Overview: A Stock Marked Down Before the Print

onsemi reports third-quarter results Monday after the close, and the setup is unusual: the shares have already been marked down as though a miss were coming. Ainvest data puts the stock down about 15% over the past month and roughly a third over the past four months — close to half below its 52-week high — even as the company’s fastest-growing business accelerates. For a chipmaker that has beaten earnings estimates in all four of its last quarters, that divergence between the trend in the business and the trend in the stock is the central question heading into the release.

The answer will not come from whether onsemi clears the bar. It will come from what the margin line says about the quality of the growth.

The Power Tree: From the Wall to the Core

The reason onsemi carries an AI label at all is power delivery. An AI rack draws electricity like a small town, and converting grid AC down to the low-voltage DC a processor needs wastes energy at every stage of the chain. Silicon carbide switches and power modules reduce that loss, and onsemi sells across the path — in its own phrasing, “from the wall to the core” of the data-center power tree.

The mechanism matters more than the marketing. As AI compute scales, the power semiconductor content behind each dollar of compute rises. Every additional conversion stage, every increase in rack density, and every tightening of efficiency budgets expands the silicon content attached to a given deployment. That is the structural argument underpinning the growth the Street is modeling: a near-40% year-over-year jump in earnings at consensus, with revenue back to roughly 10% above a year ago.

The Numbers That Define the Quarter

onsemi guided non-GAAP gross margin to 40%–42% for the quarter, a step up from the 38.4% reported in June, and set revenue at $1.65 billion to $1.75 billion with non-GAAP earnings of 81 to 93 cents a share. Consensus sits at $0.88 of non-GAAP EPS on about $1.71 billion in revenue — squarely inside the guided ranges.

Management has said AI data-center revenue should more than double this year, and the segment carrying it is guided to high-teens sequential growth. Those are the inputs. The output that decides the narrative is the gross margin print and, more importantly, the fourth-quarter guide that follows it.

The Margin Line: Operating Leverage or Allocation Optics

After 2025 — when revenue ran around $6 billion and the company was shipping into a destocking automotive and industrial market — gross margin moving back toward 40% is precisely what separates a genuine re-rate from a headline beat on thin economics.

The complication is how that margin is being produced. Semiconductor lead times have stretched from roughly 27 weeks to around 32, and onsemi has shipped AI orders ahead of automotive and industrial customers. That is simultaneously a demand signal and the specific dynamic to watch unwind. If the beat is powered by reshuffling scarce supply toward the highest-margin destination rather than by customers converting real designs into volume, the margin line will eventually reveal it. Allocation-driven margin is real cash in the short term and fragile in the medium term; design-win-driven margin compounds.

Market Context: A Growth Multiple on a Cyclical Book

The market has already priced much of the pivot. Ainvest data places onsemi near a $28 billion market capitalization on roughly 44 times trailing earnings and about 4.5 times sales — a growth-era multiple applied to a company whose traditional book remains automotive and industrial power chips, the cyclical demand that just went through a hard downturn.

That math only works if the AI power growth is durable and margins hold. Two disclosures therefore carry outsized weight: whether Q3 gross margin actually reaches the 40% threshold, and whether Q4 guidance sustains the trajectory. The first validates the quarter; the second validates the story.

The M&A Pivot: Synaptics, Ferric, and the Power-Delivery Race

Management is not waiting for clarity — it is spending the balance sheet on the pivot. Last month onsemi revised its deal to buy Synaptics to about $5.7 billion in cash at $123 a share, down from a roughly $7 billion all-stock structure, and, per the company, immediately accretive to non-GAAP earnings.

It is also reportedly in talks to buy Ferric for close to $1 billion — a startup whose regulators sit directly underneath AI processors to cut power loss. That is the same race Analog Devices entered by agreeing to buy Empower for $1.5 billion, and that Infineon joined with C2i. The common bet: value in AI is migrating from the compute chip to the power-delivery path itself.

Implications for AI Data Centers

For data-center architects, the strategic read is that power conversion is becoming an architectural discipline rather than a commodity procurement line. Silicon carbide switching, integrated power modules, and increasingly regulators placed immediately beneath the processor all attack the same problem — wasted energy at each voltage transition. As rack-level power density climbs, the penalty for conversion inefficiency compounds across the tree.

That favors vendors with coverage across the chain rather than point-solution suppliers, and it explains why consolidation is running through this segment specifically. It also means lead-time stress in power semiconductors is not a transient logistics artifact; it is a signal about where the physical constraint in AI buildouts is migrating.

Future Outlook

Three things to track after the print. First, the margin: a 40%-plus non-GAAP gross margin that holds into the fourth-quarter guide would support the operating-leverage interpretation. Second, the composition of growth: evidence of design-win conversion, rather than only priority allocation of constrained supply, would extend the durability of the AI narrative. Third, the deal cadence: closing Synaptics on the revised cash terms and advancing the reported Ferric discussions would confirm that onsemi intends to own more of the power tree rather than a slice of it.

The stock has already discounted disappointment. What it has not yet been given is proof that the margin expansion is structural — and that is the disclosure the quarter hinges on.