atlasHold
Open analysis → · PDF unavailable
Preview
# HIMX — Stock Analysis (Atlas) > Date: 2026-04-30 > Quarter: Q4 FY25 (latest reported, Feb 12, 2026) > Market cap: ~$1.91B | EV: ~$1.65B | EV/TTM Rev: 2.0x | Revenue YoY: −14.4% | TTM Revenue: $832M ## Verdict **FAILS the standard growth qualification gate, PASSES as a special-situation optionality play.** Revenue is contracting (−14.4% YoY in Q4 FY25, FY25 −8.2% YoY); gross margin is 30%, not >60%. This is not a wsm/Saul/Bear-style hyper-growth name. Two things keep me from dismissing it: (1) a sole-source position in TSMC's COUPE silicon-photonics platform — credible per Ming-Chi Kuo, with potential to take revenue from $832M → $2.4B by 2028 at 45–50% incremental gross margin; (2) a defensible automotive DDIC moat (~40% global share, >50% in TDDI). The stock trades at 2.0x EV/Rev and 13.9x EV/FCF with $258M net cash — pricing it as a cyclical legacy semi, not as an AI-infrastructure call option. **Conviction: 2/5 as a growth holding; 3/5 as a small special-situation tryout.** The gate-failure is real and disqualifies it for any persona running the orthodox six-factor screen, but the optionality is too specific and too cheap to ignore entirely. ## Qualification Gate | Criterion | Threshold | HIMX | Pass? | |-----------|-----------|------|-------| | Revenue YoY growth | >30% (>40% preferred) | **−14.4%** | **FAIL** | | Gross margin | >60% (>70% preferred) | **30.4%** | **FAIL** | | Revenue per quarter | >$50M | $203M | Pass | | Data availability | 4+ quarters in DB | 16 quarters | Pass | | Share dilution | <10% annual | ~0% (174M stable) | Pass | | GAAP profitability trajectory | Improving or positive | **Compressing** (op margin 9.7% → 3.4% YoY) | **FAIL** | **Gate verdict:** Fails 3 of 6 criteria. By orthodox growth screen, HIMX is disqualified. The remaining factors (size, data, dilution discipline) are fine but those are necessary, not sufficient. **An honest growth analyst would stop here and decline to take a position.** I continue the analysis only bec *…truncated*