STMicroelectronics STM

Technology · Semiconductors · Synthos Deep Dive · 2026-07-19

$62.06

Hold

—

The Overview

STMicroelectronics is a big European chipmaker (Franco-Italian roots, Geneva-run, NYSE-listed, reports in U.S. dollars). It makes three main things: power chips — including silicon-carbide chips that control high-voltage electricity in EVs and, increasingly, AI data centers; microcontrollers — the tiny, cheap "brains" (its STM32 family) inside billions of industrial and consumer devices; and analog chips and sensors. Its two biggest customer worlds, cars and factories, have been in a deep slump — revenue fell about a third from 2023 to 2025 and profit almost disappeared.

Two things happened next. First, the slump started ending: sales are growing again, up about 23% year over year last quarter. Second, NVIDIA named STM as one of its partner chipmakers for a new way of powering AI data centers (800-volt DC), and investors got excited — the stock roughly tripled in under a year.

Our read: the business turn is genuine, the NVIDIA connection is genuine — but the price now reflects both. Nobody, including STM, has said how many dollars of STM chips go into each AI rack, and STM is one of fourteen-plus named partners. Meanwhile Chinese chipmakers are undercutting STM's bread-and-butter power and analog chips on price. So we say Hold: keep it if you own it from lower, but this is not the moment to start a new position.

Here's what our three scores mean in everyday terms:

Downside Risk 6/10 (above average). A solid cash cushion, but almost no current profit under the stock, a big fast run-up that can unwind, and a real Chinese competitive threat.
Growth Quality 4/10 (below average). Sales are growing fast right now, but it's recovery growth after a crash, and profit margins and cash flow are still broken.
Exponential Potential 3/10 (low). The AI-power opportunity is real but unquantified and shared with many rivals; the analysts' long-run growth forecast for STM is modest.

The one big worry: China. The sharpest expert fact in our knowledge base says Chinese chipmakers are now "matching and undercutting" STM on its older-technology power and analog chips — the exact products that fund the company while it waits for the AI-power opportunity to pay.

Putting a number on it: our fair-value estimate is $65 against a current price of $62.06 — real upside if our numbers are right.

Our summary metrics

Downside Risk (lower = safer)

6/10 · High

A ~$2.8B net-cash liquidity cushion (cash + short-term investments $4.9B vs $2.1B debt) and a diversified franchise lower it, but trailing GAAP earnings are near zero (TTM P/E ~360×), FCF has been negative two consecutive years, the stock is +95% in 12 months and 53% above its 200-DMA (momentum-unwind risk), and China insourcing names ST directly — that mix is riskier than average.

Growth Quality

4/10 · Moderate

2026E revenue +21% to ~$14.3B is real, but it is recovery growth off a −31% two-year collapse (2023 $17.3B → 2025 $11.8B), with TTM net margin ~1.2%, ROE ~0.9% and negative FCF — the quality metrics are cyclically broken. Mid-cycle this is a 38-40%-gross-margin franchise, which is why the score is 4, not lower.

Exponential Potential

3/10 · Low

The 800 VDC socket is a genuine option — STM is a named NVIDIA ecosystem partner with its own 800V product push — but it is one of 14+ named silicon partners with zero quantified per-rack content, and consensus revenue only compounds ~6%/yr from 2026E to 2030E (thin outer-year coverage). A cyclical $55B-cap recovering, not accelerating.

Fair value$65 $36–$95

What does “fair value” mean?

Fair value is Synthos’s estimate of what one share is worth today, based on our model of the company’s future cash generation and the risks to it. It is not a price target or a prediction of where the stock trades next quarter — it’s the price where we think risk and reward are balanced. Above it, you’re paying for outcomes better than our base case; below it, the market is offering a margin of safety.

The model’s inputs and weightings are proprietary — they’re the product. What we publish is the output, plus an independent, public-math cross-check further down the page (“Check our number”) so you can judge it for yourself.

What do the 5 tiers mean? (Core · Tactical · Watch · Hold · Avoid)

Buy — CoreOwn it as a foundation — start or add now, size it for years, let dips be gifts.

Buy — TacticalGood price + confirmed trend + a defined exit — buy the setup, not a marriage.

WatchWe want the business, just not at this price/setup — act only when the listed trigger hits.

HoldFine to keep if you own it — no reason to buy more; new money does better elsewhere.

AvoidDon't own it — the problem is the business or the expectations, so a cheaper price won't fix it.

The Road Ahead

What we expect to matter in each window, and the evidence that would prove us wrong.

Short term 0–6 months

Neutral

Driver: The stock is digesting a huge run — +122% in 6 months, −22% off the $80 high, below the 50-DMA ($69.5), RSI 36 — and walks into Q2 earnings on 2026-07-23 having missed consensus EPS in each of the last two quarters; expectations embedded at ~48× 2026E leave little room for a soft print.
What we’re watching: A Q2 beat with gross margin holding ~34%+ and a Q3 guide consistent with the ~$14.3B 2026 consensus would stabilize the base; a third straight EPS miss or cautious auto/industrial commentary likely accelerates the pullback toward the mid-$50s.
Confidence: Low

Medium term 6–24 months

Tailwind

Driver: The cyclical recovery is visible in the reported numbers — four straight quarters of sequential/YoY improvement, Q1 2026 revenue +22.8% YoY — and consensus has EPS rebuilding from $0.18 (2025 GAAP) to ~$1.29 (2026E) and ~$2.63 (2027E), with first 800 VDC rack deployments targeted for 2027 adding a thematic kicker.
What we’re watching: The 2027 EPS dispersion is enormous ($1.80 low to $3.38 high, 6 analysts) — evidence of China price pressure spreading from trailing-edge discretes into MCU/analog, or an 800V production slip beyond 2027-28, would collapse the recovery multiple.
Confidence: Low

Long term 2+ years

Neutral

Driver: Two structural forces point in opposite directions — electrification plus AI-datacenter power (800 VDC, SiC, power discretes) is a real multi-year demand wave for exactly what STM makes, while China's aggressive insourcing of 45/90nm power/analog attacks the trailing-edge base that funds it; which force wins is genuinely contested.
What we’re watching: Quantified 800V design-win content and SiC share retention at non-Chinese OEMs would tilt this positive; sustained China share gains in power discretes and any erosion of the STM32 MCU franchise would tilt it negative.
Confidence: Low

Exponential Potential

3/10 · Low

What could take this further than the base case — and where the market may be underpricing it. Full forward-growth and acceleration math in Deeper analysis, §4 below.

Deeper analysis

Technicals, fundamentals, valuation, and the full expert-claim evidence panel — the detail behind the numbers above.

Check our number Market-implied growth ≈ 17%/yr This isn’t how we calculate fair value — it’s public math you can verify yourself. To justify today’s $62, earnings would have to compound roughly 17% a year for 10 years (9% discount rate). Analysts forecast ~-7%/yr, so the market is pricing in MORE than what the Street expects.

Reference table

Street consensus	$74.83 (median $83.50; high $100 / low $34; 15 Buy · 12 Hold · 2 Sell) — wide dispersion; the $34 low is a live bear case
Valuation	~360× trailing GAAP (cyclically meaningless) · ~48× 2026E · ~24× 2027E · ~16× 2028E · EV ~$52.6B (net cash ~$2.8B incl. short-term investments) · EV/2026E sales ~3.7× · P/B ~3.1×
Technicals	Post-blowoff digestion — $62.06, +95% 12-mo (SPY +19%), +193% off the $21.20 low, −22% off the $79.91 high (EOD basis), below the 50-DMA ($69.5), 53% above the 200-DMA ($40.6), RSI 36
Conviction	None — 0 entity-tagged KB claims. The only KB material naming ST (Dylan Patel, Feb 2026, text-mention) is bearish on its trailing-edge base
Position sizing	No new money at $62 — existing holders hold; watchlist add zone ~$50–54 (≈19–20× normalized EPS)

What the experts actually said

No independent expert claims in the Synthos knowledge base yet for STM — this dive is fundamentals- and technicals-driven, not panel-driven.

Price & moving averages 12 months · 50 & 200-day averages · 52-week range

Solid line = price · dashed line = 50-day average · dotted line = 200-day average · the two thin horizontal lines mark the 52-week high and low. Price above both averages is an uptrend.

Bollinger Bands 20-day average ± 2 standard deviations

The shaded band widens when the stock gets more volatile. Riding the upper edge = strong momentum (sometimes stretched); the lower edge = weak / potentially oversold.

RSI (14) momentum gauge · 0–100

Above 70 (overbought zone, shaded) = overbought, below 30 (oversold zone, shaded) = oversold. Currently 39.

MACD 12 / 26 / 9 · trend & momentum

The MACD line crossing above the signal line (bars flip to the up color) = momentum turning up; crossing below (bars flip to the down color) = turning down. Bar height = the size of that gap.

Relative performance vs S&P 500 & its sector (XLK (sector)), set to 100 a year ago

Solid = STM · dashed = S&P 500 · dotted = XLK (sector). A rising line means it is beating that benchmark — the sector line shows whether it is a leader or laggard within its own group.

Forward revenue & earnings actual → estimate · "FY" = fiscal year, "E" = estimate

Darker bars = actual results, brighter = analyst estimates. Taller bars to the right = expected growth.

Key stats an RIA wants

Price$62.06

Market cap$55B

P/E trailing373×

P/E FY26E / FY27E48× / 24×

EV / Sales4.3×

EV / EBITDA22.8×

Gross margin34.0%

Net margin1.2%

Dividend yield0.58%

Beta1.563

52-wk range$21 – $80

RSI(14)36

50 / 200-DMA$69 / $41

12-mo return+95% (SPY +19%)

Street target$75 ($34–$100)

Analyst grades15 Buy · 12 Hold · 2 Sell

FMP ratingB-

Next earnings2026-07-23 (Q2 2026 earnings — consensus revenue ~$3.45B, EPS $0.26 per FMP calendar; watch gross margin, auto/industrial orders, SiC trajectory and any 800V/data-center content commentary)

1. What it is

STMicroelectronics (NYSE: STM) is one of Europe's two flagship semiconductor houses (alongside Infineon), incorporated in the Netherlands, operationally headquartered in Geneva, and listed in New York, Paris and Milan. It reports in U.S. dollars — every statement in our data file carries reportedCurrency: USD, so there is no currency-blending issue in the numbers below (a meaningful share of its cost base is euro-denominated, which is an FX margin sensitivity, not a reporting quirk). CEO Jean-Marc Chery; ~49,000 employees; an integrated device manufacturer that owns most of its fabs — hence the heavy capex you'll see in §5.

Revenue mix — what our file shows and what it can't. The FMP product segmentation collapses to a single "Product" line ($11.75B of the $11.99B FY2025 total... note the FMP income statement shows FY2025 revenue $11.84B; the segment table sums to $11.99B — a small vendor reconciliation gap we flag rather than hide). For the split that matters, we must go outside the file, clearly labeled (external, STM company reporting, approximate):

Power & discrete / analog / MEMS & sensors — one of STM's two product groups; this is where the silicon-carbide (SiC) franchise lives. Company-disclosed SiC revenue was ~$1.1B in 2023 (external, STM disclosures) — high-single-digit percent of sales — and management's earlier ~$2B SiC ambition was pushed out as EV demand rolled over. This group is the 800 VDC vehicle.
Microcontrollers, digital ICs & RF — the STM32 general-purpose MCU franchise, one of the broadest MCU ecosystems in the world, plus smartphone/personal-electronics content. This is the recurring cash engine.
End markets (external, approximate): automotive and industrial together are the clear majority of demand — which is exactly why 2024-25 was so painful (both were in synchronized inventory-correction downcycles) and why the recovery now underway matters so much.
Geography caveat: the file's geographic table (Singapore $6.56B = ~55% of FY2025) is bill-to/ship-to logistics hubs, not end demand — do not read it as customer geography.

The strategic story: STM sits at the intersection of two power-electronics waves — vehicle electrification (SiC traction inverters, onboard chargers) and now AI data-center power (NVIDIA's 800 VDC architecture, where STM is a named silicon partner and issued its own 800V product push referencing NVIDIA as recently as March 2026 — verified, external). Against that: China's state-backed insourcing of trailing-edge power and analog silicon is attacking the installed base.

2. The expert thesis — why the panel is bullish (traceable)

There is no entity-tagged expert coverage of STMicroelectronics in the Synthos knowledge base. total_claims = 0, net_bullish_voices = 0, breadth 0. No tracked voice has published a distilled, traceable claim carrying STM as a tagged entity, so this dive carries no conviction rating and cites zero claim_ids — fabricating conviction is against the house standard. The verdict below is entirely fundamentals- and quant-driven.

Two pieces of context are worth stating plainly, both labeled for what they are:

The only KB material that names ST at all is bearish, and it is a text-mention, not tagged coverage. Dylan Patel (2026-02-03/05, conviction 60–70 in the source claims) argues China's power-chip makers are now "matching and undercutting Texas Instruments/ST Micro" on trailing-edge (45/90nm) power/analog, with China aggressively insourcing that node class. ST appears in the thesis text of claims tagged to other entities. We treat this as thematic context — but it is the sharpest single fact in this dive, and it points against the stock.
The 800 VDC frame comes from Synthos's own think piece (the 800 VDC data-center power architecture note on the research site), not from panel conviction. NVIDIA announced the architecture May 20, 2025 and named STM among 14+ silicon partners (Infineon, TI, onsemi, MPS, Navitas, ROHM, ADI, Renesas, Innoscience, AOS, EPC, Power Integrations, Richtek…). Named ≠ paid: no public source quantifies per-rack dollar content or design-win share for any partner, NVIDIA's claimed benefits (up to 5% efficiency, 70% maintenance, 30% TCO, 45% copper) are unaudited vendor numbers, and full production targeted for 2027 has a reported slip risk to 2028 (SemiAnalysis via CNBC, 2026-07-06; NVIDIA denies). All external, dated, verified in our 2026-07-19 research run.

Readers who require expert-panel corroboration should treat this as a quant/fundamental screen with a thematic overlay, not a conviction call.

3. Synthos scores & the Bull / Base / Bear cases

The one-glance judgment — three scores, 0–10, each anchored to real metrics (not probabilities we can't honestly calibrate):

Score	0–10	The read
Downside Risk (lower = safer)	6 · Above average	The cushion is real: cash + short-term investments $4.9B vs total debt $2.1B → ~$2.8B net cash, current ratio 3.3×, a diversified product base and an investment-grade balance sheet. Against it: trailing GAAP earnings near zero (TTM P/E ~360× — no earnings floor under the price), FCF negative in both 2024 (−$216M) and 2025 (−$52M), a stock up +95% in 12 months and 53% above its 200-DMA (momentum-unwind risk), two consecutive quarterly EPS misses, and a named structural attacker (China, §2). More risk than the balance sheet alone suggests.
Growth Quality	4 · Below average	2026E revenue +21% to ~$14.3B (consensus) and Q1 2026 +22.8% YoY are real — but this is recovery growth off a −31% collapse (2023 $17.3B → 2025 $11.8B). Current quality metrics are broken: TTM net margin ~1.2%, ROE ~0.9%, gross margin 33.9% (FY25) vs 47.9% at the 2023 peak, FCF negative. The franchise argument (mid-cycle 38-40% gross margins, STM32 ecosystem, net cash) keeps this at 4.
Exponential Potential	3 · Low	The 800 VDC socket is a genuine, verified option — but one of 14+ named partners with zero quantified content, on an architecture whose production timing (2027, possibly 2028) is itself contested. Consensus revenue 2026E ~$14.3B → 2030E ~$18.0B is only a ~6% CAGR — and the outer years are thin (2–5 analysts) and non-monotonic (2028E $18.8B > 2029E $16.7B), so we don't lean on them. A $55B-cap cyclical recovering, not accelerating.

The three cases (our own scenario model — assumptions shown; each target is a ~12–18-month fair value). We deliberately do not attach probabilities: the base case is the expected path, so a weighted blend would just restate it with false precision.

Case	Key assumptions	Fair value
Bull	Auto/industrial recovery compounds, SiC share holds at non-Chinese OEMs, 800 VDC ramps on schedule in 2027 with STM winning visible content; 2028E EPS lands near the high (~$4.16) and the market pays ~22× for a franchise with a proven AI-power leg.	~$95 (+53%)
Base (our anchor)	Consensus roughly hits — 2026E EPS ~$1.29, 2027E ~$2.63, 2028E ~$3.91. Two cross-checking methods (§6): normalized mid-cycle EPS ~$2.64 at 22–24×, and 2028E $3.91 at ~18× discounted a year — both land in the low-to-mid $60s.	~$65 (+5%)
Bear	China price pressure spreads from trailing-edge discretes into analog and brushes the MCU franchise; the recovery stalls, 2027 EPS lands near the consensus low ($1.80), the 800V option stays unquantified, and the multiple de-rates to ~18-20× a lower base. The Street's $34 low target marks the same zone.	~$36 (−42%)

Synthos fair value = the base case, ~$65 (+5%), with the full $36–$95 span as the honest range. Upside to base is ~5% — inside the noise band — while the bear leg (−42%) is as long as the bull leg (+53%). We sit below the Street's $74.83 consensus because the sell-side path largely assumes the recovery and multiple both hold while China stays contained; the Patel evidence says that containment is exactly what's in doubt. That asymmetry-neutral setup at a full price is a textbook Hold. This is a tracked call — the Forecaster Scorecard grades it once it matures.

4. Exponential Potential

Synthos separates compounders (durable returns on real demand) from exponentials (accelerating, multi-baggers-from-here). STM is a cyclical recovery story with a thematic option attached — not an exponential:

Forward growth: consensus revenue $14.30B (2026E) → $16.73B (2027E) → $18.84B (2028E) is a real +21%/+17%/+13% recovery ramp — then the estimate path flattens and even inverts ($16.7B 2029E, $18.0B 2030E on 5 analysts), leaving a ~6% 2026E→2030E CAGR. The acceleration is cyclical snap-back, not structural inflection.
The 800 VDC option (real, unquantified): NVIDIA's architecture moves facility power to 800VDC with power conversion silicon (SiC/GaN rectification, DC/DC stages) exactly in STM's wheelhouse; STM is a named partner and is marketing its own 800V line (March 2026 press release referencing NVIDIA — external, verified). But no source quantifies per-rack content for anyone, there are 14+ named partners including every serious power house, and first full production is 2027 with a reported 2028 slip risk. We treat it as a free-ish call option at today's multiple — worth monitoring, not worth capitalizing.
SiC/EV: the original exponential thesis (SiC to ~$2B+) was pushed out by the EV downcycle and Chinese SiC entrants; ~$1.1B (2023, company disclosure, external) is the last hard number.
Room to run: a $55B-cap IDM with heavy capex (17% of revenue TTM) does not multi-bag without a margin regime change; the 800V + edge-AI MCU combination could eventually provide one, but nothing in the current numbers evidences it yet.

Exponential Potential: Low (3/10). A recovering cyclical with a genuine but unpriced-in-either-direction AI-power option — the honest description is "cheap option, expensive stock."

5. Financials (real numbers — FMP annual/quarterly)

Revenue (the downcycle, in full): FY2023 $17.29B (peak) → FY2024 $13.27B (−23.2%) → FY2025 $11.84B (−10.8%) — a −31.5% two-year drawdown as automotive and industrial corrected simultaneously. For scale, FY2021 was $12.76B: the company round-tripped four years of growth.
The recovery (visible and accelerating): quarterly revenue Q1 2025 $2.52B → Q2 $2.80B → Q3 $3.17B → Q4 $3.33B → Q1 2026 $3.095B, +22.8% YoY (Q1 is seasonally softer sequentially). Consensus for Q2 2026 (reports 2026-07-23) is ~$3.45B, +23% YoY, and full-year 2026E $14.30B (+21%).
Margins & earnings (the damage): FY2025 gross margin 33.9% vs 39.3% (FY2024) and 47.9% (FY2023) — a 14-point peak-to-trough collapse. FY2025 operating income $324M (2.7% margin) vs $4.61B (26.7%) in FY2023; FY2025 GAAP net income $166.5M, diluted EPS $0.18 vs $4.46 in FY2023. Q1 2026: gross margin 33.8%, net income $37M, GAAP diluted EPS $0.04. (Basis flag: FMP's earnings-surprise feed shows Q1 2026 EPS $0.13 — an adjusted/non-GAAP basis — versus $0.04 GAAP diluted in the income statement. Both are in our file; we label whichever we use.)
One-off flag: Q4 2025 shows a GAAP net loss of −$30M driven by a $170.9M tax expense against only $146.9M of pretax income — a tax-item quarter, not an operating collapse. (The same FMP quarter carries a flat 1,000M share count, a vendor artifact; the annual weighted diluted count is 923M.)
Cash flow (the sore spot): FY2025 operating cash flow $2.15B against capex $2.20B → FCF −$52M; FY2024 FCF −$216M (OCF $2.97B, capex $3.18B); FY2023 FCF +$1.46B. Two consecutive negative-FCF years — the cost of building capacity (Agrate/Crolles 300mm, Catania SiC) through the trough. Capex is 17.2% of TTM revenue; R&D another 16.2%.
Balance sheet (the cushion): cash $2.84B + short-term investments $2.09B = $4.92B liquid, vs total debt $2.13B → ~$2.8B net cash (cash-only basis: +$704M net cash). Equity $17.8B; current ratio 3.3×. Inventory $3.14B at 136 days — still elevated for this point in a recovery; watch it.
Shareholder returns despite negative FCF: FY2025 dividends $321M + buybacks $367M ≈ $688M, funded from the investment portfolio — sustainable only because of the cushion.

6. Valuation — priced in or room?

The uncomfortable arithmetic: at $62.06, STM trades at ~360× trailing GAAP EPS (cyclically meaningless, but it means there is no current-earnings floor), ~48× 2026E EPS ($1.29), ~24× 2027E ($2.63), and ~16× 2028E ($3.91). EV is ~$52.6B netting the full $2.8B liquid cushion (cap $55.4B − net cash) — EV/2026E sales ~3.7×, EV/2026E EBITDA ~12.6×, P/B ~3.1× ($19.64 book/share). Dividend yield 0.58%. (Vendor flag: FMP's own TTM key-metrics block computes EV at $55.6B — implying ~$0.9B net debt — which contradicts the balance sheet in the same file once short-term investments are counted. We compute EV ourselves and say so.)

Our fair-value method (explicit): two cross-checks. (1) Normalized mid-cycle earnings: the 2021–2025 GAAP diluted EPS average is $2.64 — which, tellingly, is almost exactly the 2027E consensus ($2.63). At a 22–24× mid-cycle multiple (discount to U.S. analog peers for structurally lower margins and direct China exposure; premium to bare cyclicals for net cash, the STM32 ecosystem and the 800V/SiC position), that is $58–63. (2) Recovery-year comp: 2028E EPS $3.91 at ~18× = $70, discounted a year at 10% ≈ $64. Blend → base ~$65 (+5%).

Street context: consensus target $74.83, median $83.50, high $100, low $34 — a 3× spread that is itself the message: this name is a referendum on whether the recovery multiple survives China. Ratings: 15 Buy / 12 Hold / 2 Sell — the least conviction-loaded sell-side book we've covered recently. FMP's letter rating is B- (P/E sub-score 1/5 — the screen sees the same thing we do).

Bottom line: a year ago at $21–30 this was the cheap way to own the power-semi theme. At $62 the recovery is paid for and the 800V option is partly paid for; what's not paid for is the China risk. Fully priced — Hold.

7. Technicals (from the tech block)

Trend: post-blowoff digestion. $62.06 is below the 50-DMA ($69.49) and a full 53% above the 200-DMA ($40.64) — an extremely extended long-term posture even after the pullback. MACD is negative (−1.35); the momentum phase has rolled.
Location: −22.3% off the 52-week high ($79.91 EOD basis; $81.42 intraday) and +193% off the 52-week low ($21.20). The entire investment debate is about a stock that already nearly quadrupled trough-to-peak inside twelve months.
Momentum: RSI(14) 36 — approaching oversold on the daily, but "oversold" 53% above the 200-DMA is a correction within an extended trend, not a washout.
Relative strength: +94.7% 12-mo vs SPY +19.1% / QQQ +24.8%; +122.6% over 6 months vs SPY +7.7%. Massive outperformance now consolidating.
Read: technicals say let it come to you. There is no meaningful moving-average support until far below; the mid-$50s is the first zone where the post-earnings air pocket likely finds buyers, and our ~$50–54 add zone (§12) coincides with ~19–20× normalized EPS. A reclaim of the 50-DMA on a clean Q2 print would neutralize the caution.

8. Moat & competitive position

STM's moat is real but under direct attack at its edges: (1) the STM32 ecosystem — one of the world's broadest 32-bit MCU franchises with decades of design-in stickiness and toolchain lock-in; this is the highest-quality asset in the company; (2) SiC process leadership and vertical integration (including in-house substrate work) built through the Tesla-era ramp — still top-tier globally, but Chinese SiC entrants have compressed the technology gap faster than expected; (3) IDM scale in power/analog — owning fabs is a cost advantage at high utilization and a millstone at low utilization (see the 14-point gross-margin collapse); (4) customer position at essentially every non-Chinese auto and industrial OEM, plus the named seat in NVIDIA's 800 VDC ecosystem.

The attack surface is precise, and our KB's sharpest relevant fact names it: China's power-chip makers are "matching and undercutting Texas Instruments/ST Micro" on trailing-edge (45/90nm) power/analog (Dylan Patel, Feb 2026, text-mention — §2). Trailing-edge power/discrete/analog is commodity-adjacent; the defensible core is the MCU ecosystem and leading-edge SiC/GaN for demanding sockets (traction inverters, 800V rack power). The bear case is China compressing the commodity layer faster than the defensible layer grows.

Peer set: Infineon (the closest mirror — European IDM, power + auto + MCU), onsemi (SiC/auto pure-ish play; in our file at $34B cap), Texas Instruments and NXP/Renesas (analog/MCU), and the 800V partner cohort (MPS, Navitas, Power Integrations…) as theme comps. (Note: FMP's auto-generated "peers" list — CDW, Check Point, VeriSign etc. — is index-cluster noise, not a comp set; we discard it.) Versus Infineon and onsemi, STM's relative discount on recovery earnings has largely closed after the run.

9. Management, capital allocation & guidance

Capital allocation: an IDM investing through the trough — capex $2.2B in 2025 (down from $3.2B in 2024 and $4.5B in 2023 as management cut into the downturn), concentrated in 300mm (Crolles/Agrate) and SiC (Catania). Shareholder returns continued through negative FCF (dividends $321M + buybacks $367M in 2025), funded from the $4.9B liquidity pool — defensible given net cash, but it is portfolio drawdown, not earned cash.
Execution record (recent — mixed): against FMP consensus, STM beat Q3 2025 (EPS $0.29 vs $0.22) but missed Q4 2025 badly (EPS $0.11 vs $0.27) and missed Q1 2026 (EPS $0.13 vs $0.19, adjusted basis) while revenue modestly beat both quarters — the misses are margin/below-the-line, not demand. Two straight EPS misses into a stock priced at 48× current-year earnings is the single most uncomfortable pairing in this dive.
Restructuring: the 2025 GAAP quarters carry visible one-offs (Q2 2025 GAAP loss −$98M; Q4 2025 tax-driven loss −$30M) consistent with the company's announced cost-reset program (external context). The 2026E consensus EPS of $1.29 embeds that program landing.
Guidance posture: management guides quarterly (revenue and gross margin); treat forward commentary on the 2026-07-23 call — especially any first quantification of 800V/data-center pipeline — as management's own book, half-weighted by design.

10. Catalysts & what to watch

Next earnings: 2026-07-23 (Q2 2026 — four days after this dive's as-of date). Consensus (FMP calendar): revenue ~$3.45B (+23% YoY), EPS $0.26. Key lines: gross margin (33.8% last quarter; direction matters more than level), Q3 guide vs the ~$14.3B full-year consensus, inventory days (136 — should be falling), auto vs industrial order commentary, and any dollar-quantification of data-center/800V content — the single item that would most change this dive.
800 VDC timeline (external, verified): NVIDIA targets full 800VDC production 2027 (Kyber/Rubin Ultra); SemiAnalysis via CNBC (2026-07-06) reports a slip to 2028, NVIDIA denies. Watch OCP updates and partner BOM disclosures; a confirmed slip hits the theme premium across the whole partner cohort.
China trailing-edge pricing: evidence in peer prints (TI, onsemi, Infineon) of trailing-edge power/analog ASP erosion — Patel's thesis showing up in income statements — is the bear tripwire even if STM's own quarter looks fine.
SiC/EV datapoints: any refresh of SiC revenue disclosure (last hard number ~$1.1B, 2023, external) and design-win retention at non-Chinese OEMs.
FCF inflection: with capex guided down and revenue recovering, 2026 should mark the return to positive FCF; a third negative year would break the capital-returns story.

Thesis tripwires (what would change the call): downgrade to Watch/Avoid on (a) a third consecutive EPS miss or a Q3 guide implying 2026 revenue materially below $14B; (b) gross margin failing to recover through the mid-30s during 2026; (c) confirmed China price pressure reaching 32-bit MCU or leading SiC sockets; (d) an 800V production slip beyond 2028. Upgrade toward Buy — Tactical on (a) a reset into the ~$50–54 zone with fundamentals intact, or (b) quantified 800V/data-center content plus gross margin back on a path toward 38-40% — either would restore the asymmetry this price has spent.

11. Key risks

China insourcing of trailing-edge power/analog (the dominant risk): the KB's sharpest relevant fact names ST directly as being matched and undercut at 45/90nm. This attacks the revenue base that funds the SiC and 800V investments — a structural, state-backed, price-led threat, not a cycle.
Valuation/momentum unwind: ~48× current-year earnings, +95% in 12 months, 53% above the 200-DMA, no trailing-earnings floor — the stock can fall a long way on no new fundamental information.
Recovery fragility: two straight EPS misses; gross margin 14 points below peak; inventory 136 days; auto end-demand (EV mix, European production) still uneven.
800V option risk: named ≠ paid — 14+ partners, zero quantified content, contested 2027/2028 production timing, and all headline benefits are unaudited vendor numbers.
IDM capex intensity: 17% of revenue in capex means downturns burn cash (FCF negative 2024 and 2025); operating leverage cuts both ways.
Governance/FX structure: French-Italian state shareholder influence over strategy and management; euro cost base vs USD reporting is a live margin sensitivity.
No expert corroboration: zero entity-tagged panel support in the Synthos KB — the call rests on fundamentals and quant, and the only text-mention that exists is bearish.

12. Verdict, position sizing & monitoring

Hold. The question this dive was commissioned to answer — is STM the cheap, real-revenue way to own the 800V power theme, or a value trap? — resolves to: it was the cheap way at $21–30; at $62 it is the fairly-priced way, with a live structural bear case. The recovery is real (Q1 2026 +22.8% YoY, 2026E +21%), the balance sheet is genuinely strong (~$2.8B net cash), the STM32 franchise is high quality, and the 800V seat is verified. But the market has paid for all of it in advance: ~48× 2026E, ~24× 2027E, a +5% gap to our $65 base fair value, negative FCF two years running, two straight EPS misses, and China undercutting the trailing-edge base by name. The bull leg (+53% to $95) and bear leg (−42% to $36) are nearly symmetric — and symmetric risk at a full price is a Hold, not a Buy.

Sizing: no new money at $62. Existing holders from materially lower levels: hold, with the §10 tripwires as the exit discipline. For new capital, the add zone is ~$50–54 (≈19–20× normalized mid-cycle EPS of $2.64, and the first technically credible support region) — provided the fundamentals tripwires haven't fired on the way there. Quantified 800V content would justify entry at higher prices; a slogan-only 800V story does not.
Monitoring: the 2026-07-23 print is the immediate test (gross margin, Q3 guide, inventory, any 800V quantification). Formal re-score on any ±20% move, the estimate-revision cycle, peer evidence of China trailing-edge price erosion, or any 800V timeline/content news. This verdict is logged as a tracked Synthos call as of 2026-07-19 at $62.06 (2026-07-17 close).
Single biggest risk: China's trailing-edge undercutting compressing the commodity layer of STM's power/analog base faster than the defensible layer (STM32, leading SiC, 800V) grows — the bear case (~$36) marks where that scenario prices.

Provenance & disclosures

Traceability: 0 entity-tagged KB claims, breadth 0 — no claim_ids are cited because none exist for this entity. The Dylan Patel material (2026-02-03/05, conviction 60–70) names ST in thesis text of claims tagged to other entities; it is used here as labeled thematic context only, never counted as coverage. Fabricated conviction is structurally impossible (claim-ID reconciliation); this note states plainly it is fundamentals-/quant-driven.
Data as-of: fundamentals 2026-03-31 (Q1 FY2026) · estimates and calendar from the 2026-07-19 FMP pull (v800_data/STM_data.json) · price $62.06 = 2026-07-17 close per the FMP quote timestamp. Forward figures are analyst consensus (FMP), labeled as estimates throughout. 800 VDC facts (partner list, timelines, March 2026 STM press release, SemiAnalysis/CNBC slip report) are external, dated, and were adversarially verified in the 2026-07-19 Synthos research run; NVIDIA's efficiency/TCO benefit claims are unaudited vendor numbers.
Currency & entity note: all statements in the data file are reportedCurrency: USD — STM genuinely reports in U.S. dollars, so no currency blending was performed. FMP's profile lists country "CH" (Geneva operational HQ); the company is incorporated in the Netherlands and its NYSE line is a U.S.-registry share of an EU issuer. Euro-denominated costs vs USD revenue is a real margin sensitivity, flagged in §11.
Vendor-data flags (kept visible, not smoothed): (1) FMP's TTM key-metrics EV ($55.6B) contradicts the same file's balance sheet once short-term investments are counted — we compute EV (~$52.6B) ourselves; (2) the earnings-surprise feed's Q1 2026 EPS ($0.13, adjusted) differs from GAAP diluted ($0.04) — bases labeled where used; (3) Q4 2025 carries a flat 1,000M share count artifact and a tax-driven GAAP loss; (4) outer-year estimates (2029–2030) rest on 2–5 analysts and are non-monotonic — we do not lean on them; (5) the FMP product-segment table post-2016 collapses to "Product" and its FY2025 sum ($11.99B) differs slightly from the income statement ($11.84B); segment/SiC splits are therefore external and labeled; (6) the geographic table is ship-to hubs (Singapore ~55%), not end demand; (7) the earnings-calendar rows carry a lastUpdated: 2026-07-20 vendor stamp one day ahead of this dive's as-of date — a feed-timestamp quirk, noted rather than hidden; (8) FMP's auto-generated peer list is index-cluster noise and was discarded in favor of a hand-built comp set (§8).
Non-GAAP note: trailing multiples use GAAP diluted EPS; the forward P/E path uses FMP consensus EPS (the Street's adjusted basis). Normalized EPS ($2.64) is the 2021–2025 GAAP diluted average, method shown in §6.
Management caveat: STM's guidance and any 800V pipeline commentary are management's own book, half-weighted by design.
Not investment advice. Independent research, educational and informational only, never personalized. Hypothetical/forward figures are labeled; the only performance numbers Synthos will headline are the live, real-money Flagship's.
Version: 2026-07-19. Prior versions available via the deep-dive version dropdown ("based on the info at the time").