Technology · Semiconductors · Synthos Deep Dive · 2026-07-03
| Verdict | Hold — systematic Synthos tier |
| Price (2026-07-02) | $91.22 · market cap ~$35.7B |
| Synthos scores (0–10) | Downside Risk 6 · Growth Quality 5 · Exponential Potential 5 |
| Synthos fair value (base case) | ~$95 → +4% · full range $53 (bear) – $130 (bull) |
| Street consensus | $109 (high $140 / low $62; median $116; 25 Buy · 21 Hold · 1 Sell) — context, not our anchor |
| Valuation | 64× trailing EPS (trough) · ~30× FY26E · ~21× FY27E · ~16× FY28E · EV/S 6.1× · EV/EBITDA 30× |
| Exponential Potential | 5/10 · Moderate — AI-datacenter and SiC-EV lines are genuinely accelerating, but it's a cyclical still below its prior revenue peak |
| Technicals | Mixed — $91, −32% off 52-wk high, below the 50-DMA ($111), above 200-DMA ($71), RSI 38, MACD −4.9, but +70% 12-mo |
| Conviction | Low — 0 expert voices in the KB; this is a quant/fundamentals call only |
| Position sizing | Watch-list; if bought, small/tactical (~1–2%) with a plan tied to the recovery confirming |
| Next catalyst | 2026-08-03 Q2'26 earnings (Street EPS $0.71, rev ~$1.59B) |
| Single biggest risk | The cyclical recovery stalls — a beta-1.98 semi on 64× trough earnings has a long way to fall if demand rolls over |
One-line thesis. onsemi is a high-beta power-and-sensing semiconductor company that has "moved beyond the cyclical trough" (management's own words, Q1'26) with a genuinely accelerating AI-datacenter and silicon-carbide EV story — but the stock has already rallied ~70% in twelve months, trades at 64× trough earnings / ~30× FY26E, and our base-case fair value of ~$95 sits only marginally above the price, so the risk/reward says Watch, not chase.
onsemi makes the power chips and sensors that run electric cars, factory equipment, solar/energy-storage systems, and — increasingly — the power systems inside AI data centers. Think of it as the plumbing that moves and controls electricity inside modern machines.
The business is cyclical: like a rollercoaster, its sales rise and fall with the economy and the car/industrial order cycle. Sales fell from about $8.3 billion (2022–23) to about $6.0 billion (2025) as customers worked off excess inventory. Management now says the worst is over and things are turning up — and one bright spot, the AI-data-center business, more than doubled year over year.
Is the stock cheap or expensive? On last year's depressed profits it looks expensive (you pay $64 for every $1 of trailing profit). On next year's expected recovery it looks more reasonable (~$30, then ~$21 the year after). The stock has already jumped ~70% in a year, so a lot of the good news is priced in.
Our verdict is Watch — a solid company in an up-cycle, but the price already reflects much of the recovery, and it's a volatile name that can drop hard if the economy wobbles.
Here's what our three scores mean in everyday terms:
The one big worry: if the recovery stalls (weak car sales, soft industrial demand), a high-flying, high-volatility chip stock on rich trailing earnings can fall a long way.
Solid = price · dashed = 50-day average · dotted = 200-day average · amber = 52-week high/low. Price above both averages is an uptrend.
The shaded band widens when the stock gets more volatile. Riding the upper edge = strong momentum (sometimes stretched); the lower edge = weak / potentially oversold.
Above 70 (red band) = overbought, below 30 (green band) = oversold. Currently 39.
Blue crossing above amber (bars flip green) = momentum turning up; below (bars red) = turning down. Bar height = the size of that gap.
Solid = ON · 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.
Darker bars = actual results, brighter = analyst estimates. Taller bars to the right = expected growth.
Every claim reconciles to a real claim_id in the Synthos knowledge base — this is the evidence the verdict is built on, not vibes. Management (the company itself) is shown but half-weighted; one cautionary voice is included on purpose.
ON Semiconductor (Nasdaq: ON), which markets itself as onsemi, is a Phoenix/Scottsdale-based global supplier of intelligent power and sensing semiconductors — analog and discrete devices, power modules, and image sensors that handle power switching, energy conversion, signal conditioning, circuit protection, and imaging. Founded 1992, ~26,400 employees, CEO Hassane El-Khoury. The end-markets are automotive (EV powertrains, ADAS sensing), industrial (energy storage, solar, factory automation), and a fast-rising AI data center power business. Fiscal year ends late December/early January.
Segments. onsemi reports three groups — Power Solutions Group (PSG), Analog & Mixed-Signal Group (AMG), and Intelligent Sensing Group (ISG). (FMP's segment feed is incomplete for FY2025 — it captured PSG $2,805M and ISG $928M but dropped AMG for the full year. The cleaner, current read comes from the Q1'26 earnings release:)
Revenue mix (Q1'26, from the 8-K):
Geography. FMP's geographic feed for onsemi is unreliable (it reports ship-to/bill-to hubs — Hong Kong, Singapore, UK — that reflect distribution and contract-manufacturing locations, not true end-demand, and drops to a single "Other" line in FY2025). We therefore do not lean on the geographic split; the honest read is that onsemi's demand is global, Asia-weighted through the electronics supply chain, with automotive and industrial as the swing factors.
The strategic story management keeps returning to: electrification (900V EV architectures with EliteSiC, wins with Geely and NIO), software-defined vehicles (Treo-based Ethernet at a North American OEM), and the newest leg, AI-datacenter power (revenue more than doubled YoY, +30% sequentially in Q1'26).
There is no expert coverage for onsemi in the Synthos knowledge base. total_claims = 0, breadth 0, net conviction 0. No net-bullish voices and no cautionary voice have been distilled for this name.
That means this deep dive carries no claim_id citations — and, per the Synthos house standard, we will not manufacture any. Everything below is driven by the hard fundamentals (FMP filings), live analyst estimates, the technical block, and management's own earnings release (half-weighted, §9). Treat the conviction rating as Low accordingly: this is a quant/fundamentals judgment, not a conviction-panel call. Where the wider Street sits is shown as context (25 Buy / 21 Hold / 1 Sell, consensus target $109), not as our anchor.
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 · Elevated | Balance sheet is healthy — net-debt/EBITDA 0.8×, current ratio 4.9× — but beta 1.98, deep semiconductor cyclicality, a −32% drawdown already in the tape, and 64× trailing earnings on a trough base leave little cushion if demand rolls over. |
| Growth Quality | 5 · Middling | Forward EPS rebounds hard (FY26E→FY28E ~37% CAGR), but it's a cyclical recovery off a depressed base, not durable secular growth; ROE 7.4% and ROIC 5.4% are soft, and margins swung from a −$486M Q1'25 charge to healthy non-GAAP profitability. |
| Exponential Potential | 5 · Moderate | Real accelerants — AI-datacenter revenue >2× YoY, EliteSiC EV design wins — and room to run at a $36B cap, but the overall franchise is still below its 2022–23 revenue peak; it compounds cyclically, it doesn't obviously multi-bag. |
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 by definition the expected path, so a weighted blend would just restate it with false precision. Instead the cases bound the range, and the scores above summarize them.
| Case | Key assumptions | Fair value |
|---|---|---|
| Bull | Recovery accelerates; AI-datacenter and SiC/EV scale faster than modeled; FY27E EPS beats to ~$5.00 (vs ~$4.28 cons) and the market pays a cycle-peak ~26×. | ~$130 (+43%) |
| Base (our anchor) | Estimates roughly hit — FY27E EPS ~$4.28 — and a mid-cycle recovering cyclical earns a ~22× multiple. | ~$95 (+4%) |
| Bear | Recovery stalls (auto/industrial demand softens, inventory rebuilds); FY27E EPS misses to ~$3.30 and the multiple de-rates to a cycle-trough ~16×. | ~$53 (−42%) |
Synthos fair value = the base case, ~$95 (+4%), with the full $53–$130 span as the honest range. Note the asymmetry: our base sits just above the current price, our bear ($53) is near the Street's low ($62) and would be a −42% move, while the bull ($130) approaches the 52-week high. That skew — limited base-case upside, large cyclical downside — is exactly why the verdict is Watch, not Buy. Our base is slightly below the Street's $109 consensus because we haircut the multiple for cyclicality and the already-large run. This is a tracked call — the Forecaster Scorecard grades it once it matures.
Synthos separates compounders (durable high returns on capital) from exponentials (accelerating, multi-baggers-from-here). onsemi is neither cleanly — it is a recovering cyclical with two genuinely exponential product lines bolted onto a mature base:
Exponential Potential: Moderate (5/10). Own the idea — AI-datacenter power and SiC-EV are legitimate secular growth pockets — but the vehicle is a beta-2 cyclical still climbing back to its old peak. A small, accelerating pure-play on AI power would score higher; onsemi's mature analog base and cyclicality cap it at moderate.
onsemi is not cheap on trailing numbers (64× GAAP EPS, 6.1× sales, 30× EV/EBITDA) — but the trailing EPS is a trough distorted by the Q1'25 charge, so trailing P/E overstates the richness. The forward picture is the real debate: on live consensus the P/E is ~30× (FY26E, $3.08) → ~21× (FY27E, $4.28) → ~16× (FY28E, $5.79). For a cyclical rebounding into an up-leg, ~21× forward FY27 is a fair-to-full multiple — you are paying up for the recovery, not buying it cheap.
A simple reverse read: to justify today's ~$91 at a mid-cycle ~22× multiple, the market needs onsemi to earn ~$4.15+ — essentially the FY27 consensus. In other words the stock is priced for the recovery landing on schedule, with the ~70% twelve-month run having already discounted much of it. Street targets (context): consensus $109, median $116, high $140, low $62 — a wide band that itself signals how cyclical-uncertain the name is. Our $95 base FV is deliberately below consensus: we haircut the multiple for beta-2 cyclicality and the size of the prior move. FMP's own letter rating is B− (price-to-earnings sub-score 1/5 — i.e. flagged as richly valued). Not a value buy; a fairly-priced cyclical recovery where the easy money may already be made.
onsemi's moat is moderate, not wide. Its edges: (1) a leadership position in silicon carbide (EliteSiC) for EV powertrains and, increasingly, AI-datacenter power — a genuinely differentiated, capacity-intensive technology; (2) scale and breadth across power discretes, analog, and image sensors, with deep automotive design-in relationships (Geely, NIO, a North American OEM); (3) long design-win cycles in automotive/industrial that create switching costs. Against that: power and analog semis are competitive and partly commoditized, the business is cyclical, and returns on capital (ROIC 5.4%, ROE 7.4%) are currently sub-par — a moat that exists but isn't throwing off elite economics at this point in the cycle.
Peer set (FMP-supplied, market cap). FMP's peer list for onsemi is a poor comp set — it mixes in IT-distribution and software names (CDW, Check Point, SS&C, Tyler, Toast, Jacobs) that are not power-semi peers. The relevant comparables in the list are the semiconductor names: STMicroelectronics ($60.7B) — the closest direct power/analog/automotive comp — plus foundries GLOBALFOUNDRIES ($38.3B), United Microelectronics ($61.2B), and materials supplier Entegris ($22.3B). The truer peer group (not fully in this list) is STM, Infineon, Texas Instruments, and Wolfspeed in silicon carbide. Against STM, onsemi carries a richer forward multiple, justified only if its SiC/AI-power mix wins share.
Thesis tripwires (what would change the call): two consecutive quarters of sequential revenue decline; gross margin slipping back below ~35%; AI-datacenter growth stalling; or a break of the 200-DMA (~$71) that would signal the cyclical recovery is failing.
Watch. onsemi is a fundamentally healthy company (net-debt/EBITDA 0.8×, $1.4B FCF, recovering ~39% gross margin) in a genuine cyclical up-leg, with two real secular bright spots in AI-datacenter power and silicon-carbide EV. But three things keep it off the Buy list: (1) the stock has already rallied ~70% in twelve months and our base-case fair value (~$95) implies only ~4% upside; (2) it's a beta-2 cyclical on rich trailing earnings with a −42% bear case — an unattractive risk/reward skew; and (3) there is no expert conviction behind it in the KB. The recovery is real but largely priced.
claim_id citations. The verdict is fundamentals- and quant-driven only. Fabricated conviction is structurally impossible (claim-ID reconciliation); here there is simply nothing to cite, and we say so plainly.