VectorialDATA

ON — ON Semiconductor

TechnologyNorth AmericaUnited StatesBlockchain certified

Price when we picked it (Jul 16, 2026)

$88.95

Target: $113.52 (+27.6%)

P/E Ratio

64.8

P/E Forward

20.3

Dividend

—

Market Cap

$34.3B

EPS

$1.36

Consensus

Buy

What they do

onsemi makes a living selling chips that manage electricity, not chips that «think». An NVIDIA processor computes; onsemi's chips make sure power gets where it needs to go, at the right voltage, without waste.

It has three businesses:

  • Power Solutions Group (~$737 million last quarter, almost half the total): power chips, including the silicon carbide (SiC) ones that turn an electric car's battery energy into motion. Also power for AI data centers.
  • Analog & Mixed-Signal (~$540 million): chips that connect the real world (temperature, current, signals) to the digital one. This is home to Treo, its new, higher-margin platform.
  • Intelligent Sensing (~$236 million): image sensors — the «cameras» a car uses to detect pedestrians, lanes, and other cars so it can brake or park on its own.

Its number one customer is the automotive industry (~$797 million last quarter, more than half of sales), followed by industrial, and increasingly, AI data centers. It's headquartered in Scottsdale, Arizona, and has been run by Hassane El-Khoury since December 2020.

Why we like it

The thesis is simple: the cycle already turned and there's a new engine that didn't exist before.

1) The recovery has started and it's measurable. After seven straight quarters of decline in its automotive business, the January-March 2026 quarter showed total sales of $1.51 billion (above its own guidance and ~4% more than a year earlier), automotive up 5% year over year, and guidance for the next quarter of $1.535-1.635 billion (~7% more, excluding businesses it deliberately exited). Lead times stretched from 23 to 26 weeks and some products are already on allocation — a sign customers are ordering with urgency again.

2) Margins are rising by choice, not just because of the cycle. With its «Fab Right» strategy it closed and sold old factories (in July 2026 it agreed to sell two more, saving about $35 million a year). Result: three straight quarters of gross margin improvement, up to 38.5%, with factories running at only 77% utilization — if demand keeps rising, each extra point of utilization drops almost straight to profit.

3) AI is the new engine. Its AI data center business grew more than 30% in a quarter and the company expects it to double in 2026. It's working with NVIDIA on the shift to 800 volts: the new AI racks use so much power that how they're fed has to be redesigned, and that's exactly what onsemi sells.

4) It's still the #2 player worldwide in silicon carbide, with its chips in ~55% of the new electric models shown at the 2026 Beijing Auto Show, and its Treo platform (target margins of 60-70%) multiplied revenue by 2.5 in one quarter.

5) The price still doesn't reflect the full recovery. At $88.95 it trades at ~20 times expected earnings, in line with a semiconductor mid-cycle turnaround, and the 25 analysts covering it average a $113.52 target (~28% upside). It pays no dividend: it returns cash by buying back shares ($346 million last quarter alone) — a pure growth bet.

Key Risk

Risk #1 is China: it's both onsemi's biggest market and its biggest threat. Much of the electric vehicle boom is happening in China, and the Chinese government has approved a wave of state-backed local silicon carbide factories. If Chinese manufacturers flood the market with cheap, «good enough» chips, onsemi's prices and margins suffer — and any geopolitical escalation could shut that market off overnight.

Second, the automotive cycle is only just recovering: annual sales ($6.0 billion in 2025) are still ~27% below the 2023 peak, and if electric vehicles sell more slowly than expected (for example, after the US cut EV incentives), the recovery stalls.

Third, the stock has already climbed ~143% in a year and trades near highs: expectations are high, and any soft quarter would get punished hard. The P/E on past earnings (65x) looks expensive because 2025 had huge accounting charges; what matters is the 20x on future earnings — reasonable, but not a bargain.

Fourth, the Synaptics purchase ($6.2 billion in stock, the largest in its history, closing mid-2027) dilutes shareholders and adds integration risk. And fifth, it's up against giants: STMicroelectronics leads silicon carbide, Infineon is chasing the same NVIDIA deal, and Sony is pushing hard in image sensors.

This is educational and informational content, not financial advice. Always consult a qualified financial advisor before making investment decisions.

Vectorial Data picked ON on Jul 16, 2026 at $88.95.

Full Research

What onsemi is

ON Semiconductor (brand name: onsemi) was founded in 1999 as a spinoff of Motorola and today is one of the world's major manufacturers of power semiconductors: the chips that move, convert, and manage electricity. While a processor «thinks», onsemi's chips make sure the energy from a battery, an outlet, or a solar panel reaches the motor, sensor, or server at exactly the right voltage and with minimal waste.

Headquarters: Scottsdale, Arizona. CEO: Hassane El-Khoury (since December 2020, previously CEO of Cypress Semiconductor), the architect of the current transformation: fewer low-margin commodity products, more silicon carbide, intelligent sensing, and power for AI. It employs about 22,700 people worldwide after several rounds of headcount cuts.

Its technological crown jewel is silicon carbide (SiC): a material that withstands much higher voltages and temperatures than traditional silicon and loses less energy when converting it. In an electric car that translates into more miles per charge or a smaller (cheaper) battery. onsemi is the #2 player worldwide in SiC (STMicroelectronics is #1 with ~33% market share) and is vertically integrated: it manufactures everything from the raw material crystal to the finished module. In January 2025 it also bought Qorvo's SiC JFET business to strengthen its high-performance lineup.

How it makes money

Three business units (figures from the January-March 2026 quarter, total $1.51 billion):

  • Power Solutions Group (PSG) — $737 million, +14% year over year. Discrete power chips and modules, including the entire SiC business (EliteSiC brand) and AI data center power solutions. This is the heart of the thesis.
  • Analog & Mixed-Signal Group (AMG) — $540 million, -5% year over year. Analog and mixed-signal chips for automotive and industrial uses. This is home to Treo, its new 65-nanometer platform that integrates power, sensing, and communication into a single chip (1 to 90 volts, up to 175°C): Treo revenue multiplied by 2.5 in a quarter, and the company is targeting gross margins of 60-70%, well above the corporate average.
  • Intelligent Sensing Group (ISG) — $236 million, +1% year over year. Image sensors, mostly for cars: the cameras behind automatic braking, pedestrian detection, and assisted parking (ADAS).

By end market, automotive is more than half of sales (~$797 million last quarter), followed by industrial; the fastest-growing segment is AI data centers.

It pays no dividend: its policy is to return 100% of free cash flow over time through share buybacks (last quarter it bought back $346 million at an average price of $60.54 — well below the current price, a sign of good timing).

The numbers: last quarter and the full year

Full year 2025 (the bottom of the cycle): sales of $5,995 million, -15.3% versus $7,082 million in 2024 and ~27% below the 2023 peak of $8.25 billion. Reported net income collapsed to $121 million (from $1,573 million) due to charges and depreciation tied to the restructuring. But note: even in the worst year of the cycle it generated $1.4 billion in free cash flow, a record 24% margin — the company bled on paper, not in cash.

First quarter 2026 (reported in May, the turning point):

  • Sales: $1.51 billion, above the midpoint of its own guidance and ~4% more than the $1,446 million a year earlier — the first year-over-year growth after the desert crossing.
  • Automotive: $797 million, +5% year over year, the first increase after 7 straight quarters of decline.
  • AI data centers: +30% in a single quarter, nearly double a year earlier; management expects the business to double in 2026.
  • Adjusted gross margin: 38.5%, the third straight quarter of improvement; operating margin 19.1%.
  • Adjusted earnings per share: $0.64. Free cash flow: $217 million.
  • Factory utilization: 77%. Lead times: 26 weeks (up from 23), with some products on allocation — customers went from canceling orders to fighting over them.
  • Second quarter 2026 guidance: sales of $1.535-1.635 billion (~7% sequential growth excluding divested businesses), gross margin 38-40%, adjusted EPS $0.65-0.77.

The thesis: why now

1. The cycle has turned. Automotive and industrial semiconductors run on inventory cycles: in 2023 customers stockpiled chips fearing shortages, in 2024-2025 they stopped buying to burn through that inventory, and onsemi paid the price. That excess has now been worked off: orders rising, lead times stretching, some product lines on allocation. Buying a quality cyclical stock right as it comes off the bottom is historically the best time to do it.

2. It will come out of the cycle a more profitable company. The «Fab Right» strategy means keeping only competitive factories: in 2024 it restructured ~1,500 positions, and on July 7, 2026 it signed agreements to sell two more factories, saving roughly $35 million a year starting in 2027. That's why gross margin has risen for three straight quarters even with factories running at only 77% utilization: once demand recovers, operating leverage is enormous.

3. AI gave it a second engine. A next-generation AI rack can draw up to 1 megawatt; feeding it with traditional electrical architecture wastes too much power. That's why NVIDIA is driving the shift to 800 volts of direct current, and onsemi is an official partner in that ecosystem (MGX): it sells the full power-conversion chain, from the substation to the processor. Its alliance with Flex Power already covers 30+ active programs, and in gallium nitride (GaN, another advanced material) its design pipeline exceeds $1.5 billion.

4. Dominant position where it matters. Its SiC chips were in ~55% of the new electric models shown at the 2026 Beijing Auto Show. Treo is targeting margins of 60-70%. And in automotive image sensors it's a leader.

5. It's also growing by acquisition. On June 25, 2026 it announced the acquisition of Synaptics for about $6.2 billion in stock (~$7 billion including debt), its largest purchase ever, to move into «physical AI»: on-device connectivity and computing. Expected synergies of $200 million a year, accretive to earnings within 18 months of closing (expected mid-2027). The CEO estimates it opens up an incremental market of ~$30 billion.

Valuation

We bought at $88.95 (market cap: $34.3 billion). The trailing P/E (64.8x) looks scary but is misleading: 2025 earnings were artificially depressed by restructuring accounting charges (EPS of just $1.36). The useful metric is the forward P/E of ~20x — reasonable for a cyclical stock coming off the bottom with an AI business set to double, and at a discount to other power/AI names. The 25 analysts covering it have a consensus buy rating with an average target of $113.52 (~28% upside). The company itself bought back $346 million of its own shares last quarter. Honest caveat: the stock is up ~143% over the past twelve months and trades near 52-week highs — the market has already caught the scent of the recovery; what you're buying here is the continuation, not the discovery.

Risks

1. China: giant customer and subsidized competitor

Much of the electric vehicle boom is happening in China (hence the 55% Beijing figure). But the Chinese state has approved a wave of local SiC capacity — 15 new epitaxy lines and state-backed players like AscenPower. If China does to SiC what it did to solar panels (flood the market and crush prices), margins in onsemi's star business suffer. And any escalation in US-China trade tensions could cut off access to that market.

2. The automotive cycle could relapse

One quarter of growth doesn't make a recovery. Annual sales are still ~27% below the peak, and EV adoption has disappointed before (seven quarters of decline prove it). The cut to US EV incentives is pressuring demand in that market; the thesis depends on China and Europe making up the difference.

3. High expectations after +143% in a year

Near 52-week highs, any soft guidance or delay in margin recovery could trigger sharp, fast drops. It's a volatile stock: in 2025 it lost more than half its value from its 2024 highs at one point.

4. Synaptics integration

Paying ~$6.2 billion in stock dilutes current shareholders, and big semiconductor acquisitions have a mixed track record. The deal doesn't close until mid-2027: a year of management distraction and regulatory risk.

5. Margin recovery is a promise, not a fact yet

The 38.5% gross margin is still far from the ~45% the company achieved in the good times of 2024. If factory utilization doesn't rise above 77%, or if the high-margin businesses (Treo, SiC, AI) grow more slowly than expected, margin expansion falls short and the 20x forward P/E stops looking cheap.

6. Competition on every front

STMicroelectronics leads global SiC and is building new factories in Italy and China; Infineon announced its entry into the same NVIDIA ecosystem just two days later; in image sensors, Sony is pushing hard. onsemi isn't competing alone on any front.

Verdict

onsemi is the classic cycle-turn bet, with an extra ingredient that didn't exist in the last cycle: AI's demand for electricity. The bottom is behind it (growth positive again, orders accelerating, margins rising for three straight quarters), the company has become structurally more efficient by selling off surplus factories, and its NVIDIA alliance puts it in the mix for one of the decade's biggest infrastructure spending waves. At ~20 times forward earnings, with analyst targets 28% higher, the risk-reward is attractive — as long as you accept the volatility of a cyclical stock that's already run hard, and whose biggest market (the Chinese EV industry) is also its biggest competitive threat. It fits the portfolio as the power piece of the semiconductor bucket: KLA and Applied Materials make the machines, Micron makes the memory, ARM does the designs — onsemi puts electricity in motion.

More picks in Technology

See all Technology picks →

More picks from United States

See all picks from United States →
Researched: Jul 16, 2026Updated: Jul 16, 2026Next review: Jan 16, 2027

This is not financial advice. Consult a certified financial advisor.

The author may hold positions in the securities discussed.

Past performance does not guarantee future results.