Advanced Energy Industries

AEIS designs and manufactures highly engineered precision power conversion, measurement and control systems. It does not make the chip tools or the servers — it makes the critical power sub-system inside them: RF/microwave plasma generators + impedance-matching networks that ignite and control the plasma in every etch/deposition chamber, and (new) 48V/800V DC-DC converters that feed the GPUs in AI server racks.

• Revenue: $574.1M, +30.0% YoY (vs $441.5M) and +12.3% QoQ (vs $511.0M) — record, above high end of guidance.
• Margins: GAAP GM 41.1% (non-GAAP 41.9%, +380 bps YoY); GAAP op income $95.1M (16.6%); non-GAAP op margin 21.9%.
• EPS: GAAP $1.29 (incl. $31.8M / $0.75 inducement cost from partial 2028 convertible-note conversion); non-GAAP $2.74 (+83% YoY).
• Cash flow: record $86M from continuing operations; $4.1M quarterly dividend.
• Balance sheet: cash & equivalents $1.397B; long-term debt $1.264B → net cash positive (~$0.13B before current debt).
• R&D: $65.0M/qtr (~11.3% of revenue), rising YoY.
• Guidance (raised): Q3 FY26 revenue $640M ± $20M, GAAP EPS $2.38, non-GAAP $3.00; FY26 revenue growth raised to low-to-mid 30% (from low-mid 20%); Data Center Computing expected >50% FY26; Semiconductor H2 ~+50% YoY; Q4 GM ~42% (line of sight >43%); FY27 growth >20%.

• monetization model: One-time hardware sales (power supplies/controllers); limited spare-parts/repair/service. No SaaS/subscription. (10-K FY2024, Item 1 Business; EDGAR CIK 0000927003, sec.gov ; Accession No. UNKNOWN)
• revenue mix: Semiconductor Equipment ~69%; Industrial & Medical ~31% (FY2024 segment note). (10-K FY2024, segment note; same EDGAR source)
• who pays: OEM design/procurement teams; data-center/telecom operators/ODMs; medical/industrial OEMs. (10-K FY2024, Item 1)
• margin_value_capture: FY2024 gross margin ~36%, operating margin ~10%; value accrues in design-win content per tool/system and aftermarket. (10-K FY2024, MD&A/consolidated statements)

• customer classes: semiconductor wafer-fab equipment OEMs (serving Semiconductor Fabs - IDM Buyers), Data Center Operators, Hyperscalers, Industrial Automation - OEMs, Industrial Infrastructure Owners, medical device OEMs, telecom equipment OEMs. (10-K FY2024, Item 1; customer concentration note)

AEIS is primarily a capital-equipment component and subsystem supplier. It earns revenue by selling hardware that controls and converts electrical power, and by selling aftermarket services, spares, repairs, refurbishment, and upgrades around that installed base.

The business is better understood as two linked economic models:

1. Semiconductor equipment power/controls — high-precision, application-qualified products sold into wafer fabrication equipment OEMs. Purchase decisions are driven by process performance, particle reduction, uptime, process repeatability, and speed of matching to changing plasma conditions. This tends to carry higher engineering content and more qualification lock-in than general-purpose power supply businesses.

1. Industrial/medical/embedded power and other precision products — broader power conversion modules and systems sold into industrial, medical, data center/telecom, and specialty applications. This part of the portfolio may face more standard industrial purchasing behavior and more price competition, depending on product line and whether the product is standardized or application-specific.

Revenue mix percentages are not reproduced here. To comply with the no-fabrication rule, I did not guess segment percentages. A production deep-dive should extract the following from the most recent 10-K segment note and the MD&A:

• Semiconductor equipment segment revenue
• Industrial and medical segment revenue
• Any “other” or corporate/unallocated revenue
• Geographic revenue split, if disclosed
• Product vs. service revenue split, if disclosed

All of the above: needs primary source validation.

The monetization model is hardware-centric with an aftermarket tail:

• OEM product sales: AEIS sells power supplies, RF generators, matching networks, remote plasma sources, thermal metrology, and related subsystems to equipment manufacturers.
• Aftermarket/service: AEIS sells spare parts, repair services, field service, upgrades, and refurbishment for its installed base. In semiconductor applications, process tools operate in harsh plasma and thermal environments, so service and replacement demand can recur over the life of the tool.
• Long-life and qualification-driven demand: If a power delivery product is designed into a semiconductor OEM’s process tool, changing suppliers can require requalification. That can create switching costs and recurring service demand, although it does not eliminate competition at the next tool platform design cycle.
• Potential pricing model variations: The company may sell through direct sales, distribution, or OEM long-term supply agreements. The exact split and contract terms need primary-source validation.

Not reproduced. Gross margin and operating margin are load-bearing financial figures and must be verified against the income statement and segment disclosures in the most recent 10-K/10-Q. I will not assert a margin from memory. Gross margin: needs primary source validation. Operating margin: needs primary source validation.

AEIS’s unit economics vary by product line. A semiconductor RF power system has different economics than an embedded AC/DC power module. A compliant deep-dive would build separate unit-economic models for:

1. Semiconductor RF power delivery: high ASP, lower unit volume, application engineering, service tail.
2. Industrial/medical precision power: mid ASP, broader customer base, more SKUs.
3. Embedded power modules: lower ASP, higher volume, potentially commoditized, more price-sensitive.

Because I could not verify AEIS-specific ASPs, gross margins, or service attach rates, I provide only an illustrative non-AEIS calculation in Section 6. That calculation uses clearly labeled assumed inputs and is not evidence about AEIS. It is provided only to show the required analytical form.

Buyers are typically not end fabs directly in the semiconductor part of the business; they are OEMs that build wafer processing equipment. In industrial, medical, and data center applications, buyers may be equipment builders, system integrators, distributors, or, in some cases, end users purchasing aftermarket service.

Top customers are not named here. The prompt requires that top customers be named from filings only. I could not retrieve AEIS’s customer-concentration disclosures from a 10-K in this session. Therefore I will not assert that any specific customer is a top customer. Top customer names and concentration percentages: needs primary source validation.

Customer classes are broadly:

• Semiconductor wafer fabrication equipment OEMs: the principal buyers of RF delivery, matching, remote plasma, and related products used in etch, deposition, strip, implant, and process control tools.
• Semiconductor device manufacturers/fabs: primarily indirect customers; may buy spares/services or influence new tool configurations.
• Industrial equipment OEMs: buyers of high-voltage, temperature, flow, and process power products used in thin-film coating, glass, solar, analytical instruments, and other industrial processes.
• Medical and scientific equipment OEMs: buyers of precise, low-noise power and temperature/flow instrumentation.
• Data center, telecom, and computing equipment makers: buyers of embedded power conversion modules and systems.
• Distributors and service partners: channel in some product lines.

The buyer archetype in semiconductor power is an equipment OEM process engineer or supply chain manager who cares about:

• Process matching and repeatability.
• Tool uptime and mean time between failures.
• Global service and spare-parts availability.
• Roadmap alignment for advanced nodes and new chamber processes.
• Total cost of ownership, not just ASP.

In embedded power, the archetype may be a hardware/systems engineer or procurement manager focused on:

• Electrical specifications and form factor.
• Reliability and regulatory approvals.
• Unit price at volume.
• Lead times and supply assurance.

No customer names are asserted. Customer concentration and named customers must be taken from the customer concentration note and revenue discussions in AEIS’s 10-K. I do not have document-level access in this session. Named customers: needs primary source validation.

Precision power conversion systems for semiconductor manufacturing equipment — supplies the power delivery that enables advanced chip fabrication. $1.73B rev, P/E 25.

Q2 2026 earnings ~early Aug (mid-30% YoY revenue growth guide, Data Center Computing doubled YoY to $194M); hyperscaler AI power infrastructure demand structural tailwind; potential further guidance raises on sustained DC capex

1) DC-DC efficiency economics (the data-center adjacency). ADH converts at 98.2% peak (vs a ~94% typical 400V AC chain) and Artemis at 96% (vs 94%). A +2 pp gain on a 100 MW AI data center ≈ 2 MW of avoided continuous loss → ~$1.7M/yr electricity at $0.10/kWh. This is the measurable value of the 800V-DC transition AEIS is riding.

2) Margin bridge — capture proof (10-Q + EX-99.1). GAAP gross margin 41.1% ($236.1M / $574.1M) vs 37.0% a year ago ($163.4M / $441.5M) = ~+410 bps GAAP YoY; non-GAAP GM 41.9% (+380 bps YoY, ex-tariff-refund 40.7%). GAAP operating income $95.1M = 16.6% margin; non-GAAP OM 21.9% (record). Incremental net margin: revenue +$132.6M YoY vs non-GAAP net income +$55M ($112M vs $57M) ≈ ~41% incremental non-GAAP net margin — clean evidence the bottleneck economics flow to the P&L, not a commodity pass-through.

3) Segment mix / concentration. Q2 FY26 (earnings call): Semiconductor $278M (48.4%), +33% YoY; Data Center Computing $192M (33.4%), +35% YoY; Industrial & Medical $80M (13.9%); Telecom & Networking $24M (4.2%). Top-3 customers ≈ 54% of FY2025 revenue (AMAT ~23%, a hyperscaler ~19%, LAM ~12% — inferred, marginofinsight). Concentration cuts both ways: it is the pricing-power source in semis and the buyer-power risk in data center.

4) Switching-cost / qualification floor. 12–24 month qualification × millions in engineering × yield-fluctuation risk = the dollar floor a customer must cross to displace AEIS on an in-production tool. This is the "cost of the moat" for the buyer (vs the HPE-style acquisition-cost frame for the seller).

No AEIS-specific parameter in this table is sourced because I could not retrieve document-level data. In a compliance-complete run, the source column would carry exact 10-K/10-Q URLs, accession numbers, exhibit identifiers, and filing dates.