediaTek episode (2022-23). In May 2022 MediaTek agreed to acquire SIMO for ~$93.50 cash + 0.388 MediaTek shares per ADS (~$114/ADS at signing) — a clean, premium exit for shareholders. The deal was terminated in July 2023 after failing to obtain China SAMR antitrust clearance amid deteriorating US–China relations; MediaTek paid a $160M break fee. The board’s decision to sell was arguably the most shareholder-friendly capital event in SIMO’s history, and its collapse was outside management’s control. But the episode left SIMO a standalone, busted-arb stock through the NAND trough, and it is a permanent reminder of the geopolitical overhang on any future strategic transaction. (To be explicit: nothing here implies any position; this is historical fact.)

Capital return is dividend-led and pro-cyclical. The dividend per ADS went $1.55 (2021) → $1.50 (2022) → cut to $0.50 (2023 trough) → restored to $2.00 (2024-25); payout ratio ~55% (FY25) and as high as 75% (FY24). Cutting the dividend into the trough and raising it into the recovery is the textbook pro-cyclical pattern — understandable for a cyclical, but the opposite of counter-cyclical capital discipline. Buybacks are modest and also pro-cyclical: ~$46M (2021), $133M (2022), ~$0 (2023-24), ~$24M (2025) — i.e., the most buying happened near the prior peak, the least at the trough. There is no evidence of an opportunistic, valuation-aware repurchase program.

R&D is the primary capital use, and it is rising deliberately to fund MonTitan, the Gen6 4nm controller, and boot drives — the right place to spend, and the spending is producing real design wins. Capex is modest (~$55M FY25, fabless). There has been no major M&A in the period; the growth is organic.

Insider and incentive alignment. As an FPI, SIMO’s officers are exempt from Section 16 reporting, so there is no clean Form-4 insider-trading corpus; the 26 Rule-144 planned-sale notices in the five-year filing set indicate routine insider selling into strength (planned, not necessarily conviction-revealing). The widely-circulated “$89M sale” headline (May-2026) was an outside fund — Pertento Partners — trimming a 13F position, not an officer sale; it is a portfolio-rotation datapoint, not an insider signal. Founder-CEO Wallace Kou has led the company since inception, which is a genuine continuity asset; the proxy-equivalent compensation detail (and whether incentives are tied to ROIC/per-share metrics vs revenue/size) is an open item — no ROIC-linked metric is evident in the disclosure reviewed, which is a mild negative consistent with the pro-cyclical pattern above.

Verdict: conservative stewardship of a fortress balance sheet, the right R&D priorities, and a once-in-a-decade attempt to sell at a premium that died on geopolitics — but pro-cyclical dividends/buybacks and no evidence of valuation-aware, per-share-focused capital allocation. Adequate, not exemplary.

8. Changes and Headwinds — Last Two Years

• From trough to super-cycle (2024 → 2026). The dominant change: the NAND market flipped from multi-year glut to acute, AI-driven shortage. NAND prices +55–60% QoQ in Q1-2026; flash makers reallocating capacity to HBM/DRAM/enterprise and exiting low-end consumer/merchant controllers — the supply-side dynamic powering SIMO’s share gains. Strengthens the thesis near-term; is itself the cyclical risk longer-term.
• The enterprise/AI pivot became real. MonTitan moved from roadmap to production (2 customers ramping, qualifications a quarter ahead of plan); boot drives entered NVIDIA’s DPU/switch platforms; the Gen6 4nm controller secured Tier-1 design wins pre-tape-out. This is the most thesis-relevant structural change of the period. Strengthens the thesis.
• eMMC/UFS share gains structurally re-rated the mix from 25–30% (2023) to 40–45% (2025) of sales, diversifying away from the volatile client-SSD line. Strengthens.
• The 2025 inventory build / FCF collapse (+$215M inventory, FCF ~$6M) introduced working-capital and cycle-timing risk to a previously pristine balance-sheet story. Weakens / adds risk.
• The April-2025 tariff/recession panic drove the $37.89 capitulation low — a reminder of how violently the stock discounts macro and cycle fear, and of the tariff/geopolitical overhang on a Taiwan-China supply chain. Neutral-to-weakening (volatility).
• Smartphone/PC unit weakness worsened through 2026 (units now expected −10%+ vs prior −5/10%), concentrated at the low end — partly offset for SIMO by its limited low-end exposure and by share gains, but a demand headwind nonetheless. Mild headwind.
• Substrate supply tightness (BGA substrate “Ton material”) flagged by management as the binding near-term operational constraint, competing with US Tier-1 customers for limited supply. Mild headwind.

Verdict: the last two years’ changes are net thesis-strengthening on the structural axis (enterprise pivot, share gains, mix shift) and thesis-complicating on the risk axis (inventory bet, cycle-top positioning, geopolitics). The stock has more than priced the positives.

9. Risk Analysis (Risk Matrix)

The dominant risks are not solvency (the balance sheet is pristine) but valuation and cycle-timing: a high-multiple stock priced for a permanent super-cycle, sitting on a cycle-sensitive inventory bet, with a customer base that could compete. The catastrophic-loss probability is genuinely low; the “permanent impairment of capital paid at $300+” probability is not.

10. Valuation Discussion (Embedded Expectations)

No price target and no recommendation. This section frames what the current price implies.

10.1 Where the multiple sits

At $305.28 (26-Jun-2026): ~33.6M ADS → market cap ~$10.24B; net cash ~$210M → EV ~$10.03B. Against the trailing financials:

Every cross-cycle anchor — sales, book, dividend yield — says the stock is at the most expensive point in its public life, by a wide margin. The only multiple that looks “reasonable” is forward P/E if you accept a steep continued ramp.

10.2 Embedded-expectations / scenario analysis

The market is paying ~$10B EV for a business doing ~$1.06B TTM revenue at a ~15–18% operating margin. To justify $305 on a ~25x forward-earnings basis, SIMO needs roughly $400M+ of net income, i.e. ~$12/ADS — versus $169.9M (~$5.06) TTM and $122.6M (~$3.65) FY25. That requires revenue to roughly double from the FY25 base to ~$1.7–1.9B at a ~20%+ operating margin and a normalized tax rate — achievable at a sustained super-cycle peak with the enterprise pivot scaling, but it is a peak-on-peak assumption.

• Bear (cycle turns 2027-28): revenue normalizes back toward ~$1.0–1.2B, operating margin to ~12–14%, EPS to ~$4–5; the multiple compresses toward the historical 2–3x sales / mid-teens P/E. Implied value ~$90–150 (a 50%+ drawdown from $305 — entirely consistent with SIMO’s −93% lifetime max drawdown and its 2022→2023 halving).
• Base (elongated cycle, enterprise scales gradually, then a normal down-leg): mid-cycle normalized revenue ~$1.3–1.5B, ~17–18% operating margin, normalized EPS ~$6–9/ADS; a cyclical-but-growing fabless semi fairly trades ~18–26x mid-cycle → ~$150–230. The current price sits at/above the top of this band.
• Bull (AI demand = structural step-change, super-cycle persists into 2028, MonTitan/boot drives become 20%+ of a much larger revenue base): revenue $1.8–2.2B, ~22% operating margin, EPS $11–14, and the market keeps paying a growth multiple → $300–420+ (consistent with the Wedbush $400 target). This is the scenario the current price underwrites.

The price is therefore fully discounting the bull and giving little weight to the base, and essentially none to the bear. The asymmetry is unfavorable for a new buyer at $305: you are paid the bull’s upside (~+30%) for taking the bear’s downside (~−55%), in a business whose own history says the bear arrives on a schedule.

Comp context. Phison (8299.TW) is the direct merchant peer; Marvell (MRVL) is the enterprise-SSD-controller comparator (richly valued on AI); the NAND makers themselves (Micron, SK Hynix, Kioxia, Samsung) are simultaneously customers, suppliers, and the captive ceiling. Versus the broader fabless-semi peer set (SWKS, QRVO, MPWR, AEIS), SIMO screens as a higher-growth, higher-cyclicality, richer-multiple name today — closest in setup to the “richest-ever cyclical re-rated to an AI story” cohort (AEIS, KNX), but with a genuinely better secular growth option than most of them.

Verdict: the price embeds a permanent super-cycle and a flawless enterprise pivot. The structural improvements are real; they are also fully — arguably over — capitalized.

11. Variant Perception

Consensus belief (where the sell-side and the tape are). SIMO is a structural AI-storage winner whose share gains and enterprise/cloud-AI pivot (MonTitan, boot drives) have changed the business’s earnings power permanently; the memory super-cycle runs into 2027; price targets to $400. The tape agrees emphatically — +363% trailing year, near an all-time high, a crowded momentum long.

Strongest bull case. The make-vs-buy shift is a genuine, supply-side-driven secular re-rating of SIMO’s addressable market: as cash-constrained NAND makers permanently retreat from low-margin consumer controllers to chase AI/HBM, SIMO inherits share that does not come back. Layer on a real new TAM — enterprise eSSD and AI boot drives, with NVIDIA-platform design wins and Tier-1 CSP customers — and SIMO is no longer a pure NAND-cycle derivative but a structural AI-infrastructure supplier. At ~25x a plausible $12 forward EPS, $305 is not expensive if the earnings are durable.

Strongest bear case. SIMO is a fabless price-taker at the richest multiple in its history, near the top of the most violent commodity cycle in technology, sitting on a $422M inventory bet, with GAAP earnings flattered ~$90M/year by FX/subsidies/investment gains, a customer base (Micron/Kioxia/SK Hynix) that is its own most dangerous competitor, and a pro-cyclical capital-return record. The last cycle cut EPS 69%; the next one will too. The enterprise pivot is real but early, and the price already pays for it as if it were proven and scaled.

The 3–5 assumptions that matter most:

1. Is the super-cycle structural or cyclical? (Bull needs AI demand to abolish the glut; history says it won’t.)
2. Does the merchant share gain stick when the cycle turns — or do NAND makers re-insource once consumer controllers matter again?
3. Does MonTitan + boot drives + auto durably reach and exceed 20% of revenue, giving SIMO non-cyclical earnings — and does it do so with the cycle flat?
4. Will the $215M inventory build convert to cash and revenue, or to write-downs if the cycle mistimes?
5. What is normalized mid-cycle EPS — the whole valuation pivots on whether it’s ~$5 (bear) or ~$12 (bull).

Falsification. Bull is falsified if NAND ASPs roll over in 2027-28 and SIMO’s revenue/margins compress while inventory bloats — i.e., a normal cycle turn — or if a major NAND maker re-insources client/UFS controllers. Bear is falsified if enterprise (MonTitan + boot-drive) revenue scales past 20% of a growing base and holds through a NAND down-leg, demonstrating that SIMO’s earnings power has structurally de-cyclicalized.

Factor-positioning read (FactorsToday). The model locates SIMO as a high-beta (~1.36–1.58), high-idiosyncratic-vol, momentum-loaded semiconductor name (negative LowVolatility loading −0.58; positive Momentum +0.13; dominant Semiconductor/Tech sector betas; R² only ~0.22–0.26, so most of the move is stock-specific). Risk-adjusted track record is spectacular recently and dreflects the melt-up: trailing-year return +363%, but lifetime max drawdown −93% and lifetime Sharpe just 0.33 — the signature of a violently mean-reverting cyclical that is currently mid-mania. The positioning evidence supports the bear’s “crowded momentum long, late in the move” reading more than the bull’s “durable compounder” framing — without making a price call.

12. Fact vs. Interpretation Table

13. Open Questions

1. Normalized mid-cycle EPS — the entire valuation hinges on whether it is ~$5 or ~$12/ADS. What does enterprise contribution do to the trough EPS of the next cycle?
2. Composition and durability of the ~$90M/yr non-operating income — how much is recurring (interest, structural subsidies) vs. non-recurring (FX, unrealized investment gains)? Needs the 20-F note detail.
3. MonTitan/boot-drive revenue trajectory — actual dollars, not just “5–10% of revenue”; how much is shipping vs. design-win pipeline, and what are the gross margins vs. the corporate average?
4. Will the $215M inventory convert to cash over the next 2–3 quarters (watch OCF and the inventory line), or sit if demand softens?
5. Executive compensation structure — is any portion tied to ROIC or per-share metrics, or purely to revenue/size? (FPI disclosure is thinner than a DEF 14A.)
6. Customer identities behind PHISEMI / AFASTOR — are these arm’s-length module makers or related/affiliated entities? Any related-party element to the concentration?
7. Phison competitive dynamics — is SIMO’s >50% PCIe-5 client share gaining or being contested, and at what price?
8. Re-insourcing risk timing — under what conditions would Micron/SK Hynix/Samsung pull controller development back in-house?

14. What Must Be True

For the bull case (owning at ~$305 works out):

1. The AI-driven memory demand is a structural step-change that elongates the up-cycle into 2028+, not a normal glut-bound cycle. Falsification: NAND ASPs and bit-demand roll over in 2027-28 and SIMO’s revenue/margins compress — a normal cycle turn.
2. Merchant share gains stick through the next down-leg; NAND makers do not re-insource client/UFS controllers. Falsification: a major maker announces/ramps in-house controllers for consumer SSD or UFS.
3. MonTitan + boot drives + automotive durably exceed 20% of revenue and hold through a NAND down-leg, structurally de-cyclicalizing earnings. Falsification: enterprise revenue stalls below ~10% of sales or proves as cyclical as the legacy lines.
4. The $215M inventory build converts to cash and revenue, not write-downs. Falsification: inventory keeps rising while revenue growth decelerates; an inventory write-down or gross-margin hit appears.
5. Normalized mid-cycle EPS is ~$10–14/ADS, making ~25x reasonable. Falsification: a full cycle reveals normalized EPS nearer ~$5–7.

For the bear case (a de-rating / drawdown from here):

1. The current super-cycle is a cycle, and NAND normalizes in 2027-28. Falsification: bit-supply stays structurally short and ASPs hold for 3+ years.
2. The richest-ever multiple mean-reverts toward SIMO’s historical 2–3x sales / mid-teens P/E. Falsification: the market permanently re-rates SIMO as a secular AI compounder and holds 8–10x sales.
3. GAAP earnings prove lower-quality than they look (non-operating income fades; inventory pressures margins). Falsification: operating margin sustains 20%+ and cash conversion normalizes to ~1.0.

The single most decisive variable for both sides: whether SIMO’s enterprise/AI earnings can grow through the next NAND down-cycle. If they can, the bull is right and SIMO has graduated from cycle-derivative to AI compounder. If they can’t, the stock is a high-multiple cyclical at the top, and the next turn will be expensive.

15. Source Appendix

See the Source Appendix below for the full source list with URLs and access dates.

This article contains no investment recommendation and no price target outside the clearly-labeled opening opinion block. It is general information, not investment advice.

APPENDIX A — Standard Diligence Questionnaire

Silicon Motion Technology Corporation (NASDAQ: SIMO) — as of 2026-06-27

Supplemental to the research memo. Fact / Interpretation / Assumption labels applied where it matters.

General

What thoughtful questions have other investors asked? (From the Q1-2026 call and sell-side.) (1) How do the segments rank for sequential growth, and is the eMMC/UFS business diversifying away from handsets? (Yes — eMMC into automotive/smart-devices growing faster than UFS/handset.) (2) MonTitan: how many customers ramping (2 now, +5 Tier-1 CSPs by late 2026), and why the shift toward TLC/compute-SSD vs QLC? (NAND price spike made high-capacity QLC uneconomic short-term; TLC compute/KV-cache SSDs ramp first — and ship more controllers per drive). (3) Is there pull-forward demand in client SSD ahead of price hikes? (Some, but share gains dominate.) (4) Can SIMO secure enough NAND for 2026/27 growth? (Yes via 20-yr maker relationships, 3-supplier sourcing.) (5) Can it hold/expand to 50% gross margin? (Management: 50% “achievable” in 2026.) (6) Substrate supply constraint — BGA “Ton material” is the binding bottleneck. Interpretation: the buy-side debate is super-cycle durability + enterprise scaling vs. cycle-top valuation.

Cyclicality & Earnings Nature

Cyclical high or low? Fact: Near a cyclical high — Q1-26 revenue a record, +105% YoY off a 2024-25 trough; operating margin (18.2%) climbing toward the 2021-22 peak (23–27%). EPS still below the 2021 peak ($5.74) but the run-rate is closing fast. External environment or internal actions? Both — a genuine AI-driven NAND super-cycle (external) plus real merchant share gains and an enterprise pivot (internal). The external driver dominates the magnitude. Revenue stability? Fact: Low. Revenue swung 922→946→639→804→886 over 2021-25; visibility is ~1-2 quarters; no contracted recurring base. Outlook for products? Strong near-term (sequential growth guided all of 2026); structurally improving via MonTitan/boot-drive/auto; cyclically exposed to the next NAND down-leg. Market size — growing/shrinking, domestic/international? Merchant controller TAM is growing via NAND-maker outsourcing + new enterprise/AI storage TAM; 99% of revenue is international (79% China/Japan/Singapore).

Business Quality & Competitive Moat

Industry more or less competitive? Interpretation: Less competitive in the legacy merchant niche near-term (NAND makers exiting low end; “major flash makers essentially gone” from eMMC), but the captive-controller ceiling and Phison keep it from being a true monopoly; enterprise eSSD is more competitive (Marvell, captive). How profitable (ROIC/ROE)? Fact: FY25 ROE 27.1%, ROIC ~27%; trough (2023) 12.6%/12.5%; peak (2021) 57%/57%. High at the top, ordinary at the bottom; through-cycle high-teens/low-20s. Industry profitability / barriers? NAND oligopoly (6 makers); controller layer is where differentiation/margin live (gross margin ~48% through-cycle). Barriers = firmware IP, multi-generation qualification, NAND-maker relationships. Moderate, not fortress. Easily understood? Reasonably — a fabless controller designer riding the NAND cycle with an enterprise option. Undermined by low-cost labor? No — it’s an IP/design business, not labor-cost-driven; foundry/OSAT costs are the input. Do brands matter? Not consumer brands; reputation/qualification track record with NAND makers and OEMs is the relevant “brand.” Nature of competition? Design wins per product generation; Phison the direct merchant rival; captive controllers the ceiling. Switching costs? Interpretation: Moderate — re-qualification cost per generation; but customers can in-source generation by generation.

Financial Condition & Balance Sheet

Assets not on the balance sheet? Decades of firmware IP and NAND-maker relationships (intangible, unrecognized). Assumption: the real value is the IP + relationships, not the carrying assets. Off-balance-sheet liabilities? None material identified; obligations are routine (purchase commitments, OSAT). Note the embedded working-capital risk in the $422M inventory. Accounting conservatism? Mixed. Gross margin/revenue recognition appear clean; caution flag: GAAP net income is flattered by a large, volatile non-operating line (FX, government subsidies, unrealized investment gains) — ~$90M/yr, with GAAP EPS exceeding non-GAAP in Q1-26. Use operating/non-GAAP earnings. CapEx-hungry? No — fabless, capex ~$55M FY25 (~6% of sales). The capital intensity is in working capital (inventory), not fixed assets.

Capital Allocation & Management

FCF generation and use? Fact: Through-cycle FCF positive but volatile; FY25 FCF collapsed to ~$6M on the +$215M inventory build (vs $98.8M in 2023). Uses: dividend (primary), modest buybacks, rising R&D. Recent acquisitions? None material. Growth is organic. (The notable corporate event was being the target — MediaTek’s terminated 2022-23 acquisition; $160M break fee received.) Buying back shares? Modestly and pro-cyclically (~$24M FY25; ~$0 in the 2023-24 trough; $133M near the 2022 peak). No valuation-aware program. Issuing shares to insiders? SBC modest (~$26M FY25, ~3% of sales); share count roughly flat-to-down. FPI → no Section 16 Form-4 corpus; 26 Rule-144 planned-sale notices (routine insider selling into strength). The “$89M sale” was an outside fund (Pertento), not an officer. Compensation policy / motivations? Founder-CEO Wallace Kou leads since inception (continuity asset). Open question: no ROIC/per-share-linked metric evident in the FPI disclosure reviewed; pro-cyclical dividend/buyback pattern suggests size/cash-flow orientation over per-share-value discipline.

Valuation & Market Data

ADR/MLP/K-1? Fact: ADR (1 ADS = 4 ordinary shares); Cayman-incorporated, Taiwan-HQ FPI filing Form 20-F. Not an MLP/K-1. Dividend policy? Fact: $2.00/ADS in 2024-25 (cut to $0.50 in the 2023 trough, restored on recovery); ~0.66% yield at $305; payout ~55%. How profitable? See ROIC/ROE above — high at the top of the cycle. Net income vs cash from operations diverging? Fact — yes, materially. FY25 OCF $61M vs NI $123M (conversion 0.50) on the inventory build — the single most important QoE flag. Watch it reverse (or not) over the next 2-3 quarters.

Risks & Downside

What would cause the stock to decline? A NAND cycle turn (ASP/volume + inventory write-down), a valuation de-rating from richest-ever multiples, customer re-insourcing, MonTitan under-delivery, a top-customer loss, or a China/Taiwan geopolitical shock. Catastrophic loss risk? Low on fundamentals (net cash, no debt, through-cycle profitable). The realistic severe case is a 50%+ drawdown from a cycle turn + multiple compression — consistent with the −93% lifetime max drawdown and the 2022→2023 halving — not a wipeout. Total loss? Very low absent an extreme geopolitical tail (Taiwan-Strait).

Recent News & Events

Business environment changed recently? Fact — dramatically. NAND flipped from multi-year glut to acute AI-driven shortage (prices +55–60% QoQ Q1-26); flash makers reallocating to HBM/DRAM/enterprise and exiting low-end → SIMO share gains. Stock +363% trailing year to a $336.90 ATH (Jun-2026), now $305. Significant acquisitions? None by SIMO. (Historical: MediaTek’s terminated bid.) Accounting policy changes? None material identified. Recent changes — markets/facilities/management? New enterprise/cloud-AI market entry (MonTitan, boot drives into NVIDIA platforms); Gen6 4nm controller tape-out Q3-2026; automotive Ferri scaling; management team stable (Kou CEO, Jason Tsai CFO).

APPENDIX B — Source Appendix

Silicon Motion Technology Corporation (NASDAQ: SIMO) — accessed 2026-06-27

Primary sources first. Quantitative figures reconciled to SEC filings where applicable; ROIC.ai used as a cross-check aggregator (third-party, not primary).

Primary — SEC filings (EDGAR, CIK 0001329394)

• Form 20-F, FY2025 (annual report), filed 2026-04-30 — https://www.sec.gov/Archives/edgar/data/1329394/000119312526197184/d17718d20f.htm — business overview, segment mix (SSD controllers 45-50% / eMMC-UFS 40-45% / SSD solutions 0-5% in 2025), customer concentration (top-5 = 66%; >10% customers PHISEMI/Kioxia/AFASTOR/Micron), geographic mix (99% ex-US; 79% China/Japan/Singapore), risk factors, non-operating income description, FX.
• Form 6-K, Q1-2026 results, filed 2026-04-29 — https://www.sec.gov/Archives/edgar/data/1329394/000119312526188049/d285401d6k.htm — record revenue $342.1M (+105% YoY), gross margin 47.2%, operating margin 18.2%, GAAP EPS/ADS $1.98 / non-GAAP $1.58, Q2-26 guide ($393–411M, GM 48.5–49.5%, op margin 21–22%), cash $210.9M.
• Form 20-F, FY2024 / FY2023 / FY2022 — filed 2025-04-30 / 2024-04-30 / 2023-04-28 / 2022-04-25 (multi-year segment & customer history; cyclical comparison).
• Form 6-K series, FY2021–FY2026 — quarterly earnings releases (revenue/EPS/dividend history).
• Form 425 series (21 filings, 2022-2023) — MediaTek/SIMO merger communications: deal announced 2022-05-05 (~$93.50 cash + 0.388 MediaTek shares per ADS ≈ ~$114/ADS); terminated 2023-07-26 on China SAMR non-clearance; $160M break fee.
• Rule 144 notices (26 filings, 2021-2026) and Form 3/4 — insider planned-sale activity (FPI; officers largely Section 16-exempt).
• Schedule 13G/A (institutional holders), Form S-8 (2025-06-04, equity plan), Form SD (conflict minerals).

Primary — Management commentary (transcript; treated as hypothesis, validated against filings)

• SIMO Q1-2026 earnings call transcript, 2026-04-29 (via ROIC.ai get_latest_earnings_call) — Wallace Kou (CEO) / Jason Tsai (CFO). Source for: NAND +55–60% QoQ; mobile eMMC/UFS +30-35% QoQ / +140% YoY; client SSD +45% YoY; MonTitan (2 customers ramping, +5 Tier-1 CSPs by late 2026, 5–10% of revenue target, Gen6 4nm Q3-26 tape-out); boot drives into NVIDIA DPU/NVLink/Ethernet; Ferri automotive; 50% gross-margin target; ~20% diversification target; substrate (“Ton material”) constraint; 3-supplier NAND sourcing; 2027 expected tighter than 2026.

Quantitative aggregators (third-party; cross-check, reconciled to filings)

• ROIC.ai MCP — income statement, balance sheet, cash flow, profitability/per-share/valuation-multiples (FY2019–Q1-2026). Key derived figures: TTM revenue ~$1,061M, TTM NI ~$169.9M, EV ~$10.0B, decade multiple ranges (P/S 1.4–5x, P/TBV 1.5–5x). TBV ~$24.74/ADS.
• AZI price history CSV — https://azitrading.com/controls/download-data.php?t=SIMO — 5-year OHLCV, EMAs, beta. Price waypoints: $86.41 (end-2021), ~$85 (May-2022 MediaTek bid), ~$55 (Jul-2023 deal termination), $37.89 (21-Apr-2025 low), $336.90 (22-Jun-2026 ATH), $305.28 (26-Jun-2026).
• AZI valuation_index — FLAGGED GARBLED for SIMO (ADS/ordinary-share ratio confusion; reported TTM EPS $18.71 / book $106.87 / composite 57.6th pctile are ~5x off real per-ADS figures). Not relied upon (NEM-style garble).
• FactorsToday — https://www.factorstoday.com/api/{stock-loadings,leaderboard,stock-info}/SIMO — beta ~1.36–1.58; LowVolatility loading −0.58; Momentum +0.13; Semiconductor/Tech sector loadings dominant; R² ~0.22–0.26; trailing-year return +363%; lifetime max drawdown −93%; lifetime Sharpe 0.33.

Secondary — News / market data

• “This Chip Stock Soared 325%. Why One Investor Still Sold $89 Million Worth,” 2026-05-23 (Yahoo/Nasdaq, EODHD) — Pertento Partners LLP 13F trim of 738,875 SIMO shares (~$89.7M); SIMO at $276.14, ~$9.4B market cap, TTM revenue $1.06B, TTM NI $169.97M. (Confirms the “$89M sale” was an outside fund, not an officer.)
• “Wedbush Maintains Outperform on Silicon Motion Tech, Raises Price Target to $400,” 2026-06-22.
• Micron strong-results-driven semiconductor rally items (2026-06-24); semiconductor sell-off on OpenAI/sector-rotation headlines (2026-06-05, 2026-06-26) — context for the tape.

All figures as of access date 2026-06-27 unless otherwise stated. Company is a foreign private issuer reporting in USD; no SEC Form 4 insider corpus (foreign-private-issuer exemption).