Allegro MicroSystems, Inc. (NASDAQ: ALGM) — A 71%-Automotive Sensor Maker Repriced as an AI-Data-Center Compounder

Independent equity research note. Report date: 2026-07-24. Price reference: $46.03 (2026-07-24 close) · Shares out 185.4M · Market cap ~$8.53B · Enterprise value ~$8.65B 52-week range: $22.80 – $69.62 · All-time high $69.62 (2026-06-30) · All-time low $16.53 (2025-04-21) Fiscal year ends late March. FY2026 = year ended 2026-03-27.

The analysis below discusses valuation only as embedded expectations and scenarios. It contains no buy/sell recommendation and no price target anywhere in the main body. The single, deliberate exception is the Author’s Take block immediately below, which is fenced off as a subjective view.

⚡ Author’s Take

This block is the author’s own subjective opinion. It is general information, not investment advice, and not a recommendation to buy or sell any security. Everything that follows it is position-free analysis carrying no price target — it discusses valuation only as embedded expectations. Do your own research.

Verdict: AVOID-here / HOLD-if-owned. A genuinely good niche franchise whose stock has been re-classified into an asset class it does not belong to. Not a short — the business is inflecting and the balance sheet is fine. Constructive accumulation zone ~$26–34 (~5–6x EV/Sales on FY27E, ~28–34x normalized non-GAAP EPS); fair ~$35–40; demanding above ~$45. Conviction: medium-high.

Allegro is the real thing in one narrow domain: magnetic sensing. Design-in captivity is genuine — automotive sockets are ASIL-qualified, ship for a platform’s 15-year life, and repricing runs at only low-single-digit annual ASP declines. The cycle has clearly turned: five consecutive quarters of sequential growth, FY26 revenue +23% to $890M, gross margin recovering 140bp, record free cash flow of $125M, net leverage under 1.0x, and a data-center business that more than quadrupled to 10% of sales. None of that is in dispute, and it is why this is not a short.

The problem is what you are being asked to pay for it. At $46 the enterprise is valued at 9.7x trailing sales and 51x adjusted EBITDA — on a company that lost money on a GAAP basis in each of the last two fiscal years and whose gross margin is still 850bp below its own FY2024 peak. Strip the promotion and the arithmetic is brutal: to justify today’s EV at a fair-for-analog 20x EV/EBIT, Allegro must earn ~$432M of operating income, which at management’s own >55%-gross-margin target model requires ~$1.6B of revenue — 1.8x today’s, roughly four years away at the Analyst Day’s mid-teens growth rate. You are paying the FY2030 target price in FY2026, undiscounted, with no haircut for cyclicality or execution. The comparison that settles it: Melexis, the closest pure-play magnetic-sensor competitor with near-identical revenue ($826M), trades at 3.0x EV/Sales while earning a 16.0% operating margin against Allegro’s 2.8% GAAP. Allegro’s premium is not a quality premium. It is an AI-narrative premium attached to a fan-driver business that management concedes ships below fleet-average gross margin.

The framing is a momentum unwind on a still-extended base, not value and not a falling knife — and the factor data says so precisely. The stock ran +321% off its April-2025 low to $69.62 on June 30, then fell 33.9% in under four weeks to $46.03, breaking its 21- and 50-day averages; it now carries a 2.29 beta, a −1.12 low-volatility loading, and — most tellingly — its nearest factor neighbours are semicap names (MKSI, COHU, ENTG, NVMI), not its actual peers ADI, NXPI and MCHP. The market re-bucketed this stock. Yet even after a third of the value came off, it sits in the 97th percentile of its own price-to-sales history. The de-rating has removed froth, not the premium. Two more facts sharpen the point: onsemi offered $35.10/share in cash and walked away in April 2025 — a strategic buyer with full diligence would not pay today’s price — and across all 230 Form 4s filed since 2021, not one insider has bought a single share on the open market, including at $16.53, while five of them sold into this year’s rally between $41.56 and $52.72. Single bullish trigger: gross margin crossing ~53% while data center compounds >25% and auto returns to its prior peak — which would put $1.3B+ of revenue and >$1.20 of clean EPS in view and make the multiple defensible. Single bearish trigger: the data-center growth rate decelerating toward the low teens as fan drivers commoditize, or a 2027 auto air-pocket — either would expose a 97th-percentile multiple on under-earning. Tag: a good sensor company wearing someone else’s multiple.

📈 Stock Price Action — Five-Year Event Map

Allegro IPO’d at $14 in October 2020 and spent five years going almost nowhere in a straight line — then everywhere at once. From a $16.53 all-time low on 2025-04-21 the stock ran +321% to an all-time high of $69.62 on 2026-06-30, and has since given back 33.9% to $46.03 in under four weeks. The three-year annualized return is −1.4%; the six-month return is +48%. Today’s price sits 33.9% below the June high, 178% above the April-2025 low, and 31% above the $35.10 cash bid that onsemi withdrew in 2025.

Cycle narrative. (1) The IPO priced into the post-COVID shortage; content-per-vehicle and scarcity pricing carried the stock to $36 by September 2021. (2) The 2022 rate shock de-rated the whole high-multiple analog complex; nothing broke at Allegro. (3) FY2023–24 were the earnings peak — 56.1% gross margin, 20.9% operating margin, 21.1% ROIC — and the stock followed to $52.72 in July 2023. (4) Then the auto/industrial inventory correction hit: revenue fell from $1,049M (FY24) to $725M (FY25), gross margin collapsed 1,040bp, and the stock lost nearly two-thirds. The July-2024 Sanken exit — 38.8M shares repurchased, funded with incremental term debt — landed almost exactly at the cyclical top. (5) onsemi’s $35.10 all-cash proposal was withdrawn on 2025-04-14 after Allegro’s board declined to engage; the stock made its all-time low of $16.53 one week later, on 2025-04-21. (6) The recovery that followed was extraordinary and event-dense: sequential growth every quarter of FY2026, data center rising from ~2% to 14% of sales, a February-2026 Analyst Day laying out mid-teens growth and >55% gross margins, and a string of +10% single-day moves (2026-01-06 +10.8%, 2026-01-29 +9.9%, 2026-04-30 +10.1%, 2026-05-26 +10.5%, 2026-06-29 +14.6%). (7) The June-30 high of $69.62 marked the top; the unwind since has been equally violent (−9.2% on 7/1, −12.2% on 7/2, +11.5% on 7/9, −8.1% on 7/24), with Barclays’ 2026-07-20 valuation downgrade a proximate catalyst. Price moves are Fact; attributed causes are Interpretation.

1. Executive Summary

Allegro MicroSystems designs magnetic sensor ICs — position, speed and current sensors — and application-specific analog power ICs, principally motor drivers, regulators and isolated gate drivers. It sells >1,500 products, overwhelmingly into automotive (71% of FY2026 sales) and, increasingly, industrial end markets (29%), of which AI data center is now the fastest-growing component at 10% of the fiscal year and 14% of the fourth quarter. The company is fabless for wafer fabrication and owns its primary back-end assembly and test facility in the Philippines. It employs 4,250 people, is headquartered in Manchester, New Hampshire, and has been led by CEO Michael Doogue with CFO Derek D’Antilio.

The business is genuinely recovering. FY2026 revenue rose 22.8% to $890.1M after a 30.9% collapse in FY2025, with five consecutive quarters of sequential growth and a fourth quarter of $243.2M (+26% YoY). Non-GAAP gross margin improved 140bp to 49.4% and non-GAAP operating margin reached 14.1%. Free cash flow was a record $124.9M. The balance sheet was repaired: $60M of voluntary debt paydown, term loan reduced to $285M and repriced twice to SOFR + 175bp, net debt ~$116M and net leverage below 1.0x. Design wins rose more than 30% and backlog exited the year at a multi-year high.

The reported economics are considerably weaker than the promoted ones. On a GAAP basis FY2026 gross margin was 46.3%, operating margin was 2.8%, and the company lost $0.08 per share. The gap to the promoted $0.54 of non-GAAP EPS is 1,130bp of operating margin, driven principally by $47.9M of stock-based compensation — 5.4% of sales, growing every year, and unambiguously a real economic cost — plus intangible amortization, and flattered further by a 6–9% tax rate. Gross margin remains ~850bp below the FY2024 peak. Over the eight fiscal years FY2019–FY2026, Allegro earned a GAAP operating margin above 20% in only two, and lost money on a GAAP basis in two.

The industry position is real but narrow. The moat is customer captivity through automotive design-in: AEC-Q/ASIL qualification, 15-year platform lives, and re-qualification costs that make mid-platform substitution impractical. That captivity is visible in pricing — ASP declines run low-single-digit annually, and management characterised CY2026 negotiations as “very low single digit,” better than normal. But Allegro is a component supplier into a socket, not a platform owner; it is fabless, so it holds no process-cost advantage; and its moat did not prevent a 31% revenue collapse or a 1,040bp gross-margin collapse in a single year.

The valuation is the entire issue. At $46.03 the enterprise is capitalised at ~$8.65B: 9.7x trailing sales, 50.9x adjusted EBITDA, 93x GAAP EBITDA, 85x FY2026 non-GAAP EPS, 16.6x tangible book, and a 1.5% free-cash-flow yield on record free cash flow. The stock sits in the 97th percentile of its own price-to-sales history and the 94th percentile of its composite own-history valuation — after a 34% drawdown from the June high. Melexis, the closest pure-play comparable at near-identical revenue, trades at 3.0x EV/Sales while earning a 16.0% operating margin. To justify today’s EV at 20x EV/EBIT, Allegro needs ~$1.6B of revenue at its target margins — 1.8x current, roughly four years out at mid-teens growth.

This report takes no position. It sets out what the price requires to be true, what the evidence supports, and what would falsify each side.

2. Business Overview

What the company sells. Allegro’s portfolio splits into two product families of comparable size. Magnetic sensor ICs ($141M in Q4 FY2026, +21% YoY) measure position, speed and current without physical contact, using Hall-effect and increasingly tunnel-magnetoresistance (TMR) transducers. Power ICs ($102M in Q4 FY2026, +35% YoY) comprise motor driver ICs, voltage regulators and LED drivers, and isolated gate drivers (IGDs). The company also supplies photonic and 3D-sensing components — photodiodes, eye-safe lasers and readout ICs for LiDAR — though these are immaterial to current revenue.

The technical core is a proprietary BCD wafer process qualified to automotive voltage and temperature transients — Allegro cites a 120V, 175°C-capable platform integrating high-density logic and EEPROM alongside Hall and xMR transducers on a single die. This matters because it is what allows a single part to carry a configurable, embedded algorithm rather than being a passive transducer, and it is the technical basis for the 12V → 48V → 800V architecture transitions that the company argues drive content growth in both electrified vehicles and high-power AI racks.

How it makes money. Allegro sells components to OEMs and Tier-1 suppliers through a direct sales force, independent representatives, third-party distributors and consignment arrangements. Revenue is not contractually recurring — there are no subscriptions, no royalties, no maintenance streams — but it behaves like an annuity once a part is designed into a vehicle platform, because that platform ships for 10–15 years and the part cannot be economically substituted mid-life. The commercial event that matters is therefore the design win, which occurs years before revenue.

Distribution is a larger channel than casual reading suggests: per the Q3 FY2026 call, all Japanese sales, more than half of Chinese sales, and 90% of industrial sales including data center are serviced through distributors. This makes reported sell-in a noisy proxy for end demand and means channel inventory swings can amplify both directions of the cycle — as they did in FY2025.

Revenue segmentation (FY2026).

Geography. 89.7% of FY2026 sales were outside the United States, with Greater China 28.0%, Japan 17.0% and Taiwan 10.0%. On a Q4 ship-to basis: rest of Asia 30%, China 25%, Japan 17%, Europe 15%, Americas 13%. This is a company with roughly a quarter of its revenue in China and the large majority of its manufacturing and end-customer base in Asia.

Customer concentration is low. No customer accounted for 10% or more of net sales in FY2024–FY2026. The largest non-affiliated distributor was 9.4% of FY2026 sales (9.3% FY2025, 10.2% FY2024). No distributor or customer represented 10%+ of trade receivables. Sanken, the former majority holder, was 0.6% of sales in FY2024.

Verdict. A focused, technically credible component franchise with genuine product differentiation in a narrow domain, low customer concentration, and a capital-light manufacturing model. It is also a single-domain business with no recurring revenue, heavy automotive dependence, and substantial exposure to Chinese and broader Asian end markets. Good niche business; structurally cyclical delivery.

3. Industry Dynamics

Structure. Allegro competes in analog and mixed-signal semiconductors — specifically magnetic sensing and application-specific power management for motion control. The 10-K declines to name competitors, describing only “a broad spectrum of global semiconductor designers and manufacturers, ranging from large-scale diversified entities to specialized participants,” and conceding that “some of our competitors have substantially greater financial, technical, marketing, and management resources than we have.” In practice the competitive set is: Melexis (the closest pure-play magnetic-sensor comparable), Infineon, TDK/Micronas, Asahi Kasei Microdevices, and the broad-line analog houses — NXP, STMicroelectronics, Texas Instruments, Analog Devices, onsemi, Renesas and Rohm — each of which can and does attack individual sockets.

Analog semiconductors are structurally more attractive than logic or memory: process nodes are mature and long-lived, capital intensity is lower, product lives run a decade or more, and portfolios are broad enough that no single part determines a franchise. Allegro’s fabless-for-wafers model amplifies the capital-light attributes — FY2026 capex was $38.2M, 4.3% of sales — but forfeits the process-cost advantage that Texas Instruments’ internal 300mm analog fabs confer. It is an asset-light model, not a low-cost one.

Market size and growth. Management frames its automotive serviceable market at ~$8B, of which e-mobility (xEV plus ADAS) is ~$5B growing at a claimed 16% CAGR, and describes data center as growing “well north of 20%” long-term. These are company estimates, not independent data, and should be treated as such.

The content thesis. Allegro’s growth argument is content-per-unit rather than unit volume, and it is specific:

• Data center: ~$150 of Allegro content per rack today, rising to >$425 in next-generation AI configurations, as racks move from 15kW toward 1MW. The mechanism is that thermal and current-sensing requirements scale with power density rather than with server count.
• Automotive: steer-by-wire ADAS systems carry a 2–3x content uplift over conventional electric power steering; isolated gate drivers carry a 2–3x uplift as platforms move from 400V to 800V.
• Robotics: up to 150 sensor ICs and 50 power ICs in an advanced humanoid robot.

Where each sits in the capital cycle — and these are two very different places:

Automotive analog is late-trough. Distributor inventories fell ~50% over five quarters; point-of-sale and sell-in equalised in Q3 FY2026; management reports “thin” Tier-1 inventories with “no clear signs of restocking”; and Allegro’s own auto shipments were still ~20% below their prior peak as of Q3 FY2026. In Marathon’s framework this is the attractive end of the cycle — supply discipline is maximal, capital has left, and returns mean-revert upward. The FY2026 recovery is real and there is more of it available.

AI data center is early-boom. Capital is flooding in and every analog and power vendor — onsemi, TI, NXP, ADI, Infineon, Vicor, Monolithic Power, Navitas — is pitching a variant of the same 800VDC content story. Critically, the majority of Allegro’s data-center revenue today is fan-driver ICs, which the CFO concedes ship “slightly below fleet average from a gross margin standpoint.” A three-phase motor driver for a cooling fan is a legitimate product but not a defensible one: it is a commoditizable socket in a market attracting maximum competitive entry. The genuinely differentiated data-center products — high-bandwidth TMR current sensors and isolated gate drivers — are earlier: current sensors moved from ~0% to ~20% of data-center revenue during FY2026, and isolated gate drivers remain 18–24 months from material revenue.

Regulatory and structural factors. Automotive qualification (AEC-Q100, ISO 26262 ASIL-D) is a genuine barrier — it takes years and creates the switching costs described above. Against that: annual price-down expectations are contractual in many long-term agreements; roughly a quarter of revenue sits in China, where domestic sensor and power-IC competitors are subsidised and improving (management noted a robotics win where “our current sensors were selected over local alternatives,” which is also an acknowledgement that local alternatives were competing); and input costs are exposed to commodity swings — gold was a ~200bp gross-margin headwind in FY2026 alone, with freight and fuel adding more.

Verdict. A decent but not excellent industry. The design-in captivity, long product lives, mature process nodes and low capital intensity are genuinely attractive and support mid-cycle returns well above the cost of capital. But violent cyclicality (a 31% revenue collapse in one year), contractual ASP deflation, commodity input exposure, foundry dependence, and rising Chinese competition are real and recurring. Allegro’s automotive exposure sits at the favourable end of the capital cycle today; its data-center exposure sits at the unfavourable end, which is precisely inverted from how the market is pricing the two.

4. Competitive Position

Name the moat. In Greenwald’s taxonomy, Allegro has customer captivity — and only customer captivity. It does not have economies of scale (it is a $890M revenue company competing against $10B+ diversified analog houses), and it does not have a supply/cost advantage (it is fabless; its wafers come from the same merchant foundries available to others; its back-end Philippines facility is a scale asset but not a unique one).

The captivity mechanism is specific and verifiable. An Allegro current sensor or position sensor designed into a braking, steering or traction-inverter system is qualified to AEC-Q100 and ASIL-D functional-safety standards against that specific vehicle platform. Requalifying an alternative part means re-running safety validation on a system where failure is a recall event, for a component that is a small fraction of system cost. Rational Tier-1s do not do this mid-platform. The consequence is that a design win converts into a decade of revenue with only modest annual price erosion.

Is the moat visible in the financials? Partially, and the honest answer matters. The affirmative evidence:

• Pricing power inside sockets. ASP declines run “low single digit” annually, and management characterised CY2026 as “very low single-digit reductions… more favorable in 2026 than historical,” attributing this to competitor pricing dynamics and tight supply. A commodity component would face far worse.
• Peak-cycle returns are strong. FY2023 ROIC was 21.1% and ROE 85.7%; FY2024 ROIC 14.0% and ROE 39.3%. At mid-cycle utilisation this business earns well above its cost of capital.
• Gross margin at peak was 56.1% (FY2023) — a level only achievable with genuine differentiation.
• Incremental gross margin of 60–65% on recovery volumes, confirmed repeatedly and visible in the 660bp of Q4 FY2026 operating-margin expansion.

The disconfirming evidence, weighed honestly:

• The moat did not hold the line in the downturn. Revenue fell 31% in FY2025 and gross margin fell 1,040bp from the FY2024 peak. A true Greenwald franchise with high captivity should show market-share stability and margin resilience. Allegro showed neither. The company attributes this to customer-specific inventory overbuild rather than share loss.
• Allegro is still ~20% below its prior auto peak while several analog peers are at or near record automotive revenue. Management stated flatly in Q3 FY2026 that “share loss is not even a part of the conversation for us… ours is one of share gain.” That is a hypothesis, not evidence, and it sits uncomfortably against the observable data. This is the single most important unresolved question in the file.
• ROIC is currently negative-to-negligible. FY2025 and FY2026 GAAP operating margins of −2.7% and +2.8% mean the business earned less than its cost of capital for two consecutive years.
• The #1-in-magnetic-sensing claim is management’s, repeated on earnings calls, and is not corroborated by an independent share study anywhere in the filings.

Direct comparison against the closest competitor. Melexis NV is the cleanest available benchmark: an automotive-focused magnetic-sensor pure play of almost identical size ($826M TTM revenue vs Allegro’s $890M).

The reading is instructive and cuts both ways. Allegro carries a 770bp higher gross margin than Melexis — evidence of superior product differentiation and mix, and the strongest single quantitative support for the moat claim. But Melexis converts its lower gross margin into a 16.0% operating margin against Allegro’s 2.8%, because Allegro spends 23.1% of sales on R&D and 20.3% on SG&A against a much leaner Melexis cost base. Allegro is buying future growth with current profitability; that may prove correct, but as of today the shareholder owns the higher gross margin and none of the operating return — while paying 3.2x Melexis’s sales multiple.

Verdict. A real but narrow moat: genuine design-in captivity in magnetic sensing, evidenced by superior gross margins and benign pricing, but no scale or cost advantage, no demonstrated share stability through the cycle, and currently no operating return to show for it. This is a differentiated component supplier, not a platform franchise. It is meaningfully narrower than NXP’s automotive-processor captivity or Texas Instruments’ cost position — a distinction that matters enormously when Allegro trades at a premium to both on sales.

5. Growth History and Forward Opportunities

The historical record is cyclical, not compounding.

Over seven years revenue compounded at 3.0% — from $724M in FY2019 to $890M in FY2026 — with a 77% peak-to-trough swing in operating margin. That is the honest baseline. The FY2022–24 expansion was a genuine content story amplified by a once-in-a-generation shortage; FY2025 gave nearly all of it back.

Organic vs. acquired. Growth has been overwhelmingly organic. The two acquisitions — Crocus Technology (TMR sensing, closed 2023-10-31 for ~$408M of cash consideration in FY2024) and Heyday Integrated Circuits (isolated gate drivers) — were technology tuck-ins, not revenue purchases. Both map directly onto the current growth narratives, which is a point in management’s favour: TMR was ~30% of FY2026 sensor product releases, and the Heyday-derived Power-Thru IGD architecture is the basis of the 800V content story.

The FY2026 recovery, decomposed.

Five consecutive quarters of sequential growth, with steadily improving margins. The Q4 exit rate of $243.2M annualises to $973M — still below the FY2024 peak of $1,049M.

Forward opportunities, ranked by evidence quality.

1. Automotive cyclical recovery (highest confidence). Allegro is shipping ~20% below its prior auto peak into thin channel inventories with no restocking yet observed. Simply returning to the prior peak is ~$140M of incremental automotive revenue requiring no share gain and no new product. Management targets SAAR + 7–10%.
2. Automotive content growth (high confidence). Focus Auto (xEV + ADAS) grew 30% in FY2026 to $349M and is now 55% of automotive revenue. Steer-by-wire, high-voltage traction inverters, electromechanical braking and 800V platforms each carry 2–3x content uplift. Design wins rose >30% in FY2026 and backlog exited at a multi-year high. This is the most credible durable growth vector.
3. Data-center current sensors (medium confidence). Genuinely differentiated — Allegro’s 10MHz TMR current sensor was named EDN’s 2025 Sensor Product of the Year as the industry’s first — and ramping fast, from ~0% to ~20% of data-center revenue within FY2026, at above-fleet-average margins. This is the highest-quality part of the data-center story and currently the smallest.
4. Data-center fan drivers (medium confidence on volume, low on economics). The majority of current data-center revenue. Management’s liquid-cooling rebuttal is specific and credible — fans are proliferating into power supplies and network switching, which are rarely liquid-cooled, offsetting the loss of GPU-cooling fans. But this is a below-fleet-margin, commoditizable socket in a market attracting maximum entry.
5. Isolated gate drivers (medium confidence, long-dated). Differentiated Power-Thru architecture with claimed 40% efficiency gains, sampling to data-center power-supply leaders, but 18–24 months from material revenue.
6. Humanoid robotics (low confidence, longest-dated). Content of 150 sensor + 50 power ICs per robot is striking, and two Chinese robotic-joint design wins were secured in Q4 FY2026. But management’s own framing is “revenue ramp starting to happen probably two or three years out,” with unit volumes today in the “tens of thousands” and the path to millions “really up to the market.” Robotics and automation sales doubled in FY2026 from a small base.

Verdict. Mixed quality. The near-term growth is high-confidence but is substantially cyclical recovery — reclaiming lost ground, not creating new. The automotive content story is genuine, durable and the best part of the franchise. The data-center growth is real and fast but is today concentrated in the least differentiated product line, with the differentiated products 18–24 months out. The robotics story is a call option with a 2–3 year fuse, not a current earnings driver. A seven-year 3.0% revenue CAGR is the record; mid-teens forward growth is the promise.

6. Financial Quality

The GAAP/non-GAAP gap is the single most important number in this file.

The bridge is dominated by $47.9M of stock-based compensation — 5.4% of sales — plus intangible amortization and restructuring. SBC has grown every year: $42.5M (FY2024), $41.9M (FY2025), $47.9M (FY2026). This is not a one-time item; it is a recurring cost of retaining engineers, and it is settled in shares that dilute existing owners. Any valuation built on the $0.54 figure is valuing a company that pays 5.4% of its revenue in a currency the analysis has agreed to ignore.

The non-GAAP figure is further flattered by tax: the effective non-GAAP rate was 6% in Q4 FY2026, 8% for the fiscal year, and is guided to 9% for Q1 FY2027. At a normalized 20% rate, FY2026 non-GAAP EPS would be closer to $0.47. Expensing SBC at $47.9M (~$0.26/share) against that takes the genuinely economic figure to roughly $0.21–0.28 — which is broadly where independent analysis has landed (“true earnings trend at only $0.20–$0.30 per share,” Seeking Alpha, 2026-01-30).

Margins and operating leverage. The operating leverage is real and should not be dismissed. Incremental gross margin (“drop-through”) runs 60–65% and was guided near 70% for Q1 FY2027; Q4 FY2026 non-GAAP operating margin of 15.6% was 660bp above the prior-year quarter on 26% revenue growth. Management states the cost base is “built to service well over a billion dollars” of revenue with only inflationary opex growth thereafter, and has held G&A roughly flat for five years while reallocating dollars into R&D. If revenue compounds, margins expand mechanically.

But the starting point is poor and the recovery is incomplete: GAAP gross margin of 46.3% remains ~850bp below the FY2024 peak of 54.8%, and management’s own >55% target is a 3–5 year objective. The gross-margin bridge management describes is: (i) volume leverage over the fixed Philippines back-end — market-dependent; (ii) gold-to-copper wire-bond conversion, worth up to the ~200bp that gold cost in FY2026 — controllable but gated by customer requalification; (iii) ongoing factory efficiencies. Only one of the three is within management’s control.

Cash flow.

FY2026 free cash flow was a record in both dollars and margin, helped by a $69.7M working-capital release (notably a $49.4M swing in prepaid assets and $47.3M in payables) that will not repeat at the same magnitude. Underlying cash generation is healthy but the 14.0% margin flatters the run-rate. Capex has normalised to ~4.3% of sales after the FY2024 capacity build.

Balance sheet (as of 2026-03-27). Cash $168.8M; term loan $285.0M maturing October 2030 at SOFR + 1.75% with 0.00% amortization; finance leases $18.6M; net debt ~$116M; net leverage below 1.0x; $256M undrawn revolver; current ratio 3.45x. Total equity $956.4M, but goodwill ($203.3M) and other intangibles ($238.7M) total $442.0M, leaving tangible equity of ~$514M and tangible book value of ~$2.78 per share. Retained earnings are negative (−$68.5M) — a legacy of the Sanken share repurchase accounting, not of operating losses. This is a comfortable, unlevered balance sheet that poses no solvency risk and provides genuine optionality for tuck-in M&A or buybacks.

Working capital. Q4 FY2026 DSO of 35 days (from 40) and inventory of 128 days (from 133) are both improving, but 128 days of inventory is high in absolute terms and the cash conversion cycle of 144 days ties up meaningful capital. Inventory rose modestly in absolute dollars during the recovery, which is appropriate.

Returns on capital — the sharpest indictment. ROIC was 17.7% (FY2022), 21.1% (FY2023), 14.0% (FY2024), then not meaningful in FY2025 and FY2026 as operating income went negative and then near-zero. A business earning 21% ROIC at the peak and negative returns at the trough, averaging perhaps 10–12% through the cycle against a cost of capital that is higher for a 2.29-beta equity, is a cyclical business, not a compounder. That distinction is the whole argument.

Verdict. Economics do improve with scale — decisively — but the through-cycle record is mediocre and the reported profitability is far weaker than the promoted profitability. Balance sheet strength is a genuine positive. Free cash flow is real. But two consecutive years of GAAP losses, 850bp of unrecovered gross margin, 5.4%-of-sales stock compensation excluded from the headline number, and a sub-10% tax rate propping up non-GAAP EPS mean the “earnings” being capitalised at 85x are substantially an accounting presentation.

7. Capital Allocation

The defining transaction: the Sanken exit. Sanken Electric had controlled Allegro since 1990 and still held a majority stake after the 2020 IPO. On 2024-07-23 Allegro agreed to repurchase 38,767,315 shares from Sanken in a privately negotiated transaction priced at the underwriters’ price in a concurrent public equity offering, executed in two closings, with the second conditioned on receiving at least $300M of incremental term loans. Sanken reimbursed Allegro’s expenses and paid a $35.0M facilitation fee, credited to additional paid-in capital.

Strategically this was necessary and correct: it removed a controlling shareholder whose interests were not aligned with minority holders, eliminated the controlled-company overhang, and created a genuine public float. Allegro extracted a $35M fee from the seller, which is good negotiating.

The timing and financing were poor. Net debt went from ~$31M (Q1 FY2025) to ~$238M (Q3 FY2025), and the company took on that leverage immediately before revenue fell 31% and operating income went negative. Total FY2025 equity repurchase outflow was $853.9M against $665.9M of issuance. The company bought back nearly 39M shares at what proved to be close to the cyclical top, then watched its stock fall to $16.53. The counterfactual is instructive: onsemi, trebled its buyback to repurchase 28M shares at ~$49 at the trough. Allegro did the opposite.

Since then, capital allocation has been disciplined. FY2026 saw $60M of voluntary debt paydown, the term loan reduced to $285M, and two repricings down to SOFR + 175bp. Net leverage is below 1.0x. There is no dividend and no meaningful open-market buyback — appropriate for a company that just delevered and whose stock trades at 97th-percentile valuations.

M&A record. Crocus Technology (TMR magnetic sensing) closed 2023-10-31; FY2024 cash for acquisitions was $408.1M. Heyday Integrated Circuits (isolated gate drivers) was earlier. Both were technology acquisitions rather than revenue purchases, and both have demonstrably fed the current roadmap — TMR reached ~30% of FY2026 sensor product releases and underpins the differentiated current-sensor ramp; Heyday’s architecture is the Power-Thru IGD platform. This is genuinely good tuck-in M&A: bought capability, integrated it, and it shows up in products. Goodwill of $203M against these has not been impaired.

R&D and SG&A intensity. R&D rose to $205.8M in FY2026, 23.1% of sales, up from 16.8% in FY2024 — a large step-up funded partly by holding G&A flat for five years and reallocating. This is a deliberate bet: spend through the trough to capture the content cycle. It is defensible strategy, but it is why Melexis earns a 16% operating margin on a lower gross margin while Allegro earns 2.8%. Shareholders are funding an option, and the option’s value depends entirely on whether the design wins convert.

Incentives and insider behaviour. Compensation is heavily equity-weighted — $47.9M of SBC against 4,250 employees — with annual grants vesting in May and tax withholding at $43.10 in 2026. The insider record is the concerning part:

Across all 230 Form 4s filed since July 2021, there is not one open-market purchase (code P) by any insider. Not at the $16.53 all-time low in April 2025, not at any point in five years.

Meanwhile, five insiders sold discretionarily into the 2026 rally: Lynch 16,711 shares at $45.46 (2026-05-14), Kent 2,642 at $41.56 (2026-05-18), Webster (CAO) 5,217 at $44.89 (2026-05-21), Coleman (SVP GM Products) 4,500 at $45.72 (2026-05-22), and Madormo (SVP Worldwide Sales) 5,000 at $52.72 (2026-06-03). The dollar amounts are individually modest and post-vest sales are routine. But the asymmetry is unambiguous: insiders have sold into a 3x rally and never once bought weakness.

Board refresh. Robert Willett (appointed 2026-05-13) and Brian White (2026-06-17) joined as independent directors — a reasonable post-Sanken governance normalisation.

Verdict. Mixed, trending better. The tuck-in M&A is genuinely good and the post-2025 deleveraging is disciplined. The Sanken exit was strategically right but financed with debt at a cyclical peak and immediately followed by a 31% revenue collapse — the opposite of counter-cyclical capital allocation. The R&D step-up is a defensible bet that has not yet paid. And with the stock having tripled, no insider has demonstrated conviction with their own money in five years.

8. Changes and Headwinds — Last Two Years

Strategic and structural.

• Sanken’s complete exit (July 2024): 38.8M shares repurchased plus a concurrent public offering; the controlled-company structure ended and the float was created. Financed with incremental term debt. Strengthens governance; weakened the balance sheet at the wrong moment.
• onsemi’s $35.10/share all-cash proposal (~$6.9B), withdrawn 2025-04-14. onsemi cited “the reluctance of Allegro’s board of directors to fully engage and explore our proposal” and concluded there was “no actionable path forward.” Allegro’s all-time low of $16.53 followed one week later. Thesis-relevant in both directions: it evidenced strategic value in the asset, and it establishes what a fully-informed strategic buyer would pay — 31% below today’s price.
• January 2025 global restructuring, extended June 2025, substantially completed in FY2026: repositioning to lower-cost regions, facility consolidation, workforce rebalancing. Strengthens the cost base.
• Polar Semiconductor stake diluted to ~10.2% following the PSL recapitalization; equity-method losses continue to flow through other non-operating expense ($18.6M in FY2026, $37.1M in FY2025). Weakens reported earnings below the line.
• Debt refinancings: Amendment No. 3 (2025-02-06) created a $375M 2030 tranche; Amendment No. 4 (2026-01-21) replaced it with $285M maturing October 2030 at SOFR + 175bp with zero amortization. Strengthens materially.

Operational.

• FY2025 trough and FY2026 recovery: revenue $1,049M → $725M → $890M. Five consecutive quarters of sequential growth into FY2027.
• Data center from ~2% to 14% of quarterly sales in roughly six quarters, with current sensors going from ~0% to ~20% of that business.
• Sales force reorganised by end market rather than geography during FY2026.
• Management changes: Ian Kent promoted to SVP Operations and Jamie Haas to VP/CTO (February 2026); two new independent directors (May, June 2026).
• Analyst Day, 2026-02-18 (Boston): 3–5 year target model of mid-teens revenue growth and gross margin above 55%. This event, more than any earnings print, appears to have crystallised the market’s re-rating.

Headwinds, current and live.

• Gold cost: a ~200bp gross-margin headwind in FY2026, with freight and fuel adding more. The gold-to-copper conversion mitigation requires customer requalification and will phase in over multiple quarters.
• The destocking tailwind is exhausted. Sell-in and POS equalised in Q3 FY2026; distributor inventories are already down ~50%. Growth from here must come from end demand, not channel refill.
• China is ~25–28% of revenue amid rising domestic competition and geopolitical risk.
• Mature-node foundry tightness was raised by analysts on the Q4 call; management downplayed it but conceded select price increases and surcharges beginning at the end of Q1 FY2027.
• Auto shipments remain ~20% below the prior peak while several peers are at records — an unresolved competitive question.
• Valuation-driven de-rating: Barclays downgraded to Equal-Weight on 2026-07-20 with a $48 price target maintained; the stock is −33.9% from its June 30 high.

Verdict. On balance these changes strengthen the business and weaken the stock’s setup. Governance, cost structure, balance sheet, product portfolio and end-market diversification are all better than two years ago — materially so. But the two cheapest sources of growth (channel refill and margin recovery off a trough) are largely spent, and the valuation has moved far faster than the fundamentals.

9. Risk Analysis

The risk that matters. Items 1 and 2 are the thesis. Allegro is not at risk of failing — the balance sheet is sound and the franchise is durable. It is at risk of being a fine business that delivers poor shareholder returns because the entry multiple already discounts four years of flawless execution. Item 8 (single-site back-end) is the tail risk most likely to be underpriced by the market.

10. Valuation Discussion

Where the stock trades. At $46.03 (2026-07-24), with 185.4M shares outstanding and ~$116M of net debt:

Own-history context. Per AZI’s own-history percentile ranks (2026-07-23, at $50.07): composite 94.0th percentile, price-to-sales 97.0th percentile (10.44x), price-to-book 91.0th percentile (9.70x). The P/E percentile is null because trailing GAAP EPS is negative. At $46.03 these moderate only slightly — P/S ~9.6x, P/B ~8.9x. After a 33.9% drawdown, Allegro still trades near the richest sales multiple in its listed history. This is the single most important valuation fact in the report, and it should be read as own-history context only — never cross-sectionally.

Peer context.

Allegro trades at a substantial premium to every one of its peers on sales while generating the lowest operating margin of the group. Both onsemi and NXP trade at their own 95th/96th-percentile valuations; Allegro is more expensive than both on sales, less profitable than both, and at a higher own-history percentile.

Embedded-expectations analysis — what must be true. This is the cleanest way to frame the question. Take the enterprise value of $8.65B and ask what operating income justifies it at various multiples, then what revenue that requires at management’s own target margins.

At a fair-for-analog 20x EV/EBIT, today’s price requires Allegro to generate ~$1.60B of revenue at a 27% operating margin — 1.8x FY2026 revenue, roughly 4.2 years away at the Analyst Day’s mid-teens growth rate. Even granting a generous 25x multiple, the requirement is $1.28B of revenue (1.44x current) at full target margins.

Note what this bridge assumes: that revenue compounds at mid-teens without a cyclical interruption, that gross margin recovers the full 850bp to above 55%, that operating margin reaches ~27% (a level Allegro has never achieved — the FY2023 peak was 20.9% GAAP), and that the market still pays 20–25x EBIT at the end of it. You are being asked to pay the FY2030 price today, undiscounted, with no margin of safety.

A discounted cross-check. Assume management delivers the target model in full: FY2030 revenue of $1.60B, 55% gross margin, 27% operating margin, $432M EBIT, 21% tax, ~$340M net income on ~190M shares = ~$1.79 of EPS. At a 25x exit multiple that is ~$45/share in 2030 — today’s price. Discounted back four years at a 10% cost of equity (charitable for a 2.29-beta stock), the present value is ~$31. To get to $46 today on a discounted basis you need either an exit multiple above 35x or growth materially faster than mid-teens.

Scenario analysis.

The base case — which embeds genuinely good execution — lands below the current price. The bull case requires the robotics option to pay and a 30x multiple to persist.

What is the market underwriting correctly? That the cyclical trough is behind, that operating leverage is powerful (60–65% incremental gross margins are real), that the data-center content story is genuine, that the balance sheet is sound, and that the automotive content roadmap (steer-by-wire, 800V, electromechanical braking) will drive above-SAAR growth. All correct.

What is it underwriting incorrectly? Three things. First, it is applying an AI-infrastructure multiple to a business that is 71% automotive and whose data-center revenue is majority-composed of below-fleet-margin fan drivers. Second, it is treating non-GAAP EPS as economic earnings when that figure excludes 5.4%-of-sales stock compensation and rests on a 6–9% tax rate. Third, it is pricing no cyclicality at all into a company that delivered a −31% revenue year eighteen months ago.

11. Variant Perception

What consensus believes. Eleven-to-twelve covering brokers rate the stock “Moderate Buy” (eight-to-ten buys, one-to-two holds, one sell). The consensus narrative: Allegro is a structural winner in the electrification-and-AI-power content cycle, with data center compounding >20%, automotive content growing at SAAR + 7–10%, powerful operating leverage carrying gross margin above 55%, and robotics as a free option. The stock hit an all-time high on 2026-06-29 on a “Best Idea” rating. Consensus price targets implied ~30% upside as recently as May 2026.

The strongest bull case. Fixed costs are built for “>$1B” of revenue; incremental gross margins are 60–65%; Q4 FY2026 already showed 660bp of YoY operating-margin expansion on 26% revenue growth. If revenue compounds at 15–20% while gross margin marches to 55%+, operating income goes from $25M (FY2026 GAAP) to $400M+ within four years — a 16x increase — and the current multiple looks prescient rather than absurd. Three independent growth vectors support this: automotive content (steer-by-wire and 800V at 2–3x uplift, with design wins up >30% and backlog at multi-year highs), data center (content scaling from $150 to $425 per rack as racks go from 15kW to 1MW, with differentiated TMR current sensors and isolated gate drivers still to ramp), and robotics (150 sensors + 50 power ICs per humanoid). Allegro is genuinely #1 in a niche with real design-in captivity, evidenced by a gross margin 770bp above Melexis’s, and it is the only pure-play way to own magnetic sensing in a US listing.

The strongest bear case. You are paying 9.7x sales and 51x adjusted EBITDA for a company that lost money on a GAAP basis in each of the last two years, whose seven-year revenue CAGR is 3.0%, whose gross margin is still 850bp below its own peak, and which has never in its history earned the 27% operating margin the current price requires. The closest comparable trades at 3.0x sales with five times the operating margin. A strategic buyer with full diligence offered $35.10 and walked. Not one insider has bought a share in five years, while five sold into the rally. The data-center business — the entire reason for the re-rating — is 14% of sales, majority fan drivers, below fleet-average margin, in the most competitively crowded corner of semiconductors. The channel-refill and trough-margin tailwinds are spent. And the stock has already fallen 33.9% from its high in four weeks while remaining in the 97th percentile of its own valuation history — meaning the de-rating has barely begun.

The 3–5 assumptions that actually matter.

1. Does gross margin reach 55%+? Everything depends on this. It requires volume leverage (market-dependent), the gold-to-copper conversion (controllable but slow), and continued factory efficiency. At 50% rather than 55%, the target-model EBIT falls by roughly a third and the valuation bridge breaks.
2. Does data center compound >20% for multiple years, and does its mix shift toward current sensors and IGDs? If it stays fan-driver-dominated, it is dilutive to margin even while additive to revenue.
3. Is the ~20%-below-peak automotive shortfall inventory or share? Management says inventory. If it is share, the automotive content story — 71% of revenue — is compromised at the root.
4. Does the cycle hold through 2027–28? The model requires uninterrupted mid-teens growth. One down year of the FY2025 variety resets the entire bridge by two-plus years.
5. Does the market keep paying an AI multiple for a 71%-automotive company? This is the factor question, and it is the one most likely to resolve first.

The factor-positioning read. The quantitative evidence is unusually clean here and it supports the bear framing on positioning specifically. The stock carries a raw beta of 2.29, a Market loading of +1.66, a Semiconductors industry loading of +1.35, and a LowVolatility loading of −1.12 (R² 0.544) — it is a high-beta, anti-defensive vehicle, not a quality compounder. Idiosyncratic volatility is 41.5% annualised. The risk-adjusted record is revealing: +119% annualised over six months, +46.9% over one year, but −1.4% annualised over three years, and a −68.6% maximum drawdown over five. Most telling of all, its nearest factor neighbours are semicap and AI-cycle names — XSD (0.964), MKSI (0.956), COHU (0.956), ENTG (0.951), NVMI (0.951), ACLS (0.947) — while its actual fundamental peers rank below them (ADI 0.943, NXPI 0.936, MCHP 0.935). The market has statistically re-bucketed Allegro out of auto analog and into the AI-capex complex. That is precisely the crowded-trade signature that precedes violent de-ratings when the narrative cools — and the 33.9% drawdown since June 30, on a broken 21- and 50-day moving average, suggests the cooling has started. This is evidence about where consensus is offsides, not a price call.

Where I think consensus is wrong. Not on the business — consensus has the operating leverage and the content story broadly right. Consensus is wrong on classification. It has priced a $890M-revenue, 71%-automotive, GAAP-lossmaking component supplier as though it were an AI-infrastructure compounder, on the strength of a data-center business that is 14% of sales and majority-composed of the least differentiated product in the portfolio. When the AI-capex factor cools — as it began to on 2026-07-01 — the stock reverts toward what it fundamentally is, and what it fundamentally is trades at 3–6x sales, not 9.7x.

12. Fact vs. Interpretation

13. Open Questions

1. Is the ~20% automotive shortfall versus peak inventory or share loss? Management denies share loss categorically. Several analog peers are at or near record automotive revenue. No third-party share data resolves this. This is the highest-value unresolved question in the file — it determines whether 71% of revenue is a recovering franchise or an eroding one.
2. What is the actual gross margin of the data-center business? Management says fan drivers are “slightly below fleet average” and current sensors “slightly better.” Without the split, it is impossible to know whether data-center growth is accretive or dilutive to the 55% target.
3. Is the “#1 in magnetic sensing” claim verifiable? It appears only as management assertion. Melexis, Infineon, TDK/Micronas and AKM would each contest it in specific sub-segments.
4. What is the real recurring tax rate? Non-GAAP EPS rests on a 6–9% rate. FY2026 GAAP shows a −$248K tax benefit on a −$14.9M pretax loss, with domestic operations losing $28.4M and foreign earning $13.5M. What happens to EPS when the rate normalises?
5. How much of the FY2026 record FCF is repeatable? A $69.7M working-capital release contributed materially. Normalised FCF is likely closer to $80–100M, implying a ~1.0–1.2% yield rather than 1.5%.
6. Why has no insider bought a share in five years? Including at $16.53. Is this a comp-structure artifact or a genuine signal?
7. What did onsemi see in diligence? Its proposal was withdrawn citing board non-engagement, not diligence findings — but a strategic buyer’s $35.10 valuation is a data point about intrinsic value that deserves explanation at $46.03.
8. How exposed is AMPI (Philippines) as a single point of failure? The 10-K flags it explicitly. What is the realistic recovery time and what second-source OSAT capacity exists?
9. What are the actual contractual ASP-decline commitments in the long-term agreements management referenced? “Low single digit” is a characterisation, not a disclosure.
10. Does the Polar Semiconductor equity-method drag persist? $18.6M (FY2026) and $37.1M (FY2025) of other non-operating expense on a now-10.2% stake.

14. What Must Be True

Bull case — what must be true

1. Gross margin must reach 55%+ within 3–5 years, requiring volume leverage over the Philippines back-end, full realisation of the gold-to-copper conversion (worth up to ~200bp), and continued factory efficiency gains — all while absorbing low-single-digit annual ASP declines.
2. Revenue must compound at mid-teens or better without a cyclical interruption, reaching ~$1.3–1.6B by FY2030.
3. Data center must sustain >20% growth AND shift mix toward high-bandwidth TMR current sensors and isolated gate drivers, so that it becomes margin-accretive rather than dilutive.
4. Automotive must return to and exceed its prior peak at SAAR + 7–10%, confirming the shortfall was inventory rather than share.
5. Operating margin must reach ~27% — a level never achieved in company history (peak: 20.9% GAAP in FY2023).

Falsification test (bull): Two consecutive quarters in which non-GAAP gross margin fails to expand sequentially while revenue grows, or a data-center growth rate printing below 15% year-over-year. Either breaks the operating-leverage bridge on which the entire valuation rests. A secondary test: any quarter in which automotive revenue declines year-over-year in a non-seasonal period.

Bear case — what must be true

1. The multiple must compress from 9.7x EV/Sales toward the 3–6x range where comparable magnetic-sensor and auto-analog franchises trade.
2. Data-center growth must decelerate or prove margin-dilutive, exposing it as a fan-driver commodity ramp rather than a differentiated content franchise.
3. Gross margin must stall in the 50–52% range, short of the >55% target, leaving operating margin in the mid-to-high teens rather than the high twenties.
4. The automotive cycle must fail to deliver a return to peak, whether from a SAAR downturn, an xEV air-pocket, or genuine share loss.
5. The AI-capex factor must cool, removing the re-classification premium that carried the stock from $16.53 to $69.62.

Falsification test (bear): Non-GAAP gross margin crossing 53% while revenue grows >20% year-over-year, with data center sustaining >25% growth and its current-sensor/IGD mix exceeding ~40% of data-center revenue. That combination would put $1.3B+ of revenue and >$1.20 of clean EPS credibly in view within three years and would make today’s multiple defensible rather than heroic. A secondary test: any insider making a material open-market purchase, which would materially undercut the conviction argument.

15. Source Appendix

(Full source list in Appendix B.)

Primary sources

• Allegro MicroSystems, Inc., Form 10-K for fiscal year ended March 27, 2026, filed 2026-05-21. https://www.sec.gov/Archives/edgar/data/866291/000119312526233537/algm-20260327.htm
• Allegro MicroSystems, Inc., Form 10-K for fiscal year ended March 28, 2025, filed 2025-05-22.
• Allegro MicroSystems, Inc., Form 10-Q for the quarter ended December 26, 2025, filed 2026-01-30.
• Allegro MicroSystems, Inc., DEF 14A, filed 2026-06-24.
• SEC Form 4 corpus, CIK 0000866291 — 230 filings from 2021-07-01 to 2026-06-18, parsed in full 2026-07-24.
• Allegro MicroSystems Q4 FY2026 earnings call transcript, 2026-05-07 (via ROIC.ai).
• Allegro MicroSystems Q3 FY2026 earnings call transcript, 2026-01-29 (via ROIC.ai).
• Allegro MicroSystems, “Reports Fourth Quarter and Fiscal Year 2026 Results,” 2026-05-07.
• onsemi, “onsemi Withdraws Proposal to Acquire Allegro MicroSystems,” 2025-04-14. https://investor.onsemi.com/news-releases/news-release-details/onsemi-withdraws-proposal-acquire-allegro-microsystems

Quantitative data

• ROIC.ai: income statement, balance sheet, cash flow, profitability ratios, enterprise value, valuation multiples — ALGM and MLXSF (Melexis), retrieved 2026-07-24.
• AZI Trading price history CSV, ALGM, 1,439 sessions from 2020-10-29, retrieved 2026-07-24.
• AZI Trading valuation_index own-history percentile ranks, ALGM, 2026-07-23.
• FactorsToday factor model: stock-loadings, leaderboard, stock-info, stock-specific-vol, related-stocks — ALGM, retrieved 2026-07-24.

Secondary

• Barclays downgrade to Equal-Weight, price target $48, 2026-07-20 (via MarketBeat, GuruFocus).
• “Allegro MicroSystems: Enthusiastically Pricing In Green Shoots,” Seeking Alpha, 2026-01-30.
• “Allegro MicroSystems Spikes To All-Time High On ‘Best Idea’ Rating,” Investor’s Business Daily, 2026-06-29.

All figures reconciled to primary filings where available. Third-party aggregated data is used as a cross-check and is not primary. This note contains no investment recommendation and no price target outside the clearly labeled Author’s Take block.

APPENDIX A — Standard Diligence Questionnaire

Allegro MicroSystems, Inc. (NASDAQ: ALGM) — as of 2026-07-24

Supplemental to the note above. Answers are labeled Fact / Interpretation / Assumption where the distinction matters.

General

What thoughtful questions have other investors asked about this company?

The sell-side Q&A across the FY2026 calls converged on five genuinely good questions, and management’s answers ranged from strong to evasive:

1. “Why are you still 20% below peak in automotive when peers are at records — is this share loss?” (Quinn Bolton, Needham, Q3 FY2026). Management’s answer — “share loss is not even a part of the conversation for us… ours is one of share gain,” attributing the gap to specific customers who overbuilt inventory — is an assertion without corroboration. This remains the best question asked and the least satisfactorily answered.
2. “How much of the data-center pipeline is still fan-driver versus current sensors and gate drivers?” (Needham, Q4 FY2026). Answered honestly: the majority is still fan drivers; current sensors went from ~0% to ~20% of data-center revenue during FY2026; isolated gate drivers are 18–24 months from material revenue.
3. “Walk us from 50% gross margin to the >55% target — what’s controllable versus volume-dependent?” (Needham, Q4 FY2026). Answered with unusual candour: leverage (market-dependent), gold-to-copper conversion (controllable, ~200bp), factory efficiencies. Only one of three is within management’s control.
4. “If data center grows 40% again in June, the rest of the business is down 3–4% — so is data center slowing?” (Timothy Arcuri, UBS, Q4 FY2026). A sharp catch. Management conceded data center will be “16%, 17% of total sales” in Q1 FY2027, implying 20–25% sequential growth — a deceleration from ~40%, framed as “the law of large numbers.”
5. “Mature-node foundry is getting tight — are you seeing cost pressure?” (UBS, Q4 FY2026). Management downplayed foundry pressure but conceded gold and fuel headwinds and disclosed select price increases and surcharges beginning at the end of Q1 FY2027.

Interpretation: The buy-side debate is not about whether the business is good. It is about whether a 71%-automotive company deserves an AI multiple, and whether the 55% gross-margin target is achievable. Short interest and the Barclays downgrade (2026-07-20, Equal-Weight, $48) indicate the valuation debate is live.

Cyclicality & Earnings Nature

Are earnings at a cyclical high or low? Fact: Decidedly closer to a low, though off the bottom. FY2026 GAAP operating margin was 2.8% versus a FY2023 peak of 20.9%. Gross margin at 46.3% is ~850bp below the FY2024 peak of 54.8%. Revenue of $890.1M is 15% below the FY2024 peak of $1,049.4M. Automotive is shipping ~20% below its prior peak. Interpretation: This is a mid-recovery earnings base, not a peak — which is the strongest argument for the bulls, and precisely why the valuation is so demanding: you are paying a peak-cycle multiple on a trough-to-mid-cycle earnings stream.

Driven by the external environment or internal actions? Both, with the external dominating. The FY2025 collapse was an industry-wide automotive and industrial inventory correction — distributor inventories fell ~50% over five quarters. The FY2026 recovery is partly channel normalisation (now exhausted — sell-in and POS equalised in Q3 FY2026) and partly genuine internal execution: design wins up >30%, backlog at multi-year highs, the Jan-2025 restructuring completed, 140bp of gross-margin improvement delivered against a 200bp gold headwind.

How stable are revenues? Fact: Not stable. Revenue over eight fiscal years: $724M, $650M, $591M, $769M, $974M, $1,049M, $725M, $890M. Peak-to-trough decline of 30.9% in a single year. Seven-year CAGR of 3.0%. There is no contractually recurring revenue — no subscriptions, no royalties, no maintenance. Interpretation: Revenue behaves like an annuity at the individual-socket level (10–15 year platform lives) but like a commodity cycle in aggregate, because order rates swing with Tier-1 build schedules and channel inventory.

Outlook for products/services? Q1 FY2027 guidance is $245–255M (+23% YoY at the midpoint), non-GAAP gross margin 50–51%, non-GAAP EPS $0.19–0.23. The 3–5 year Analyst Day model (2026-02-18) targets mid-teens revenue growth and gross margin above 55%. Assumption: these are management targets, not commitments, and the operating-margin level implied (~27%) exceeds anything the company has achieved.

How big will this market be — growing, shrinking, domestic or international? Management sizes automotive SAM at ~$8B (e-mobility ~$5B, claimed 16% CAGR) and describes data center as growing “well north of 20%.” These are company estimates without third-party validation. The market is overwhelmingly international: 89.7% of FY2026 sales were outside the United States, with Greater China 28.0%, Japan 17.0% and Taiwan 10.0%.

Business Quality & Competitive Moat

Is the industry getting more or less competitive? More. Three vectors: (i) every broad-line analog vendor — TI, ADI, NXP, ST, Infineon, onsemi, Renesas — is pitching the same 800VDC data-center content story into the same sockets; (ii) Chinese domestic sensor and power-IC suppliers are improving, and management’s own robotics win description (“selected over local alternatives”) confirms they now compete credibly; (iii) mature-node foundry capacity is tightening, raising input costs. Offsetting this: automotive qualification barriers remain genuinely high, and pricing was better than normal in CY2026 negotiations.

How profitable is the business (ROIC, ROE)? Fact: ROIC was 17.7% (FY2022), 21.1% (FY2023), 14.0% (FY2024), then not meaningful in FY2025 (operating margin −2.7%) and FY2026 (+2.8%). ROE was 39.3% in FY2024, negative thereafter. Interpretation: Through-cycle ROIC is perhaps 10–12% — above the cost of capital for a stable business, but marginal for a 2.29-beta equity. This is the single clearest evidence that Allegro is a cyclical, not a compounder.

How profitable is the industry — how many competitors, what barriers to entry? Analog semiconductors are structurally the most attractive sub-sector in semis — mature nodes, long product lives, lower capital intensity, broad portfolios. Peer profitability spans a wide range: NXP ~55% gross / ~25% operating margin; Melexis 38.6% gross / 16.0% operating; onsemi ~38% gross. Allegro’s 46.3% gross margin sits mid-pack but its 2.8% operating margin is bottom-of-class, because it spends 23.1% of sales on R&D. Barriers to entry are application-specific rather than structural: AEC-Q100/ASIL-D qualification, multi-year design cycles, and safety-critical validation. Capital is not the barrier — Allegro is fabless.

Can the business be easily understood? Yes. Allegro sells two families of chips — magnetic sensors and analog power ICs — into cars, factories, and data centers. Revenue is design-win driven; margin is volume-leverage driven; the cycle is the automotive inventory cycle. It is one of the more transparent business models in semiconductors. The complexity lies entirely in the valuation and the accounting presentation, not the operations.

Can it be undermined by foreign low-cost labor? Not by labor — but by foreign capital and subsidised competition, yes. Allegro already manufactures back-end in the Philippines and buys wafers from Asian foundries; it is the low-cost model. The genuine risk is Chinese state-supported sensor and power-IC vendors moving up the performance curve, in a market where ~25–28% of Allegro’s revenue sits.

Do brands matter? No, not in the consumer sense. What matters is the qualification record and the applications-engineering relationship. “Allegro” carries weight with a Tier-1 braking-system engineer because of a two-decade zero-defect record and local coil-design expertise, not marketing. This is a reputational asset but it is not a brand moat.

What is the nature of competition? Competition happens at the design-in, years before revenue, and is decided on time-to-market, application expertise, quality/reliability record, product performance (bandwidth, accuracy, package size), production capacity and price — the 10-K’s own list. Once won, the socket is protected for the platform’s life. So the industry is intensely competitive episodically and barely competitive in between. Price competition manifests as contractual annual ASP step-downs of low single digits.

Customers’ switching costs? Genuinely high mid-platform, genuinely low at the next platform. Replacing a qualified ASIL-D sensor in a shipping braking system means re-running functional-safety validation on a recall-risk component — economically irrational for a part that is a small share of system cost. But at the next platform refresh, every socket is contestable again. Interpretation: This is customer captivity with a 10–15 year half-life, continuously renewing but never permanent.

Financial Condition & Balance Sheet

Assets not fully recognized on the balance sheet? Yes, meaningfully. The design-win backlog and qualified socket base are the company’s most valuable assets and appear nowhere on the balance sheet — Allegro exited FY2026 with backlog at a multi-year high and design wins up >30%, representing years of contracted-by-convention future revenue. Likewise the BCD process IP and TMR portfolio: internally developed R&D is expensed ($205.8M in FY2026), so decades of accumulated process know-how carries zero book value. Conversely, the acquired Crocus/Heyday intangibles are on the balance sheet at $442.0M including goodwill — meaning the acquired IP is capitalised while the (larger) organic IP is not.

Off-balance-sheet liabilities? Nothing unusual identified. Operating and finance leases are recognised ($18.6M of finance leases). There is a defined-benefit pension obligation ($8.7M) — small. No securitisations, no material guarantees, no VIEs beyond the ~10.2% Polar Semiconductor equity-method stake, which is on-balance-sheet as a long-term investment ($25.8M) and whose losses flow through other non-operating expense.

How conservative is the accounting? Mixed — the GAAP statements are clean; the promoted non-GAAP presentation is not. GAAP accounting appears unremarkable: no aggressive revenue recognition (product shipped, largely to distributors with standard reserves), inventory at 128 days with appropriate reserves, no capitalised development costs, no impairments deferred. But the non-GAAP presentation is aggressive: it excludes $47.9M of stock-based compensation (5.4% of sales, growing every year) and applies a 6–9% tax rate, converting a GAAP loss of $0.08 into a promoted $0.54. That is a 1,130bp operating-margin gap. Investors capitalising the $0.54 at 85x are valuing an accounting construct.

How CapEx-hungry is the business? Not very — this is a genuine strength. FY2026 capex was $38.2M, 4.3% of sales, versus $124.8M (11.9%) in FY2024 when back-end capacity was built. Being fabless for wafers means no fab cycle to fund. Maintenance capex is probably 3–4% of sales. This is what allowed free cash flow of $124.9M on 2.8% GAAP operating margins.

Capital Allocation & Management

How much FCF does the business generate, how does management use it, what is the philosophy? Fact: FY2026 FCF was a record $124.9M (14.0% of sales); prior years $20.8M (FY2025), $56.9M (FY2024), $113.4M (FY2023). Caveat: FY2026 benefited from a ~$69.7M working-capital release that will not repeat at that scale; normalised FCF is probably $80–100M. Uses in FY2026 were almost entirely debt reduction — $60M of voluntary paydown, taking the term loan to $285M and net debt to ~$116M, with two repricings to SOFR + 175bp. No dividend. No meaningful buyback. The CFO’s stated philosophy: balance liquidity (targeting roughly six months of opex + capex in cash) against paying down debt, “because that’s just accretive to EPS.” Interpretation: Sensible, unglamorous, appropriate for a company that just delevered.

Significant acquisitions recently? Crocus Technology (TMR magnetic sensing) closed 2023-10-31 — FY2024 cash for acquisitions $408.1M. Heyday Integrated Circuits (isolated gate drivers) earlier. Both are technology tuck-ins rather than revenue purchases, and both demonstrably fed the current roadmap: TMR was ~30% of FY2026 sensor product releases and underpins the differentiated current-sensor ramp; Heyday’s architecture is the Power-Thru IGD platform now sampling to data-center customers. This is good M&A — capability bought, integrated, and visible in products. Goodwill of $203.3M has not been impaired.

Buying back shares? Not currently, and the history is unflattering. The only material repurchase was the 38,767,315 shares bought from Sanken on 2024-07-23 — a strategically necessary transaction to end the controlled-company structure, executed at the underwriters’ price in a concurrent offering, with Sanken paying a $35.0M facilitation fee. But it was funded with incremental term debt immediately before a 31% revenue collapse, taking net debt from ~$31M to ~$238M within two quarters. Total FY2025 equity repurchase outflow was $853.9M against $665.9M of issuance. Interpretation: The company bought back a third of its shares near the cyclical top and had no capacity to buy at the $16.53 trough — the inverse of counter-cyclical allocation. (Compare onsemi, which repurchased 28M shares at ~$49 at its trough.)

Issuing large amounts of new shares to insiders? Yes, and it is growing. SBC was $47.9M in FY2026 (5.4% of sales), up from $41.9M (FY2025) and $42.5M (FY2024). Share count has crept from 184.0M (Q3 FY2025) to 185.4M (Q4 FY2026) despite no offsetting buyback, and the diluted count guided for FY2027 is 187M. Interpretation: ~0.8–1.0% annual net dilution — not egregious for a semiconductor company, but it is a real transfer of ownership that the promoted EPS figure entirely excludes.

Compensation policy of directors/management? Heavily equity-weighted, with annual grants vesting in May (2026 grants vested 2026-05-13 with tax withholding at $43.10). CEO Michael Doogue received 160,835 shares across two grants in May 2026; CFO Derek D’Antilio 59,877. Open question: the specific performance metrics governing PSU vesting were not extracted in this pass and warrant review in the 2026 DEF 14A (filed 2026-06-24) before drawing conclusions on alignment quality.

Motivations of management? Fact — the clearest available signal is the insider record: across all 230 Form 4s filed since July 2021, there is not a single open-market purchase (code P). Not one insider bought a share at the $16.53 all-time low in April 2025, nor at any point in five years. Meanwhile five insiders sold discretionarily into the 2026 rally: Lynch 16,711 @ $45.46 (05-14), Kent 2,642 @ $41.56 (05-18), Webster (CAO) 5,217 @ $44.89 (05-21), Coleman 4,500 @ $45.72 (05-22), Madormo 5,000 @ $52.72 (06-03). Interpretation: The amounts are individually small and post-vest selling is routine, so this should be weighted modestly. But the asymmetry — sold into a 3x rally, never bought a trough — is a genuine absence of demonstrated conviction and stands in contrast to peers whose management bought their own bottoms.

Valuation & Market Data

Is the stock an ADR, MLP, or K-1 issuer? No. Allegro MicroSystems, Inc. is a Delaware-incorporated US domestic filer listed on NASDAQ Global Select, CIK 0000866291, CUSIP 01749D105, ISIN US01749D1054. Ordinary common stock, 1099 reporting, no K-1. (Its former parent Sanken Electric is Japanese, but Sanken fully exited in 2024.)

Dividend policy? No dividend and none expected. The only distribution in company history was a $400M pre-IPO dividend paid to Sanken in FY2021. Current capital priorities are liquidity and debt reduction.

How profitable is the business? Restating for this section: FY2026 GAAP gross margin 46.3%, GAAP operating margin 2.8%, GAAP net margin −1.7%, GAAP EPS −$0.08. Non-GAAP: gross margin 49.4%, operating margin 14.1%, EPS $0.54, adjusted EBITDA 19.1%. Peak profitability (FY2023): 56.1% gross, 20.9% operating, 19.2% net, 21.1% ROIC. Interpretation: currently sub-par, historically capable of good-not-great returns.

Is net income diverging from cash from operations? Yes, dramatically — and in the shareholder-friendly direction. FY2026 GAAP net income was −$14.7M while cash from operations was +$163.1M — a divergence of ~$178M. The reconciliation is legitimate and fully explained: $67.6M of depreciation and amortization, $47.9M of stock-based compensation, $6.6M of impairment, and a $69.7M working-capital release, less $12.0M of deferred tax. Interpretation: This is the normal signature of a company with heavy non-cash charges (amortization of acquired Crocus/Heyday intangibles, plus SBC) operating near breakeven on GAAP. It is not an accounting red flag. But note the direction of the SBC argument: management adds SBC back to justify non-GAAP EPS and it inflates operating cash flow — the same $47.9M is counted as a benefit twice in the bull presentation.

Risks & Downside

What factors would cause the stock to decline? In descending order of probability: (1) multiple compression from the 97th percentile of its own price-to-sales history toward peer norms — already underway, −33.9% since 2026-06-30; (2) data-center growth decelerating below ~15%, or its mix staying fan-driver-dominated and margin-dilutive; (3) gross margin stalling in the 50–52% range short of the >55% target; (4) an automotive air-pocket in 2027–28, for which the FY2025 precedent (−31% revenue) is the template; (5) AI-capex factor cooling, removing the re-classification premium — the stock’s nearest factor neighbours are semicap names, so it will trade with them; (6) evidence of genuine automotive share loss resolving the ~20%-below-peak question adversely.

Risk of a catastrophic loss? Low, in the business; moderate, in the stock. The business faces no solvency risk: net debt ~$116M, net leverage <1.0x, $256M undrawn revolver, no maturity before October 2030, $168.8M of cash, and positive free cash flow even in the trough year. The genuine catastrophic-operational tail risk is the single-site back-end concentration at AMPI in the Philippines, which the 10-K flags explicitly against earthquake, fire, flood, pandemic, strikes, civil unrest and energy shortage. A prolonged AMPI outage would impair the ability to assemble and test essentially the whole product line. The stock, separately, has a demonstrated −68.6% maximum drawdown over five years and a −64% drawdown in 2023–24 — a 50%+ decline from here would be entirely within its own historical range and would require no fundamental catastrophe, only multiple normalisation.

Chance of a total loss? Very low. This is a cash-generative, low-leverage, IP-rich business with a strategic acquirer having bid $35.10/share in cash as recently as 2025. The floor is not book value ($5.16/share) or tangible book ($2.78/share) but strategic value — Allegro’s magnetic-sensing franchise is a scarce asset that onsemi, Infineon, TI or ADI would plausibly bid for again at a sufficiently depressed price. Total loss would require a technology displacement of magnetic sensing itself, for which there is no evidence.

Recent News & Events

Has the business environment changed recently? Yes, materially and in both directions. Improving: five consecutive quarters of sequential revenue growth; data center from ~2% to 14% of quarterly sales in six quarters; design wins up >30%; backlog at multi-year highs; ASP declines in CY2026 negotiations “very low single digit” and better than historical, which management attributes to competitor pricing dynamics and tight supply. Deteriorating: the channel-destocking tailwind is exhausted (sell-in and POS equalised in Q3 FY2026, distributor inventories already down ~50%); gold cost a ~200bp gross-margin headwind with freight and fuel adding more; mature-node foundry tightening; and select price increases/surcharges beginning at the end of Q1 FY2027 — which is both a margin defence and a sign of input-cost pressure.

Significant acquisitions? None since Crocus (October 2023). Management has stated M&A remains part of the strategy and the deleveraged balance sheet now supports it. The more significant recent M&A event was Allegro as a target, not an acquirer: onsemi’s $35.10/share all-cash proposal (~$6.9B) was withdrawn on 2025-04-14, with onsemi citing “the reluctance of Allegro’s board of directors to fully engage and explore our proposal” and concluding there was “no actionable path forward.” Allegro’s all-time low of $16.53 came one week later. At $46.03 the stock trades 31% above that withdrawn bid.

Change in accounting policies? None identified in the FY2026 10-K. The non-GAAP presentation is consistent period-over-period, which makes the comparisons in this report valid even where the presentation itself is aggressive.

Recent changes — new markets, facilities, management?

• New markets: AI data center (10% of FY2026 sales, from ~2%), robotics and automation (revenue doubled year-over-year off a small base, with two Chinese robotic-joint design wins in Q4 FY2026), energy infrastructure.
• Facilities: January 2025 global restructuring — repositioning to lower-cost regions and facility consolidation — extended June 2025 and substantially completed in FY2026. Back-end capacity added in the Philippines ($17M of Q4 FY2026 capex) to address building delinquency in auto and industrial.
• Organisation: the sales force was reorganised by end market rather than geography during FY2026.
• Management: Ian Kent promoted to SVP Operations and Jamie Haas to VP/Chief Technology Officer (2026-02-03). Robert Willett appointed independent director 2026-05-13; Brian White appointed independent director 2026-06-17 — a reasonable post-Sanken governance normalisation.
• Capital structure: Amendment No. 4 (2026-01-21) refinanced into a $285M tranche maturing October 2030 at SOFR + 1.75% with 0.00% amortization.

This appendix is supplemental to the note above and contains no investment recommendation and no price target.

APPENDIX B — Source Appendix

Allegro MicroSystems, Inc. (NASDAQ: ALGM) — research date 2026-07-24

Every non-obvious claim in this note traces to a source below. Primary sources take precedence over aggregated data; where a third-party figure conflicts with a filing, the filing governs.

1. Primary — SEC filings

Corpus statistics (filings since 2021-07-01, excluding 424B/FWP/144 noise):* 377 filings — 230 Form 4, 59 8-K, 18 Form 3, 15 10-Q, 13 SC 13G/A, 6 DEF 14A, 6 DEFA14A, 5 10-K, 4 SD, 4 ARS, 2 S-3ASR, 2 8-K/A, 1 Form 5, 1 Form 425, 1 10-K/A. Local mirror: output/ALGM/sources/.

2. Primary — Management communications

3. Primary — Third-party corporate

4. Quantitative data providers

5. Secondary — financial media and sell-side reporting

6. Analytical frameworks applied

7. Derived figures — calculation basis

Every figure below is the author’s calculation, not a reported number. Inputs are cited above.

8. Verification notes and unresolved items

1. Corrected during research: an initial reading placed the onsemi withdrawal in 2026. The correct date is 2025-04-14, verified against onsemi’s own press release (item 3.1). ALGM’s all-time low of $16.53 followed on 2025-04-21.
2. Transcription artifact, not a management claim: the Q3 FY2026 transcript renders the CFO as saying “we did $1.48 billion in revenue in FY 2024.” The audited figure is $1,049.4M. Treated as a transcription error and not relied upon.
3. Contradiction carried forward: management asserts “share loss is not even a part of the conversation” (item 2.2) while conceding automotive shipments are ~20% below peak at a time when several analog peers report record automotive revenue. Unresolved; recorded above Open Questions and flagged as an Assumption in the Fact-vs-Interpretation table.
4. Unverified management claim: the “#1 position in magnetic sensing” (item 2.1) is not corroborated by an independent share study in any filing. Classified as an Assumption.
5. Not an audited disclosure: data-center revenue is given only as a percentage of sales. The ~$89M FY2026 figure is derived.
6. Not extracted in this pass: the specific performance metrics governing PSU vesting in the 2026 DEF 14A (item 1.6). Flagged as an open item in Appendix A rather than asserted.
7. Third-party data authority: ROIC.ai, AZI and FactorsToday are aggregated/statistical sources, not primary. For this US filer, EDGAR and the 10-K/10-Q govern; where the memo cites a ratio that drives a verdict, the underlying statement line items support it. No material discrepancy between ROIC.ai and the FY2026 10-K was identified during reconciliation.

Source list compiled 2026-07-24.