valuation / STM
2026-06-23 · DCF · medium · source: weilab/model-builder@2026-06-23
| wacc | 8.52% |
| cash_m | 3,000 |
| debt_m | 2,900 |
| currency | USD |
| terminal_g | 2.70% |
| pe_terminal | 16 |
| period_label | Market-implied base · 結構性 800V SiC 缺貨 (DCF) |
| forecast_path | [{"eps":3.22,"year":"CY26E","fcf_usd_m":1822,"ebitda_usd_m":4891,"revenue_usd_m":16035},{"eps":3.99,"year":"CY27E","fcf_usd_m":2508,"ebitda_usd_m":5952,"revenue_usd_m":18601},{"eps":4.43,"year":"CY28E","fcf_usd_m":3190,"ebitda_usd_m":6607,"revenue_usd_m":20647},{"eps":4.38,"year":"CY29E","fcf_usd_m":3426,"ebitda_usd_m":6675,"revenue_usd_m":21886},{"eps":4.37,"year":"CY30E","fcf_usd_m":3538,"ebitda_usd_m":6779,"revenue_usd_m":22980}] |
| ai_800v_sizing.note | onsemi self-reported content doubled $50k->$100k/1MW rack; STM named NVIDIA 800VDC primary. Earlier build used $10k -> corrected. |
| ai_800v_sizing.stm_direct_ai_sic_pct_of_rev | 2-4% |
| ai_800v_sizing.stm_sic_slice_per_rack_usd_k | 50 |
| ai_800v_sizing.ai_800v_power_semi_tam_2030_usd_b | 2.6 |
| ai_800v_sizing.onsemi_power_semi_per_1mw_rack_usd_k | [50,100] |
| pe_applied_fwd | 22 |
| revenue_cagr_5y | 10.30% |
| scenario_layers.base.price | 66.91 |
| scenario_layers.base.thesis | franchise recovery + structural 800V SiC re-rate (durable terminal GM ~43%); reconstructs ~spot |
| scenario_layers.base.confidence | medium |
| scenario_layers.bear.price | 29.64 |
| scenario_layers.bear.thesis | shortage fizzles / 800V SiC capacity arrives fast / auto+industrial stays soft |
| scenario_layers.bear.confidence | low |
| scenario_layers.blue_sky.price | 103.51 |
| scenario_layers.blue_sky.thesis | full structural super-cycle; ASP+GM spike persists (terminal GM ~45.5%), high SiC capture |
| scenario_layers.blue_sky.confidence | rumor |
| shortage_window.end_year | CY28E |
| shortage_window.start_year | CY26E |
| shortage_window.peak_gross_margin | 46.50% |
| shortage_window.normalized_gross_margin | 43.00% |
| valuation_basis | Two-stage DCF (CY26E-CY33E explicit). Full-franchise recovery off the FY24 trough + a STRUCTURAL multi-year 800V SiC shortage (2027-2030 ramp, competing with EV for fab capacity). Shortage window CY26E-CY28E spikes gross margin (base 46.5%, blue 53%), then re-rates DURABLY to a structurally higher terminal (base ~43%, blue ~45.5%) — not back to old ~40% mid-cycle. Base mid reconstructs ~spot (-14.6%; gap closed from -27% once 800V sized at onsemi's $50-100k/1MW-rack content, $2.6B 2030 TAM). Direct AI-SiC revenue is only ~2-4% of total — the thesis works through the shortage lifting ASP/GM across the WHOLE SiC/power book (the MLCC-2018 mechanism), not the direct slice. |
| growth_years_stage2 | 3 |
| revenue_cagr_stage2 | 4.00% |
| shares_outstanding_m | 900 |
| ebitda_margin_terminal | 28.50% |
| market_implied_capture_pct | 76.2 |
| market_implied_eps_at_spot | 3.56 |
**結構性 800V SiC 缺貨 thesis** — STM 為 top-3 SiC 供應商、NVIDIA 800VDC 點名 primary。800VDC AI 機櫃（Kyber／Rubin Ultra，2027 起 600kW–1MW）需 1200V SiC MOSFET。onsemi 自報單機櫃 power-semi 內容已從 \$50k 倍增至 \$100k／1MW，2030 年 800V SiC/GaN TAM ~\$2.6B。關鍵：這是**跨 2027–2030 的結構性缺貨**（還與 EV 搶同一批 fab 產能），複製 MLCC-2018 劇本 —— 缺貨把**整本 SiC／power book 的 ASP/毛利持久墊高**，非一次性 blip。**重估**（修正前一版把 800V 當 \$10k/rack、僅 0.8% 營收的過度暫時化）：毛利結構性 re-rate（base 終端 ~43%、blue ~45.5%），三層由 \$29.72/\$57.35/\$83.36 變為 **\$29.64 / \$66.91 / \$103.51**。**Reverse-step** — base \$66.91 vs spot \$78.39 為 **−14.6%**（gap 從 −27% 收斂、更貼市場），base CY27E 淨利對 spot 22x 隱含 ~\$3.2B NI 之 +12% 內 → 守錨非質疑市場。**主要風險** — 全繫於 SiC 定價權；Infineon／onsemi 產能更快到位或車用／工業續弱則崩回 ~\$30 bear。直接 AI-SiC 營收僅 ~2–4%，論點靠整本毛利持久性而非該切片。Confidence：base medium、blue rumor；歷史財報數字為估計值，待 20-F 核對。來源：onsemi IR 2025、ST newsroom、techinvestments.io。
