The Inflection Is Confirmed and AI Data Center Doubles — But After a 143% Run to 33x Earnings, We Harvest the Outperform Call and Downgrade to Hold

Results vs. Consensus

Q1 2026 Scorecard

Year-Over-Year Comparison

Quarter-Over-Quarter Comparison

Quality of Beat

Revenue: The $1,513M result (above the $1,490M guide midpoint, +5% year-over-year) is the cleanest top-line of our coverage, and the composition is what matters. It is the first year-over-year growth in seven quarters, it came "better than normal seasonality" in a seasonally down quarter, and it absorbed ~$50M of planned non-core exits. The Power Solutions Group grew 14% year-over-year, automotive grew 5% year-over-year (on content, not units, with China passenger-vehicle volumes down 6%), and AI data center grew 30% sequentially. The recovery is no longer a guide; it is in the printed number.

Margins: Gross margin of 38.5% (up 30bps sequentially in a seasonally down quarter) is the third consecutive quarter of expansion and confirms the FabRight self-help thesis. Utilization jumped to 77% from 68% as the company ramped quickly to meet stronger demand, a demand-driven move (not a die-bank build). Management reiterated the full bridge toward 50%+ (utilization at 25–30bps per point, ~200bps from FabRight, ~200bps from mix, ~200bps from the divested fabs starting in 2027) and, importantly, promised "larger step functions" in margin through the rest of 2026 than the modest 30bps seen in Q1.

EPS and cash: Non-GAAP EPS of $0.64 beat consensus by a nickel; the GAAP loss of ($0.08) reflects a $329M restructuring charge for the FabRight footprint actions and is not an operating concern. Free cash flow was $217M (14% of revenue) on capex of just 1.4% of revenue, and the company repurchased $346M of stock (~160% of free cash flow) at an average price of $60.54, well below where the stock now trades. The opportunistic, below-market buyback is a credit to capital discipline.

Segment Performance

End-Market Revenue — Q1 2026

Business-Unit Revenue — Q1 2026

AI Data Center — The Engine: +30% QoQ, Doubling in 2026, ~10x Rack Content

AI data center revenue grew more than 30% sequentially, nearly double the high-teens management expected entering the quarter, and onsemi now expects to double AI revenue year-over-year in 2026 (off the >$250M 2025 base). Growth came from breadth: across multiple XPU vendors and all leading hyperscalers, spanning the full PowerTree from high-voltage conversion to intelligent power stages, protection/control, and system integration. The FlexPower partnership alone now spans 30-plus active programs (intermediate bus converters, power supplies, battery backup, supercapacitors, 800-volt DC). Management quantified the content opportunity: a 120kW rack carries ~$9,500 of onsemi content today, while an 800-volt/HVDC rack carries ~$115,000, roughly 10x more inside the rack, with additional grid-to-rack content (including solid-state transformers) on top.

"Our revenue grew more than 30% quarter over quarter, nearly double our expected growth rate entering the quarter… we now expect our AI data center revenue to double year over year in 2026." — Hassane El-Khoury, President & CEO

Assessment: AI data center has crossed from "promising option" to "primary growth driver." A 30% sequential print against a high-teens expectation, a guide to double for the year, and a ~10x rack-content step at higher voltages give onsemi a genuine secular growth leg. As the only broad-based U.S. power-semiconductor supplier with wall-to-core breadth, it is positioned to compound share. This is the engine that justifies a growth multiple, and it is precisely why the stock has re-rated, which is the heart of the valuation question.

Automotive — Content Over Units, China Leading

Automotive grew 5% year-over-year (roughly flat sequentially), the first year-over-year growth after seven quarters of decline, with management now "shipping to natural demand." The growth is a content-and-share story rather than a unit-volume one: China passenger-vehicle volumes fell 6% in Q1, yet onsemi's China automotive revenue grew. SiC content reached ~55% of new EV models shown at the 2026 Beijing Auto Show, with expanded Geely and NIO collaborations on 900-volt platforms, and a new IGBT-based traction-inverter win with a North American OEM moving to direct semiconductor sourcing. Treo-based 10BASE-T1S Ethernet for zonal architecture entered production at a North American customer (more than 30 Treo devices per platform).

Assessment: Auto has fully transitioned from destocking headwind to content-driven growth, and the content angle is the durable one: zonal architecture, image sensors, and SiC/IGBT traction content grow regardless of flat global vehicle volumes. The ~55% Beijing Auto Show SiC share is a strong forward indicator for second-half production. Replenishment has not yet begun, leaving cyclical upside on top of the content growth.

Industrial — Broad Strength, AI Halo

Industrial was $417M, down 6% sequentially on Chinese New Year seasonality but ahead of expectations, with broad-based strength in traditional industrial for a second consecutive quarter. The AI "halo effect" is lifting adjacent infrastructure: energy storage systems are guided to grow more than 40% year-over-year in 2026 with market share approaching 60%, driven by SiC hybrid modules (the Sungrow partnership combining EliteSiC and FS7 IGBTs for utility-scale solar and storage).

Assessment: Industrial's two consecutive quarters of broad strength, layered with the AI-adjacent ESS growth (40%+ at ~60% share), make it a quieter beneficiary of the same AI-power theme. The segment's recovery plus the secular ESS tailwind give it a more durable growth profile than a pure cyclical read would suggest.

Key Topics & Management Commentary

Overall Management Tone: The most unequivocally confident posture of our coverage. Management opened by calling the quarter "a clear inflection point," declared the company has "moved beyond the cyclical trough on a path to recovery," and for the first time stated plainly that the second half will outgrow the first half ("not flattish… a good outlook"). The confidence is now backed by printed year-over-year growth, accelerating AI, expanding margins, and improving order signals (book-to-bill, lead times, short-lead-time orders all positive). This is a management team describing a recovery it can see in the numbers, not just the order book.

1. "Moved Beyond the Cyclical Trough on a Path to Recovery"

"This quarter marks a clear inflection point for ON Semiconductor Corporation. Improving demand signals, accelerating AI data center growth, and sustained gross margin expansion demonstrate that the structural changes we made over the past several years are now translating into tangible financial results." — Hassane El-Khoury, President & CEO

The recovery is now explicit and quantified: first year-over-year revenue growth in seven quarters, better-than-seasonal Q1, a Q2 guide implying ~7% sequential growth ex-exits (above seasonal), and a stated expectation that the second half outgrows the first. Management cited improving book-to-bill, lead times, and short-lead-time orders as the basis for confidence, and contrasted the outlook explicitly with the more muted second-half it experienced in 2025.

Assessment: This fully resolves the central bear point from our initiation. The down-cycle is over, the recovery is in the printed numbers and the forward guide, and the second-half-outgrows-first-half framing is the most forward-leaning growth statement management has made. From a fundamental standpoint, the thesis we built has played out. The only remaining question is whether the equity, after its run, still offers favorable risk/reward, which it largely does not.

2. AI Data Center: The Content Math Behind the Doubling

Beyond the 30% sequential growth and the guide to double in 2026, management laid out the content economics that underpin the multi-year opportunity. A traditional 120kW rack carries ~$9,500 of onsemi content; an 800-volt/HVDC rack carries ~$115,000, roughly 10x. As architectures move to higher-voltage distribution and "collapse the conversion tree," onsemi's breadth across silicon, SiC, SiC JFET, GaN/vGaN, and Vcore point-of-load maps onto more of the rack, and there is incremental content from the rack all the way to the grid (including solid-state transformers).

"At the 800-volt or high-voltage rack, we are thinking about roughly $115,000 of content. So although our content is almost 10x inside the rack, there is additional incremental content from the rack all the way to the grid that we also participate in." — Hassane El-Khoury, President & CEO

Assessment: The ~10x content step at higher voltages is the single most powerful element of the bull case and the clearest justification for the stock's re-rating. If 800-volt architectures proliferate and onsemi holds its wall-to-core breadth, AI revenue can compound for years off the doubling 2026 base. This is real, and it is also now substantially reflected in a ~33x forward multiple, which is the crux of our downgrade.

3. The Margin Bridge and "Larger Step Functions" Ahead

"We expect expansion throughout the remainder of this year, and probably larger step functions than what you saw here in the first quarter." — Thad Trent, CFO

Gross margin expanded a third consecutive quarter to 38.5%, and management guided to continued sequential expansion through 2026, with larger steps than Q1's 30bps as utilization rises (each point worth 25–30bps), FabRight benefits compound (~200bps), favorable mix builds (Treo at 60–70% gross margin; ~200bps long-term), and divested-fab benefits begin in 2027 (~200bps). Input-cost headwinds are being offset by surgical pricing actions that hit the P&L in the second half. Stacked up, management reiterated the math gets "over 50%."

Assessment: The margin story is now firing on every cylinder: utilization recovery, structural cost-out, and high-margin new-product mix all pulling the same direction, with pricing as a second-half tailwind. The path toward the mid-40s in 2026–2027 and 50%+ longer term is credible. This is a powerful earnings-growth engine, and it is the reason FY26–FY27 EPS estimates should keep rising, supporting the stock even at a full multiple.

4. Second Half Outgrows First Half

"We are expecting the second half to outgrow the first half. I am not talking about flattish; I am talking about a good outlook… That was not the sentiment that I personally had last year. In contrast, we see a much better outlook than we did last year." — Hassane El-Khoury, President & CEO

Management declined to guide full-year seasonality but stated clearly that the second half will outgrow the first, driven by program ramps already underway (the ~55% SiC-share China EV models just released and entering production in H2, plus AI data center and industrial). It explicitly contrasted this with the disappointing second half of 2025.

Assessment: A second-half-outgrows-first-half statement is the strongest growth signal of our coverage and de-risks the FY26 estimate trajectory. Combined with the Q2 guide (above seasonal ex-exits) and the doubling AI line, it points to FY26 EPS comfortably above the prior-year level. The earnings momentum is real; our concern is purely that the stock has already paid for it.

5. Capital Allocation: Opportunistic Buyback Below Market

onsemi repurchased $346M of stock in Q1 (~160% of free cash flow) at an average price of $60.54, well below the ~$102 level at which the stock now trades. The company has a $6B authorization and a 100%-of-free-cash-flow return policy, with capex held at just 1.4% of revenue in Q1 (mid-single-digit percent for the foreseeable future, including 2027). Cash and short-term investments were ~$2.44B with $3.9B of total liquidity.

Assessment: Buying back ~160% of free cash flow at an average of $60.54, against a stock now near $102, is excellent capital discipline that flatters per-share metrics. It also implicitly frames management's own valuation reference point in the low-$60s, which is notable context for the equity's current ~$102 level. The buyback supports a floor, but at the current price it is far less accretive than the Q1 repurchases were.

6. Portfolio Exits: Done in 2026, None Beyond

onsemi exited ~$50M of non-core revenue in Q1 and guided another $30–40M in Q2, completing the planned ~$300M of 2026 exits. Management confirmed there will be no further exits beyond 2026, and that the long-standing growth algorithm (market growth minus ~5% for exits) applies only through 2026, after which the core resumes above-market growth.

Assessment: The exit program reaching completion in 2026 removes a multi-year overhang. From 2027, the reported revenue line should grow at or above market without the self-inflicted drag, which is a cleaner growth story. This supports a higher-quality 2027 revenue base, reinforcing the structural improvement, though again it is a development the market appears to have anticipated.

7. Utilization, Capacity, and the Capex-Light Recovery

Utilization rose to 77% (from 68%) as onsemi ramped quickly to meet demand. Management noted that "fully utilized" is just over 90% (low-90s) and would require revenue 25–30% above current levels, and that capital intensity stays at mid-single-digit percent of revenue through 2027 because the capacity for the ramping new products (Treo at East Fishkill, SiC, JFET) is already in place. The company has "a lot of capacity" and does not need to add.

Assessment: A capex-light recovery is the ideal financial setup: revenue can grow 25–30% before onsemi approaches full utilization, and that growth drops through to gross margin (25–30bps per utilization point) and free cash flow with minimal incremental capital. This is the operating-leverage payoff of the prior years' investment, and it materially raises the through-cycle free-cash-flow profile. It is a genuine positive and, again, one the market has largely priced.

Guidance & Outlook

The Q2 2026 guide of $1.535–1.635B (midpoint $1.585B) is a 5% sequential increase, and excluding the incremental $30–40M of non-core exits, revenue grows ~7% at the midpoint, well above normal seasonality. By end-market, management expects automotive roughly flat (still no replenishment cycle), industrial up mid-single-digits on broad strength, and "Other" (AI data center) up mid-teens sequentially. Gross margin is guided to a 39% midpoint (+50bps), with EPS at a $0.71 midpoint (+11% sequential).

The trajectory: Q2's above-seasonal guide, the stated second-half-outgrows-first-half framing, the doubling AI line, and "larger step functions" in margin point to FY26 non-GAAP EPS in the low-$3.00s and rising, with FY27 potentially in the low-to-mid-$4.00s as margins march toward the mid-40s and AI scales further. The earnings trajectory is firmly upward; the rating question is whether the current ~$102 share price (and the multiple it implies) already discounts that trajectory, which we believe it largely does.

Restocking as remaining upside: Management again noted it has not seen the automotive replenishment cycle ("if that were to happen later in the year, that would be a good thing"). That coiled spring is genuine upside to the guide, but it is also the kind of upside the market is now inclined to price in advance after a 90%-YTD run.

Analyst Q&A Highlights

When the Secular Drivers Become Visible, and the Margin Stair-Steps

The opening exchange pressed on when the secular drivers (AI, grid, zonal) become the dominant visible revenue driver, and whether margin would move in larger steps rather than a slow linear ramp. Management argued the secular drivers are already showing up, and the CFO promised larger margin steps ahead.

Q: "How do you think those are really going to show through to investors, and when do those become the dominant driver of revenue…? … People expect a little bit more stair-steps than kind of a slow linear ramp on the gross margin side. Is that something we should expect in the second half[?]"
— Ross Seymore, Deutsche Bank

A: "If you think about the AI data center, you are already seeing it in 2026… doubling our revenue from last year… [CFO Trent:] We expect expansion throughout the remainder of this year… and probably larger step functions than what you saw here in the first quarter."
— Hassane El-Khoury, CEO / Thad Trent, CFO

Assessment: Management's answer is bullish and credible: the secular drivers are already in the numbers (AI doubling, Treo +2.5x), and the margin path accelerates with larger steps through 2026. For a buyer at the trough this was the dream scenario. For a buyer at ~$102, it confirms the earnings momentum that the multiple has already capitalized, which is the tension at the center of our downgrade.

How the Second Half Plays Out Versus a Soft 2025

A question probed whether the second half of 2026 would fade as 2025's did, or sustain. Management gave its most forward-leaning growth statement of our coverage.

Q: "Last year we saw the analog industry had a decent first half and then things started to get a little more muted in the second half. Do you think this year's second half plays out differently…? Should we be expecting seasonal or better-than-seasonal trends in the second half?"
— Vivek Arya, Bank of America

A: "We are expecting the second half to outgrow the first half. I am not talking about flattish; I am talking about a good outlook… That was not the sentiment that I personally had last year. In contrast, we see a much better outlook than we did last year."
— Hassane El-Khoury, President & CEO

Assessment: An unambiguous second-half-outgrows-first-half statement, explicitly contrasted with the soft 2025, materially de-risks the FY26 trajectory. It is exactly the confirmation the cyclical thesis needed. It also means the consensus FY26 estimates the stock is being valued against are well-supported, leaving little room for the kind of upside surprise that would justify chasing the stock here.

How Big Is AI, and Can onsemi Scale Without More Acquisitions?

A question asked for AI as a percent of sales and whether onsemi has the scale and resources to be a major player organically, or needs more M&A. Management framed AI off the doubling 2025 base and pointed to the 2025 acquisitions (Aura/Vcore, the SiC JFET asset) as having already built the platform.

Q: "How big is it for ON Semiconductor Corporation right now? Is it something like mid-single-digit percent of sales?… do you think you have the scale and the internal resources… or do you think you will need some inorganic resources[?]"
— Vivek Arya, Bank of America

A: "Last year, we talked about $250 million in AI revenue. We just mentioned that we will be doubling that this year… We have done some inorganic acquisitions in 2025 that are playing to our advantage… We absolutely have the focus that we need to be a major player in power for AI data center."
— Hassane El-Khoury, President & CEO

Assessment: Doubling AI revenue to ~$500M-plus in 2026 (roughly 8% of sales) makes it the company's most important growth driver, and management believes the technology platform is already in place organically (with the 2025 tuck-ins filling gaps). This is the disclosure that most underpins the AI-growth narrative, and by extension the premium multiple the stock now carries.

The Content Economics of the High-Voltage Rack

A question on where the biggest incremental AI content opportunity sits (rack versus point-of-load versus outside the data center) drew specific content-per-rack figures.

Q: "As architectures move towards higher-voltage distribution, how should we think about the biggest incremental content opportunities…? Is the larger dollar opportunity still outside and at the rack, or is the VCORE point-of-load side becoming a more material contributor?"
— Analyst, Needham & Company (on behalf of N. Quinn Bolton)

A: "At the rack… a 120-kilowatt rack at roughly $9,500 of content. At the 800-volt or high-voltage rack, we are thinking about roughly $115,000 of content. So although our content is almost 10x inside the rack, there is additional incremental content from the rack all the way to the grid… which is exactly in our sweet spot."
— Hassane El-Khoury, President & CEO

Assessment: The ~10x content step ($9,500 to ~$115,000) at higher-voltage racks is the most quantitatively striking disclosure on the call and the analytical core of the AI bull case. It frames a multi-year content-growth runway that, if onsemi holds its wall-to-core breadth, supports compounding AI revenue. It is also the figure most responsible for the stock's re-rating, and thus the one the market is now paying a premium to own.

Automotive: Unit Demand Normalization Versus Content Growth

A question asked how much of the automotive improvement is true unit-demand normalization versus content growth, and whether regional dynamics differ. Management was explicit that it is a content-and-share story.

Q: "As automotive begins to recover, how much of the improvement… is true unit demand normalization versus content growth from things like image sensors and zonal architectures? And beyond China, are you seeing any meaningful differences by region?"
— Analyst, Needham & Company (on behalf of N. Quinn Bolton)

A: "In China specifically, Q1 is seasonally down. The number of passenger vehicles was down 6%. Our revenue was actually up. Therefore, that tells you it is a content story… We are more leveraged to content than SAAR, and in certain areas, we are gaining share."
— Hassane El-Khoury, President & CEO

Assessment: The China data point (passenger-vehicle volumes down 6%, onsemi China auto revenue up) is the cleanest evidence that auto growth is content-and-share-driven rather than unit-dependent, which makes it more durable through a flat global SAAR. The zonal 10BASE-T1S win is net-new content that did not exist a year ago. This is a high-quality growth driver, and like the others, it is now embedded in the stock.

Are the Portfolio Exits Still on Track, and Any Slippage Risk?

A question revisited the ~5%-exit growth algorithm and whether the exits might slip again as they did years ago. The CFO confirmed completion in 2026 with none beyond.

Q: "Is that still the right way to think about it?… A few years ago… it took you longer than anticipated… Any risk of that happening again this year[?]"
— Joshua Buchalter, TD Cowen

A: "No change… we have exited approximately $50 million in Q1. There is another $30 million to $40 million here in Q2… roughly $300 million… So we are done in 2026… I do not plan on that changing for the rest of this year."
— Thad Trent, CFO

Assessment: Completing the exits in 2026 with none beyond removes the multi-year base-shrinking overhang and sets up a cleaner 2027 in which reported revenue grows at or above market. It is a meaningful structural positive that improves the quality of the forward growth, and it reinforces the view that 2027 is the year the "above-market growth, net of exits" framing fully kicks in.

Utilization Headroom and a Capex-Light Recovery

A question on whether capital intensity needs to rise to support 2027–2028 growth drew a clear "no": onsemi has the capacity in place and can grow substantially before approaching full utilization.

Q: "At what point do you need to take up your capital intensity above the 5% level… to support the growth in 2027 and 2028?"
— Gary Mobley, Loop Capital

A: "I do not anticipate any change to our capital intensity… mid-single-digit percentage of revenue for the foreseeable future… To get to fully utilized… just over 90%… we would need revenue that is 25% to 30% higher than where we are today… we have a lot of capacity here."
— Thad Trent, CFO

Assessment: The capex-light recovery is the ideal financial backdrop: 25–30% of revenue headroom to full utilization, with that growth dropping through to margin and free cash flow at mid-single-digit capital intensity. It substantially raises the through-cycle free-cash-flow profile and underwrites the buyback. It is a genuine structural positive, and, consistent with the theme of this quarter, it is one the market has already rewarded.

What They're NOT Saying

1. No full-year 2026 revenue guide: Despite the "second half outgrows first half" framing, management would not put a number on full-year revenue or AI as a precise percent of sales, leaving the magnitude of the recovery to be triangulated from quarterly guides.
2. AI composition undisclosed: Management declined to break the AI doubling into GaN versus SiC versus Treo contribution ("not breaking it down to that level by product family"), keeping the margin mix within the AI growth opaque.
3. Restocking still undated: Automotive replenishment "has not happened yet"; management frames it as upside "if that were to happen later in the year" but will not date it, so the biggest remaining catalyst is unscheduled.
4. No comment on the share price: Management did not address the ~90% year-to-date run, but the $346M Q1 buyback at an average of $60.54 (versus the ~$102 current price) implicitly frames the company's own valuation reference far below the market price.
5. Pricing benefit deferred and vague: Surgical pricing actions to offset input costs are described as a second-half P&L event, but management gave little detail on magnitude or breadth.

Market Reaction

• Pre-print setup (the dominant fact): ON closed at $102.04 on May 4, having risen ~66% over the prior 30 days and ~88% year-to-date (and ~143% over the trailing twelve months), at an all-time high and the top of its 52-week range ($38.41–$103.03). The print followed one of the sharpest melt-ups in the large-cap semiconductor space.
• Reaction session (May 5): ON gapped down 4.3% to $97.65, traded a wide $96.08–$105.81 range (−5.8% to +3.7%), and closed at $102.67, up just 0.6% (+$0.63), on volume of 22.7M shares versus a 9.6M-share 30-day average (2.4x normal).
• Market context: The S&P 500 rose 0.8% on the session, so onsemi roughly matched the market despite a clear beat-and-accelerate quarter.

A +0.6% close on a quarter that beat on revenue and EPS, accelerated AI to +30% sequential, raised the AI full-year outlook to a double, expanded margins, and guided Q2 above seasonal is the single most important signal in this report. When a great quarter at an all-time high cannot move the stock, the market is telling you the good news is already priced.

Exhaustion at the highs: The intraday pattern (gap down 4.3%, recover, close roughly flat) is classic late-stage price action: aggressive sellers meet committed buyers, and the result is stasis at an elevated level. After a 66% 30-day run, the bar for a positive surprise had become nearly unclearable, and even an excellent quarter could not clear it.

The buyback reference point: Management repurchased stock in Q1 at an average of $60.54, roughly 40% below the current price. That is not a forward signal, but it is a useful anchor: the company was an aggressive buyer in the low-$60s and is now watching its stock trade near $102. The reaction and the buyback context together inform our valuation-driven downgrade.

Street Perspective

Debate: Has the Stock Run Too Far, or Is This the Start of a Growth Re-Rating?

Bull view: onsemi has transformed into an AI-power growth company. AI revenue is doubling, margins are expanding with "larger step functions" ahead, the second half outgrows the first, restocking is still to come, and capex is light. At ~33x FY26 and ~23x FY27, that is not expensive for a company compounding earnings 20%+ with a secular AI tailwind; the re-rating from cyclical to growth multiple is justified and has further to run as AI scales.

Bear view: The stock is up ~143% in twelve months and ~66% in a single month, to an all-time high at ~33x forward earnings, a multiple it has never sustained. AI is ~8% of sales; the other ~92% is a cyclical recovery that, while real, is well-understood. A muted reaction to a great quarter signals exhaustion. Any sector rotation, a stalled restocking cycle, or a single margin disappointment could trigger a sharp reversion from these levels.

Our take: The bull case on the business is correct; the bear case on the stock is more compelling at this price. The re-rating from a cyclical-trough multiple to a growth multiple has largely happened, capturing the inflection we identified at Q4. From ~$102, further gains require both continued earnings momentum (likely) and a held-or-expanded premium multiple (uncertain after a parabolic run), while the downside from a melt-up unwind is material. That is a balanced-to-unfavorable 12-month setup, which is the definition of a Hold.

Debate: How Durable Is the AI Data Center Growth?

Bull view: The ~10x content step at 800-volt racks, the wall-to-core breadth (the only broad-based U.S. power supplier), 30-plus FlexPower programs across all hyperscalers and multiple XPU vendors, and the doubling 2026 guide make this a multi-year compounder. As 800-volt architectures proliferate, onsemi's content per rack and share both rise.

Bear view: AI power is a competitive field with well-capitalized incumbents, the ramp is gated by the XPU vendors' 800-volt deployment timelines, and design-in breadth is not the same as sustained sole-source dollar share. A doubling off a small base is impressive but the absolute dollars (~$500M) are still modest relative to the ~$6B company, and the growth rate will inevitably decelerate.

Our take: Durable and real, but already priced. We are confident AI compounds for onsemi over multiple years given the content math and breadth. The issue is not whether AI grows; it is that the market has extrapolated that growth into a premium multiple on the whole company, leaving little margin of safety if the ramp is merely good rather than spectacular.

Debate: Is the Margin Path to the Mid-40s (and 50%+) Achievable?

Bull view: Three consecutive quarters of expansion, utilization at 77% with 25–30% revenue headroom to full, ~200bps each from FabRight, mix, and divested fabs (2027), plus Treo at 60–70% gross margin and a capex-light model. "Larger step functions" are coming. The mid-40s in 2026–2027 and 50%+ longer term are credible, driving strong EPS growth.

Bear view: Q1's expansion was a modest 30bps, the bulk of the path still leans on a utilization recovery, input-cost headwinds are real, and the 50%+ target has been a multi-year "someday" number. Mid-40s may take longer than bulls expect if the recovery is gradual.

Our take: Bull on the destination, and this is the strongest part of the story. The margin engine is real and compounding, and it is the primary reason FY26–FY27 EPS estimates should keep rising. That rising-EPS trajectory is what makes us comfortable with a Hold rather than a more cautious rating despite the rich multiple: even if the stock does not outperform, the earnings should grow into the valuation over time.

Model Implications & Valuation Framework

Valuation framework: At the post-reaction price of ~$102.67 and our FY26 non-GAAP EPS estimate of ~$3.00–3.30, ON trades at roughly 31–34x forward earnings, and ~21–25x our preliminary FY27 estimate of ~$4.10–4.80. This is a full growth multiple for a company that, AI aside, is still substantially a cyclical recovery, and it sits well above the ~22x at which we upgraded just one quarter ago and the low-$60s at which management itself was buying back stock in Q1. The business quality has improved and the AI leg justifies a premium to the historical multiple, but ~33x FY26 prices in continued flawless execution.

Price-target framework (12-month):

• Base case ~$100–108: ~22–24x our preliminary ~$4.50 FY27 EPS (or ~32x ~$3.15 FY26). Recovery and AI proceed as guided, margins reach the low-40s, the multiple holds. Implies roughly flat to ~5% from the current price.
• Bull case ~$120–132: ~27–29x a ~$4.60 FY27 EPS if AI accelerates beyond the doubling, restocking begins, margins reach the mid-40s, and the growth multiple expands further. Implies ~17–29% upside.
• Bear case ~$72–80: ~22–25x a ~$3.15 FY26 EPS if the melt-up unwinds, the multiple reverts toward the mid-20s on FY26, or a margin/restocking disappointment hits a richly valued stock. Implies ~22–30% downside.

The base case implies roughly flat-to-modest upside from $102.67; the bear-case downside (~22–30%) is wider than the base-case upside, an unfavorable skew after a parabolic run. That asymmetry, against a business we genuinely admire, is the quantitative basis for stepping aside from the Outperform call and moving to Hold.

Thesis Scorecard Post-Earnings

Grading the standing thesis (carried from the Q4 Outperform). The unusual feature this quarter: nearly every operational pillar confirmed or strengthened, yet the rating moves down, because the share price has outrun the (excellent) fundamentals.

Overall: The operating thesis is fully validated. All three original bear points are resolved, and all three bull pillars strengthened. But a new and now-dominant risk has emerged: valuation. The stock has re-rated from a cyclical-trough multiple to a full AI-growth multiple, capturing the inflection we upgraded for, and the risk/reward is now balanced-to-unfavorable.

Action: Downgrade to Hold. Harvest the Outperform call. Re-engage on a pullback or as FY27 estimates grow into the multiple.

Bottom Line: A Great Business, a Full Price — Taking the Win

Rating decision: We are downgrading onsemi to Hold from Outperform. To be clear about what this is and is not: it is not a negative call on the business, which delivered its best quarter in years and validated every element of our thesis (the cycle has turned, margins are expanding on self-help, AI is doubling, the operating model has powerful leverage). It is a discipline on risk/reward. We upgraded to Outperform at the Q4 print at ~$67 because the inflection was arriving and unpriced; the stock has since returned roughly 52% to ~$102, the thesis has played out faster and further than we modeled, and at ~33x FY26 earnings after a 143% twelve-month run, the easy money has been made.

The rating framework we apply is explicit that a strong quarter can still be a Hold when valuation is full, and that is exactly the situation. From ~$102, the path to outperformance requires both continued earnings momentum (which we expect) and a held-or-expanded premium multiple (which is far less certain after a parabolic melt-up to an all-time high). The market's muted +0.6% response to a clear beat-and-accelerate quarter is the tell: the good news is priced. The downside from a multiple reversion or a melt-up unwind (~22–30%) now exceeds the base-case upside, an unfavorable skew. We would rather book the gain from the Outperform call and wait for a better entry than chase the stock here.

What would move us back to Outperform: A meaningful pullback (toward the low-$80s or below) that restores the risk/reward; the automotive restocking cycle beginning and driving estimates materially higher; AI data center scaling well beyond the doubling guide; or FY27 earnings growing into the multiple such that the forward P/E compresses to the low-20s at a similar price.

What would move us to Underperform: Evidence the recovery is stalling (a Q3/Q4 guide-down, or the second half failing to outgrow the first); margin expansion stalling well short of the promised "larger step functions"; or a sharp deterioration in the AI narrative (program losses, share erosion) against a still-premium multiple.

Signposts for Q2 2026 earnings (early August 2026):