IMOS · ChipMOS TECHNOLOGIES INC. · Stock research

Current Thesis

ChipMOS is a Taiwanese OSAT (outsourced assembly and test) that runs commodity DRAM, NAND/flash, bumping, and display-driver ICs through the back end a thin-margin supply-chain layer levered to memory unit growth, not an HBM advanced-packaging supplier. The narrative being bought is that a persistent AI memory demand/supply imbalance keeps back-end utilization tight and pricing firm. Fifteen days ago that leg looked like it was maturing: May revenue had decelerated to +17.7% YoY and slipped -3.1% MoM. The 2026-07-10 release rewrote that read. June revenue of NT$2,538.4M (US$79.7M) rose +37.2% YoY and +6.5% MoM a record monthly figure since 2014 and Q2 revenue of NT$7,383.1M (US$231.8M) came in +28.7% YoY, ahead of Q1's +25.4%. The second derivative flipped positive at the exact moment the bear case wanted it to keep rolling over, and price has bought back the ~17% June fade to sit near ~$68, roughly 7% under the $73.97 high.

Bullish and bearish views on ChipMOS TECHNOLOGIES INC.

The model's bull view on ChipMOS TECHNOLOGIES INC. (IMOS), in brief: June 2026 revenue +37.2% YoY (US$79.7M, reported 2026-07-10) a record monthly print since 2014 and a sharp re-acceleration from May's +17.7%; the monthly slope a momentum book wants turned back up. The bear view: Margins remain the unanswered question. Q1 gross margin was ~13.8%; a commodity OSAT does not earn HBM economics, so a revenue record can coincide with flat-to-down profit. A 2026-07-06 note framed the setup explicitly as "AI revenue growth vs. profitability crisis." Valuation… Both cases follow in full.

Bull Case

• June 2026 revenue +37.2% YoY (US$79.7M, reported 2026-07-10) a record monthly print since 2014 and a sharp re-acceleration from May's +17.7%; the monthly slope a momentum book wants turned back up.
• Q2 2026 revenue US$231.8M, +28.7% YoY and +6.5% QoQ (2026-07-10) quarterly YoY growth accelerated versus Q1's +25.4%, off the highest quarterly base since 2014.
• Q1 2026 revenue US$216.4M, +25.4% YoY with net profit NT$504.9M (EPS NT$0.72, reported 2026-04-10) the earnings inflection off the depressed 2025 memory trough is real, and management cites long-term supply agreements plus new capacity absorbed by existing customer forecasts.
• Broad back-end leverage: testing, assembly, bumping, and display-driver lines all track memory unit growth without single-customer concentration; if the DRAM/NAND upcycle holds into 2H26, utilization and pricing compound together.
• Price confirmation: the ADS reclaimed the entire June drop to ~$68.07 (2026-07-08, prior close $65.85) while the memory complex (Micron, SK Hynix) and OSAT peers (ASE, Amkor, Powertech) traded broadly bid cluster strength underneath the move.

Bear Case

• Margins remain the unanswered question. Q1 gross margin was ~13.8%; a commodity OSAT does not earn HBM economics, so a revenue record can coincide with flat-to-down profit. A 2026-07-06 note framed the setup explicitly as "AI revenue growth vs. profitability crisis."
• Valuation carries a full cycle. ~83x trailing earnings against a historical 8–12x band means a complete memory re-rating is already in the price; multiple compression alone can remove 20–30% even if revenue holds.
• The stock has 4x'd from a $15.06 52-week low to a $73.97 high; entering near the top of that range is buying after the move, with price stretched well above a rising 20-EMA sitting in the low-$60s.
• AI content is indirect. ChipMOS is a second-order beneficiary of the memory cycle via commodity DRAM/NAND and display-driver ICs; if positioning re-rates from "AI packager" to "commodity DDIC back-end," the premium loses its anchor.
• The revenue line is known through June; the profit line is not until 2026-08-11. The single fact the bear thesis turns on whether the ~13.8% margin expands with record utilization stays hidden until that print.

Setup & Price Structure

Recovered and re-testing the highs. After tagging $73.97 around 2026-06-22 and fading ~17% to ~$60 by late June, the ADS reclaimed most of the drop and traded ~$68.07 on 2026-07-08, leaving a rough $60–$74 consolidation shelf built over three weeks. That reframes the mid-June high from a failed parabolic top into a base-and-resume attempt, with the June sales print as the fundamental catalyst that pulled buyers back. Price sits above a rising 20-EMA estimable in the low-$60s and about 7% below the record high extended after a 4x, but the tape reads constructive, with buyers absorbing the June fade. The 2026-06-29 ex-dividend (US$0.783/ADS) mechanically clipped ~$0.78 and should not be read as a break. The structural line that matters is the $58–$60 shelf: hold it and the resumption stays intact; lose it on a weekly basis and the June breakout base fails.

Catalyst Calendar (next 30 days)

• ~2026-08-08 to 2026-08-10 (est.): July 2026 monthly revenue release. The freshest momentum tell another 30%+ YoY print confirms the re-acceleration; a slip back toward the mid-teens revives the digestion read.
• 2026-08-11: Q2 2026 full results (P&L, gross/operating margin, guidance). The binary. Revenue is already known and strong; this print reveals whether the ~13.8% gross margin expands with record utilization or stays capped the crux of the entire bull/bear split.
• 2026-07-24: dividend payment date (US$0.783/ADS). Not a price catalyst; noted for completeness.

What Would Change Our Mind

The bull read breaks on a weekly close below $58, which loses the $58–$60 consolidation shelf the recovery to ~$68 launched from and reopens the failed-parabola interpretation. Fundamentally, it inverts if the 2026-08-11 Q2 print shows gross margin compressing while revenue grows the "revenue without profit" outcome the 2026-07-06 note flagged or if July monthly revenue decelerates back below +20% YoY, marking the June record as a pull-forward rather than a new trend. A theme flip in the memory complex (Micron/SK Hynix rolling over, DRAM contract pricing turning) would remove the second-order pull-through that underwrites the entire premium.

Correlation Notes

ChipMOS is a second-order node, so its tape lags the primaries. Watch memory makers Micron (MU) and SK Hynix for the demand and pricing signal, and OSAT peers ASE (ASX), Amkor (AMKR), and Powertech (PTI) for back-end utilization peer rollover typically leads IMOS lower, while cluster strength confirms its up-legs. The name carries small-cap ADR mechanics: the 1:20 ADR ratio (each ADS = 20 Taiwan common shares) means US-listed EPS feeds routinely garble per-share data, so reconcile against the NT$ figures (Q1 EPS NT$0.72). Broad semiconductor beta (SOX/SMH) and the USD/TWD cross both move the ADS independent of company news.