Pegatron
Contract design + assembly of finished electronics for global OEMs (smartphones, notebooks/AI PCs, game consoles, AI servers, automotive electronics). High-volume, low-margin (gross ~4%, operating ~1%); revenue concentrated in a few large customers (Apple the largest), with net income amplified by non-operating investment income from subsidiaries/affiliates. Shifting mix from consumer devices toward higher-value AI servers and auto to lift structural margin.
Earnings, margins, COGS & capex
A ~NT$1.1T (~$35B) revenue assembler running on ~4% gross and ~1% operating margins - economics typical of contract manufacturing where scale, working-capital discipline, and customer retention matter more than pricing power. FY2025 net income attributable to owners (NT$14.4B) exceeded operating income (~NT$11.6B) because of sizable non-operating/investment income - a recurring feature of Pegatron's earnings and a reason operating trends matter more than headline EPS. Revenue has drifted down since 2022 on softer iPhone/PC volumes; the FY2026 thesis is a mix shift into AI servers (guided ~tenfold growth) and automotive to re-accelerate the top line and slowly widen margins.
Income statement — where each revenue dollar goes
% of revenueOf every $1 of revenue, ~96¢ is cost of goods and ~3¢ operating expense, leaving ~1¢ of operating profit (~1¢ net).
Revenue trend
Margins
structurally thin; management targeting gradual expansion via AI-server/auto mix
compressed; leverage sensitive to volume and program mix
flattered by non-operating investment income above operating profit
COGS structure
COGS is ~96% of revenue - dominated by purchased components (processors, memory, displays, casings, PCBs) and direct labor at assembly sites in China, Taiwan, India, Vietnam, and increasingly the US. Component cost pass-through and customer-owned-material arrangements mean gross margin is driven by assembly value-add and utilization rather than input pricing. AI-server racks carry higher content value per unit (GPU-dense) but Pegatron's value-add slice remains a thin assembly/integration margin.
Capex
Asset-light relative to revenue (~1-2%); FY2026 capex weighted toward a new US AI-server production line (Texas), automation/robotics at existing sites, and automotive-electronics capacity. Historically funded from operating cash flow; the server/US ramp is pressuring near-term FCF to roughly breakeven.
Latest earnings
Miss - EPS ~33% below consensus, revenue ~1.3% below; net profit down >60% YoY and the weakest Q1 EPS in ~7 years, attributed to seasonal off-quarter weakness in consumer devices
Full-year 2026: AI-server revenue to grow ~tenfold YoY (8-10x base, >12x if peak spend); server shipments up each quarter with 2H > 1H; overall operations expected to rise sequentially quarter by quarter; 2Q26 expected stronger on AI PC demand; automotive to keep gaining as share of mix. May 2026 monthly revenue (NT$95.96B, +12% YoY) hit a 2026 high on server momentum.
- Q1 2026 EPS
- NT$0.58
- Q1 2026 revenue
- NT$244.11B
- FY2025 EPS
- NT$5.39 (down ~15% YoY)
- AI-server FY2026 guide
- ~10x YoY (8-10x base, >12x optimal)
- May 2026 revenue
- NT$95.96B (2026 high)
Growth drivers
- AI servers - guided ~tenfold (8-10x base case, potentially >12x if hyperscaler spend stays at peak) revenue growth in 2026 off a small base, driven primarily by Nvidia's GB300 platform (plus GB200/B200) shipping to neocloud providers and hyperscale CSPs; new US server line onlining in 2026, though some shipments may slip to early 2027 on component lead times
- Automotive electronics - guided toward double-digit % of revenue, with Tesla a key client
- AI PC refresh cycle — Windows-refresh + on-device-AI notebook upgrade wave expected to lift 2H 2026 PC/notebook volumes
- Geographic diversification — Vietnam, US, and India capacity to serve customer 'de-risk-from-China' mandates (though Tata now controls 60% of the India iPhone unit)
- Game consoles and smart/IoT devices as steadier base-load volume
Bull & bear
A cash-generative Apple assembler trading at a fraction of sales with an under-appreciated, credentialed pivot into Nvidia AI-server racks and Tesla automotive - if the ~10x server ramp and US line convert into durable higher-value-add revenue, the mix shift could re-rate a stock the market currently prices as a declining low-margin ODM.
- Guided ~tenfold AI-server revenue growth in 2026 off a small base (8-10x base case, potentially >12x), with real Nvidia GB300/GB200/B200 rack wins for neocloud and CSP customers and a new US production line
- Automotive electronics guided toward double-digit % of revenue with Tesla as anchor - a second, structurally higher-value leg
- Large gross cash balance and modest interest-bearing debt fund the server/US/auto capex without balance-sheet strain; the investment portfolio backstops net income
- Trades at ~0.2x sales on a ~$7B market cap and ~$35B revenue - the AI mix-shift optionality is cheaply priced on a sales basis (note: NOT cheap on ~19-20x TTM earnings)
- China+1 diversification (Vietnam, US) aligns Pegatron with the exact supply-chain resilience customers are paying for
The core business is a mature, share-losing, ~1%-operating-margin iPhone assembler in secular revenue decline, and the AI-server story - real but low-value-add and fiercely contested - can multiply revenue without meaningfully moving profit, leaving a low-quality earnings stream propped up by non-operating income and trading at ~19-20x depressed TTM earnings.
- Operating margin ~1% and gross ~4% mean even a ~10x revenue segment adds little profit if it's thin assembly value-add on GPU-heavy BOMs
- Lost iPhone allocation - Foxconn holds >50% and China-based Luxshare surpassed Pegatron to #2; Apple concentration cuts both ways
- Revenue has fallen for multiple years; Q1 2026 net profit down >60% YoY to the weakest Q1 EPS in ~7 years, and TTM earnings have compressed
- FY2025 net income attributable exceeded operating income - earnings quality leans on non-operating/investment gains, not the core operation
- AI-server capacity race (Foxconn, Quanta, Wistron/Wiwynn, Supermicro, Dell, HPE) risks commoditizing the very growth engine bulls are paying for; the US line and automation raise capex while FCF sits near breakeven
What it is worth
Sanity-check via EV/sales and P/E on a low-margin EMS/ODM, adjusted for the AI-server mix-shift optionality. On ~$35B revenue and a ~$7B market cap, the stock trades at ~0.2x sales - appropriate for ~1% operating margins - but at ~19-20x TTM earnings after the Q1 2026 profit drop, so it is NOT cheap on an earnings basis. The debate is entirely about whether the AI-server/auto mix shift lifts through-cycle margin and earnings quality.
iPhone share erosion continues, the AI-server ramp commoditizes at near-zero incremental margin, FX/tariffs bite, and non-operating-dependent earnings quality keeps the multiple from re-rating even as trailing earnings stay depressed.
Core iPhone/PC stays flat-to-declining while servers/auto grow off a small base; revenue re-accelerates modestly but margins stay ~1%, keeping the stock range-bound as a low-EV/sales ODM with cash support.
AI-server ~10x ramp + auto scale prove durable and lift blended operating margin toward ~1.5-2%+, growing earnings so the elevated trailing P/E normalizes downward on a cheap, cash-rich, low-EV/sales base.
As a thin-margin assembler, Pegatron is valued on earnings power and mix, not sales multiples. Non-operating investment income inflates net income above operating profit, so screen on operating trends. The ~19-20x TTM P/E reflects depressed trailing earnings rather than a growth premium; a durable re-rate requires demonstrable margin expansion from AI servers/auto, not just revenue growth. Not financial advice.
SWOT
Strengths
- Entrenched top-tier Apple ODM/EMS relationship — hard-won qualification and trust across multiple product lines
- Global multi-country manufacturing footprint (China, Taiwan, Vietnam, US, plus a minority India stake) that fits customer China+1 de-risking
- Early, credentialed position in Nvidia's AI-server ecosystem (GB300/GB200/B200 rack integration)
- Solid liquidity — large gross cash balance and modest interest-bearing debt, plus a non-operating investment portfolio that cushions net income
Weaknesses
- Razor-thin margins (gross ~4%, operating ~1%) leave almost no error budget for cost or FX shocks
- Heavy customer concentration on Apple — has lost iPhone allocation to Foxconn and Luxshare over recent cycles (now #3)
- Revenue in multi-year decline off the 2022 peak — core consumer-device demand mature, and Q1 2026 profit fell >60% YoY
- AI-server value-add is a thin assembly slice of a GPU-cost-dominated bill of materials - revenue can ~10x without proportional profit
Opportunities
- AI-server buildout (hyperscaler + neocloud capex supercycle) as a genuine second growth engine
- US onshoring wave — a Texas server line positions Pegatron for 'made-in-USA' data-center demand and tariff insulation
- Automotive electronics secular content growth (Tesla and broader EV/ADAS)
- AI-PC refresh lifting notebook ASPs and volumes into 2027
Threats
- Intensifying AI-server competition for capacity from Foxconn, Quanta, Wistron/Wiwynn, Supermicro, Dell, HPE - commoditization risk
- Continued iPhone share loss to Foxconn (>50%) and China-based Luxshare (now #2)
- US-China tariff / export-control / supply-chain fragmentation raising cost and complexity
- Customer in-sourcing and pricing pressure; FX (NT$ strength) directly compressing already-thin margins
Moats, dependencies & bottlenecks
Moats
Multi-year certification, tooling, and trust make Pegatron hard to displace quickly - but Apple deliberately multi-sources and has shifted share to Foxconn/Luxshare, capping pricing power.
China/Vietnam/US footprint is expensive to replicate and matches China+1 demand; the durable edge is flexibility, not cost leadership. India control was ceded to Tata (60%) in Jan 2025.
Being a qualified rack integrator is valuable but widely shared among Taiwanese ODMs - a foothold, not an exclusive moat.
Component sourcing, logistics, and inventory turns are the real operating skill in razor-margin EMS; consistent execution is the durable asset.
Dependencies
Customer concentration Largest customer; iPhone/consumer-device orders drive the base. Share loss to Foxconn/Luxshare is the primary top-line risk.
Platform/supplier + demand engine AI-server growth is entirely tethered to Nvidia GPU platforms (GB300/GB200/B200) and allocation; also Pegatron's key component input for servers.
Server ramp depends on CSP (Microsoft/Amazon/Meta/Google) and neocloud (e.g., CoreWeave) data-center spend continuing.
Automotive customer Anchor client for the growing auto-electronics segment; EV demand cyclicality flows through.
Advanced-chip and memory/display availability gate build volumes; component cost mostly passes through but shortages hit utilization.
Geopolitical/regulatory Export controls, tariffs, and China+1 mandates dictate footprint economics; the US-line strategy is a hedge that also raises cost.
Advantages
- Deep, certified Apple/OEM relationships and proven high-volume assembly execution
- Diversified global footprint aligned with customer de-risking (Vietnam/US)
- Early credentialed entry into the Nvidia AI-server and Tesla automotive value chains
- Strong liquidity and low interest-bearing leverage to self-fund the mix-shift capex
Weaknesses
- ~1% operating margin with minimal cost-shock buffer
- Concentrated, share-losing iPhone franchise (now #3) in a mature market
- Earnings quality dependent on non-operating investment income
- AI-server value-add is thin and heavily contested - revenue growth may not translate to profit
Bottlenecks
- Structurally low gross/operating margin caps profit conversion from any revenue growth
- Nvidia GPU allocation and CSP capex timing gate the AI-server ramp regardless of Pegatron's capacity (component lead times may push some shipments to early 2027)
- Intense competition for finite AI-server assembly capacity and skilled labor (especially at the new US site)
- FX sensitivity (a stronger NT$ directly compresses thin margins)
- Customer multi-sourcing limits pricing power and locks in commodity-like returns
Top signals & trends
Top signals
Real, quantified pivot; watch for conversion to profit, not just revenue.
Off-season weakness in the core; weakest Q1 EPS in ~7 years.
Early evidence the server ramp is showing up in the topline.
Monetizes/de-risks India exposure but cedes control of a growth-geography asset.
Diversification away from iPhone dependence progressing.
Structural erosion of the profit base.
Trends
Primary bull driver; Nvidia rack demand from CSPs and neoclouds is Pegatron's second growth engine.
New Texas line positions Pegatron for domestic data-center and tariff-insulated demand.
Drives demand for Pegatron's diversified footprint but raises cost/complexity.
Mature smartphone base offset by a 2026-27 AI-PC notebook upgrade wave.
Secular per-vehicle content tailwind with Tesla anchor.
Many qualified rack integrators competing for the same AI-server work compresses value-add.
Ecosystem & competitor graph
Suppliers feed the company; customers pull from it. Line thickness shows the strength of each tie (supply-chain dependency, customer earnings contribution). Hover to isolate a tie.
GPU platforms (GB300/GB200/B200) that are the core content of Pegatron's AI-server racks.
Upstream advanced-chip foundry whose output gates system availability.
DRAM/NAND and other memory in servers, PCs, and phones - largely pass-through cost.
Largest customer; iPhones and other devices - the profit base.
Key automotive-electronics client anchoring the auto segment.
Hyperscale CSP buying AI-server racks; Amazon (AMZN), Meta (META), and Alphabet (GOOGL) are peer data-center buyers driving the same demand.
Named neocloud driver of Pegatron's ~10x 2026 AI-server guidance.
Largest iPhone assembler (>50% share) and the scale leader in AI servers; Pegatron's primary rival in both smartphones and servers.
China-based; surpassed Pegatron to become the #2 iPhone assembler, taking share directly from it.
Leading AI-server ODM with strong Nvidia CSP relationships - direct competitor for the server growth Pegatron is chasing.
Wistron (fellow ex-Apple ODM) and its Wiwynn hyperscale-server arm compete for the same AI-rack and cloud business.
US-listed AI-server specialist competing for hyperscaler/neocloud rack orders.
Notebook/PC and server ODM peers overlapping in AI-PC and general-server work.
US OEM system vendors that both compete for and partner around AI-server demand.