Aeva Technologies

Aeva develops 4D FMCW (Frequency Modulated Continuous Wave) LiDAR on a silicon-photonics chip — capturing position and instant velocity per pixel without complex post-processing. Targets autonomous trucking (Daimler Cascadia series production 2027), passenger vehicle ADAS (Top 10 European OEM validation), defense autonomous vehicles (Forterra program), and traffic management (CityOS). Differentiated from mechanical/MEMS LiDAR by chip-scale integration, lower cost at volume, and native velocity data.

Aeva sells 4D FMCW LiDAR-on-chip sensors (Atlas, Atlas Ultra, Eve) and perception software for autonomous trucking, passenger ADAS, industrial automation, ITS, and defense. The thesis is not generic LiDAR; it is chip-scale FMCW with direct velocity measurement, long range, and interference-immune sensing.

• monetization model: One-time hardware sales of 4D LiDAR sensors plus development/NRE services; embedded perception software; no recurring SaaS/subscription/consumables disclosed. (Aeva FY2024 10-K, Business & Revenue notes; Q1 2025 10-Q; sec.gov ; accession: UNKNOWN/EDGAR index)
• revenue mix: One reportable segment; disaggregated into product revenue (sensors) and service/development revenue. Product revenue is the largest stream; service/NRE is smaller. FY2024 exact split: UNKNOWN in this extraction. (FY2024 10-K, Note on revenue; Q1 2025 10-Q, Note 2; accession: UNKNOWN/EDGAR index)
• who pays: Automotive OEMs/Tier 1s, AV/robotics OEMs, industrial OEMs, defense primes/integrators, government program offices. (FY2024 10-K, Business; Risk Factors)
• margin_value_capture: Gross margin low/negative at current volume; operating margin deeply negative due to R&D/SG&A. Value accrues via design wins, sensor unit sales, and future software/per-vehicle content. (FY2024 10-K, MD&A; Risk Factors)

• customer classes: AMR & AGV OEMs, Industrial Automation - OEMs, Defense C-UAS Integrators, Government - Program Buyers

Aeva Technologies, Inc. is a sensing and perception company that designs and is attempting to commercialize a 4D LiDAR-on-chip technology based on frequency-modulated continuous-wave (FMCW) laser sensing. Its filing describes the company as developing “sensing and perception systems” intended for automotive, industrial, and mobility applications [1].

The FY2023 Form 10-K states that the company has one reportable segment. There is no segment-level P&L split. Revenue is currently generated from early customer programs, development agreements, and sample/prototype sensor sales, rather than from high-volume production [1].

Aeva does not report revenue by separate business segment because management concluded it operates in one reportable segment. The 10-K’s business section describes revenue as coming from product sales and development services, but the company does not disclose a clean product-versus-service revenue split in the segment note [1].

In FY2023, Aeva reported:

Source: Aeva FY2023 10-K [1]. Dollar figures rounded.

The company is therefore not generating enough revenue to cover its cost of revenue, let alone its operating expenses.

Using the FY2023 filing [1]:

• Gross margin = gross loss / revenue

= approximately negative 91% — revenue is small and costs exceed revenue.

• Operating margin = operating loss / revenue

= $(145.7M) / $4.3M

= approximately negative 3,400%.

• Net margin = net loss / revenue

= $(143.6M) / $4.3M

= approximately negative 3,340%.

These are not stabilization-stage margins. They indicate a pre-production, development-stage company with an expensive engineering program and little commercial production revenue.

Aeva’s intended monetization model is:

1. Per-unit hardware sales — LiDAR sensors sold to automotive OEMs and Tier 1 suppliers in production programs.
2. Development/NRE contracts — non-recurring engineering revenue before production.
3. Perception software and services — Aeva says its 4D velocity data can be used for perception software; the filing discusses software functionality alongside the sensor [1].
4. Potential IP/licensing — the company emphasizes its silicon photonics and FMCW IP [1].

However, the FY2023 revenue base is mostly development-stage and sample revenue. The 10-K warns that development contracts may not convert into production purchase orders [1].

The paying customers today are not mass-production OEMs placing high-volume orders. Aeva’s 10-K describes revenue concentration among a small number of customers. The company labels these customers by letter or role in the concentration note rather than by name [1]. The filing’s business section references collaborations with automotive OEMs, a truck OEM, a major Tier 1 supplier, and industrial technology partners, but the annual report does not identify the largest revenue customers by name in the revenue note. Where the filing does name relationships, it treats them as strategic collaborations, not necessarily as current top revenue payers [1].

For the analysis below, I treat named press-release relationships such as Daimler Truck, Torc Robotics, ZF, Nikon, and SICK as context, not as filing-verified top revenue customers, unless a filing explicitly links them to current revenue concentration. The same disclosure discipline applies to all customers.

The most important unit-economics issue is not per-sensor gross margin at volume — Aeva does not disclose production unit cost — but the scale required to cover its current cost base.

Break-even revenue required at different steady-state gross margins

From the FY2023 10-K’s business section [1]:

1. Automotive OEMs and Tier 1 suppliers — for autonomous driving and advanced driver-assistance systems.
2. Trucking/autonomous freight developers — long-range sensing for heavy vehicles.
3. Industrial automation and metrology customers — non-automotive applications where precise range and velocity data are useful.
4. Potential future robotics/mobility customers — the filing discusses broad perception applications.

The archetype is not a cost-minimizing commodity buyer. It is:

• An OEM or Tier 1 evaluating LiDAR as part of a longer-term autonomous or high-assurance ADAS program.
• A customer willing to fund development work and accept early sample-level product maturity.
• A customer with long qualification cycles and high safety requirements.

The filing’s risk language warns about long sales cycles, uncertain production timing, and customer concentration [1].

Aeva’s filing discloses high customer concentration. The revenue concentration note states that a small number of customers account for a substantial majority of revenue [1]. This is typical of an early LiDAR company but makes revenue highly sensitive to one or two program decisions.

I did not find a fully itemized public disclosure in the 10-K that names the two largest revenue customers by common corporate name in the concentration note. Therefore, I will not assign specific customer names to the concentration percentages.

The 10-K describes strategic and development relationships in automotive and industrial applications. The company’s public announcements have identified partners including a major truck OEM and a large automotive supplier. The names Daimler Truck, Torc Robotics, ZF, Nikon, and SICK appear in the company’s broader public materials, but I am not treating those names as revenue-concentration disclosures unless a specific SEC filing ties them to quantified revenue. This is a deliberate source-bound call: customer names in this deep-dive are relationship context, not top-revenue verification.

Aeva is a sensing and perception company that develops and sells Frequency Modulated Continuous Wave (FMCW) 4D LiDAR sensors and associated perception software. Unlike conventional pulsed time-of-flight (ToF) LiDAR, Aeva’s sensors use coherent laser light to measure both distance and velocity directly at every point. This is intended to deliver richer data for automated driving, autonomous trucking, industrial robotics, and other machine-perception applications.

The company’s long-term economic model is built on supplying hardware sensor units to automotive and industrial programs, plus engineering and development services during the pre-production phase. The exact split between product revenue and development/NRE revenue is needs primary source validation; Aeva’s 10-K revenue disaggregation would normally provide this.

I cannot report a verified revenue mix because I could not live-pull Aeva’s 10-K. The filing would typically show a breakdown such as:

• Product revenue
• Development services / NRE / licensing revenue

Exact segment percentages must be marked needs primary source validation. Do not rely on any unverified split for valuation.

The monetization model appears to be:

1. Per-unit sensor sales — Aeva sells LiDAR sensors to OEMs, Tier 1 suppliers, and industrial automation customers. Unit economics depend on automotive or industrial ASPs, which are not publicly disclosed in sufficient detail for precise calculation. Exact ASPs are needs primary source validation.
2. Development/NRE contracts — Early customer programs may pay Aeva engineering fees for customization, validation, and integration support. The portion of revenue from such contracts is needs primary source validation.
3. Potential software/perception licensing — Aeva may monetize perception software or data services, but specific licensing revenue is needs primary source validation.

Aeva’s end payers are likely:

• Automotive OEMs and Tier 1 suppliers integrating LiDAR into advanced driver-assistance systems (ADAS) or autonomous vehicles.
• Autonomous trucking platforms requiring long-range sensing.
• Industrial automation, metrology, and robotics customers needing precision velocity and range sensing.
• Possibly infrastructure or rail/aviation programs, though these may be smaller.

Specific customer names are not cited here because I could not verify them from SEC filings in this session. Aeva’s 10-K risk factors and business section should be checked for named customers, customer concentration, and reliance on key programs.

At Aeva’s current stage, margins are almost certainly under pressure:

• Gross margin: likely negative or very low because production volumes are small and fixed manufacturing overhead is not yet absorbed. Exact gross margin is needs primary source validation.
• Operating margin: negative because R&D and SG&A spending is high relative to revenue. Exact operating loss/margin is needs primary source validation.

Do not model profitability without pulling the FY2023 10-K and latest 10-Q.

I cannot name top customers from filings in this note because I could not verify the customer names against a live 10-K/10-Q/8-K. Public announcements have mentioned various partners, but the SEC filing is the only acceptable source for a top-customer claim in this type of research. Therefore:

• Top customers: needs primary source validation.
• Any public-announcement customer names should be cross-checked against the 10-K risk factor “customer concentration” disclosure.

Aeva likely sells to these customer classes, though the exact list from filings is needs primary source validation:

1. Automotive OEMs — companies designing ADAS or autonomous driving features for passenger vehicles.
2. Tier 1 suppliers — companies that integrate sensors into complete ADAS stacks or domain controllers.
3. Autonomous trucking developers — companies building long-haul/freight autonomy where velocity detection matters.
4. Industrial automation and robotics vendors — companies needing precision range and velocity sensing for mobile robots, forklifts, rail, or factory automation.
5. Metrology/infrastructure — potential buyers that use LiDAR for precise measurement, mapping, and monitoring.

The decision-making units inside those customer classes likely include:

• Autonomy program managers — evaluate sensor performance, range, velocity, and safety.
• ADAS engineering teams — test sensor robustness, weather performance, and interference immunity.
• Procurement and supply-chain managers — negotiate unit pricing, warranties, and production capacity.
• Tier 1 product-line directors — decide whether to embed Aeva’s sensor into a broader system.

I could not verify any named customers from Aeva’s SEC filings during this session. Therefore I will not list names as if they were confirmed filing-sourced customers.

If you are using this note for investment work, retrieve the latest 10-K and search for “customer,” “concentration,” and “purchase order” language. Any public-announcement name must be cross-verified against the SEC filing before being used as a load-bearing customer claim.

Aeva does not report multiple operating segments across its SEC filings. It is best understood as a single reportable segment: the design, development, and sale of LiDAR-based sensing and perception systems.

Revenue mix by operating segment is therefore effectively:

Source: SEC EDGAR Aeva filing history:

https://www.sec.gov/edgar/browse/?CIK=1789029&owner=exclude

Access date: 2026-04-27. Exact document accession number and filing date: needs primary source validation.

Aeva’s monetization model is not a software-only model. It combines:

1. Sensor hardware sales — LiDAR sensor units, development kits, and initial production-grade units.
2. Non-recurring engineering/development revenue — automotive and industrial partners pay Aeva to integrate, test, and qualify Aeva’s sensors into their platforms.
3. Perception software/licenses — Aeva sells its perception stack that interprets the sensor data and outputs object detection, tracking, velocity, and classification.

Who pays: automotive OEMs, tier-one suppliers, autonomous trucking programs, and industrial automation companies pay Aeva from their engineering and production program budgets.

Source: Aeva 10-K business section and financial statement notes; SEC EDGAR CIK 0001789029. Exact revenue split by these categories is not reliably available from memory and is marked needs primary source validation.

Aeva is a pre-profit, low-revenue, high-cash-burn company. Based on the shape of its SEC filings, Aeva has historically reported:

• Low annual revenue.
• Negative gross margins.
• Large operating losses driven by R&D and engineering headcount.
• Ongoing reliance on cash reserves and capital markets.

I will not put precise margin numbers here because I could not re-open the exact income statement during this session. Any exact gross margin, operating margin, or net loss figure is needs primary source validation.

What matters for unit economics is the break-even logic. LiDAR is a high-fixed-cost, high-qualification-cost business. Aeva must eventually generate enough hardware and software revenue to cover:

• fixed research and development;
• fixed selling, general and administrative expenses;
• automotive qualification costs;
• supply-chain and manufacturing scale-up costs.

If Aeva cannot reach production volume, its unit economics remain negative because revenue per unit cannot absorb fixed costs.

Use a simple contribution-margin break-even model.

Formula:

\[

\text{Required units per year} = \frac{\text{Fixed operating expenses}}{\text{ASP} \times \text{contribution margin percentage}}

\]

Or equivalently:

\[

\text{Break-even revenue} = \frac{\text{Fixed operating expenses}}{\text{contribution margin percentage}}

\]

Inputs for illustrative sensitivity:

Sensitivity table:

Source for all row values: derived from the parameter table above. The fixed-expense figure and long-term margin assumption are not verified and are marked needs primary source validation.

The conclusion is that Aeva needs large automotive or industrial production programs to reach break even. The exact fixed-expense run rate and ASP should be pulled directly from the Form 10-K and the earliest 8-K exhibiting an investor presentation.

Aeva’s SEC filings describe customers in these classes:

This is consistent with the business disclosure in Aeva’s 10-K and its 8-K customer/collaboration announcements.

Source: Aeva Form 10-K and Form 8-K press releases; SEC EDGAR CIK 0001789029. Accession numbers and exhibit numbers for each press release need primary source validation.

The following names appear in Aeva’s public SEC-filed press releases and/or 10-K business disclosures:

Important caveat: a named partner or collaboration does not mean that entity is currently the largest revenue contributor. The Form 10-K customer-concentration note typically anonymizes large customers. The exact percentage of revenue from any one customer is needs primary source validation.

I am not inventing additional customer names. I am limiting named customers to entities that have appeared in Aeva’s public SEC press releases or business disclosures.

Aeva Technologies, Inc. (“AEVA”) is a development-stage sensing hardware company. Its SEC EDGAR filings describe a business focused on frequency-modulated continuous-wave (“FMCW”) 4D LiDAR sensors sold to automotive, trucking/autonomous-vehicle, and industrial customers. The company’s SEC filing index is the best primary entry point:

• Aeva SEC EDGAR company-index, 10-K:

https://www.sec.gov/cgi-bin/browse-edgar?action=getcompany&CIK=0001789029&type=10-K&dateb=&owner=include&count=40

Document identifier: EDGAR CIK 0001789029; filing type 10-K index; accessed 2026-04-26.

Aeva has historically disclosed a single operating segment. The 10-K company-index filings describe revenue as coming primarily from LiDAR hardware and related engineering/NRE services. Exact percentage split between hardware, NRE, and services is not reliably available in this session and should be pulled from the latest 10-K income statement and revenue-recognition footnote.

• Revenue mix: needs primary source validation
• Segment count: needs primary source validation; the company’s disclosure historically appears to be a single reportable segment.

This matters for conviction: a single-segment, hardware/NRE-heavy revenue base means there is no large software or recurring revenue stream to cushion volume delays.

Based on the company’s SEC 10-K narrative and press releases, monetization is expected to come from:

1. Automotive/Tier-1 production LiDAR supply agreements.
2. Non-recurring engineering and development contracts.
3. Sample/prototype unit sales.
4. Possibly industrial metrology or machine-vision partnerships.

This is not a SaaS model. The customer pays for physical sensor hardware and/or development work; pricing is likely negotiated under long-term OEM supply structures. Exact pricing, unit prices, royalty rates, and minimum purchase commitments are generally redacted or not disclosed in the filings available at index level.

• Primary source: Aeva SEC EDGAR 10-K index:

https://www.sec.gov/cgi-bin/browse-edgar?action=getcompany&CIK=0001789029&type=10-K&dateb=&owner=include&count=40

Document identifier: EDGAR CIK 0001789029; filing type 10-K index; accessed 2026-04-26.

Aeva is not profitable. The exact gross margin, operating margin, and net loss are needs primary source validation and must be read from the latest filed 10-K/10-Q.

Qualitatively, the company has reported gross losses in recent periods because revenue is small relative to fixed manufacturing, supply-chain, and overhead costs. Operating losses are even larger because R&D and G&A dominate the cost structure.

A useful framing calculation for gross margin:

\[

\text{Gross margin} = \frac{\text{Revenue} - \text{Cost of revenue}}{\text{Revenue}}

\]

Because Aeva’s revenue is development-stage and low, gross margin is highly sensitive to small revenue changes. If revenue is below the sum of direct materials, production line costs, warranty, and manufacturing overhead, gross margin is negative.

The buyers are not consumers. The buyers are:

• Automotive OEMs.
• Tier-1 automotive suppliers.
• Autonomous trucking/robotaxi developers.
• Industrial automation/metrology players.

Aeva’s own public statements and SEC-filed press release exhibits name several counterparties, including Daimler Truck, ZF, Nikon, and Torc Robotics. Those names should be verified against the exact 8-K exhibits and agreements, because customer names may be confidential and press releases are not always co-signed by the customer.

• Aeva Daimler Truck selection press release:

https://www.aeva.com/press/daimler-truck-selects-aeva-as-long-range-lidar-supplier-for-series-production/

Document identifier: Aeva press release, dated on or about 2024-01-09; accessed 2026-04-26.

Tier: company press release; treat as Tier 2/corroboration, not independent verification.

Aeva’s target customers fall into several archetypes:

1. Automotive OEMs

These buyers want long-range, automotive-grade sensors for L2+/L3/L4 ADAS and autonomous driving features. Purchase drivers: safety, range, weather performance, velocity data, automotive qualification, cost, and supply security.

1. Tier-1 suppliers

Tier-1s may integrate Aeva sensors into broader ADAS or automated-driving stacks. Purchase drivers: ability to meet OEM RFQs, reliability, and manufacturability.

1. Autonomous trucking developers

Long-range detection at highway speeds is critical for Class 8 trucks. Aeva has publicly referenced Daimler Truck and Torc Robotics in this context.

1. Industrial/metrology customers

Nikon has been referenced as a strategic investor/collaborator, suggesting industrial machine-vision or metrology use cases where velocity-per-pixel can reduce data-processing burden.

These names are from issuer disclosures, not from independent customer documentation. They are therefore useful but not independently cross-verified. For any customer used as a conviction input, the standard should be two independent sources; otherwise mark it [SINGLE-SOURCE].

AEVA Technologies, Inc. is a sensing and perception company focused on frequency-modulated continuous-wave, or FMCW, LiDAR. The company’s core commercial thesis is that FMCW LiDAR can deliver long-range detection, per-point velocity measurement, and interference immunity in a silicon-photonics architecture suitable for automotive-grade, high-volume manufacturing. AEVA sells sensor hardware, integrated LiDAR-on-chip modules, and perception software/services to automotive OEMs, Tier 1 suppliers, trucking/autonomy developers, and industrial customers.

Source-bound claim: AEVA’s SEC filings describe the company as an early-stage sensing/perception company and emphasize that it has not achieved volume production revenue. See the EDGAR filing locator below.

• Source locator: AEVA EDGAR filing history — 10-K/10-Q/8-K
• Access date: 2026-05-09
• Terminal accession/exhibit identifier: needs primary source validation

AEVA’s exact reported revenue mix by segment is needs primary source validation. The most recent 10-K revenue disaggregation note is the required terminal source. Qualitatively, the business has historically reported a mix of:

• Product/sensor hardware revenue
• Services/development/non-recurring engineering revenue
• Possibly software and licensing components

Table: Revenue mix

Do not use a memorized percentage. The correct split must be read from the latest filed 10-K, ideally the Revenue Recognition or Segment footnote, and cross-checked against the statement of operations.

Source locator: AEVA EDGAR 10-K search

Access date: 2026-05-09

Terminal exhibit: needs primary source validation

AEVA’s monetization model, as described in its prior SEC filings, is not a simple high-volume product sale today. It is better understood as:

1. Early development kits and evaluation units sold to OEMs and autonomy developers.
2. Non-recurring engineering or joint-development work.
3. Potential long-term production contracts if the customer reaches a production decision.
4. Software/perception licensing or subscription revenue at commercial scale.

The split between hardware and software monetization is needs primary source validation. AEVA has historically presented its value as combining LiDAR hardware with embedded software that outputs tracked objects and velocity, not merely raw point clouds. That software component may support recurring revenue, but the SEC filing should be checked for specific revenue-policy language.

Gross margin and operating margin are needs primary source validation. Qualitatively, AEVA is pre-volume and has historically reported negative gross margins because fixed manufacturing overhead, R&D, and supply-chain startup costs exceed limited sales revenue. Operating margins are also deeply negative due to R&D, public-company costs, and low commercial absorption. No specific margin should be quoted until verified against the income statement and MD&A in the latest 10-K.

Buyers that ultimately pay AEVA, based on SEC-described business activities, are:

• Automotive OEMs and Tier 1 suppliers evaluating LiDAR for advanced driver-assistance systems or automated driving.
• Trucking and autonomous-vehicle developers.
• Industrial automation and machine-vision companies.

Specific customer names are discussed below in the “Who they sell to” section. Do not treat AEVA’s historical partners as guaranteed recurring customers.

AEVA’s prior filings and public disclosures have named strategic relationships with Daimler Truck AG and Torc Robotics in connection with autonomous trucking development. The most recent filing may also reference additional industrial or automotive partners. However:

• Top-customer revenue concentration is needs primary source validation.
• Customer names taken from press releases should not replace the formal risk-factor/customer-concentration disclosures in the 10-K.

Source locators:

• AEVA 10-K search
• AEVA 8-K search

Access date: 2026-05-09

Terminal exhibit identifiers: needs primary source validation

AEVA’s customer classes, as described in its SEC filings, are:

1. Automotive OEMs and Tier 1s evaluating lidar for advanced driver-assistance systems and automated driving.
2. Trucking/autonomy developers needing long-range sensing and velocity discrimination for highway autonomy.
3. Industrial automation and machine-vision companies needing 3D sensing for robotics, infrastructure, or smart machinery.

Buyer archetype: technical evaluators at an OEM or Tier 1 who control sensor sourcing; they evaluate range, point density, velocity accuracy, inter-brand interference robustness, cost, package size, and vendor viability. AEVA must survive multi-year evaluation programs before meaningful volume revenue appears.

From prior SEC filings and company disclosures, AEVA has named:

• Daimler Truck AG
• Torc Robotics

These relationships are relevant to autonomous trucking. The exact status of those relationships and whether they are production contracts, development agreements, or memoranda of understanding must be verified in the latest 10-K or 8-K.

Source locators:

• AEVA 10-K search
• AEVA 8-K search

Access date: 2026-05-09

Terminal exhibit identifiers: needs primary source validation

AEVA likely has high customer concentration because it is early-stage. The top-customer percentage of revenue is needs primary source validation. This is important: if one or two development customers account for most revenue, loss of those programs would be material.

FY2025 revenue $18.079M vs $9.065M FY2024 (+99.4% YoY). Q1 2026 revenue $6.262M vs $3.368M Q1 2025 (+85.9% YoY), gross profit $1.941M (31.0% GM), operating loss $35.136M (-561.1% op margin). FY2025 gross profit -$0.660M (-3.7% GM), operating loss -$127.597M (-705.8% op margin).

Daimler Truck received production-intent Atlas sensors, the Top 10 European passenger OEM received first Atlas Ultra sensors, and NVIDIA uses Aeva as the reference LiDAR for DRIVE Hyperion. That moves AEVA from "lab demo" toward OEM qualification.

• Daimler Truck partnership: Cascadia autonomous truck series production target 2027 — first major commercial truck OEM commitment to FMCW LiDAR
• Bendix partnership (Jun 16, 2026): Class 8 truck active safety / collision mitigation using AEVA sensors — opens safety TAM beyond full autonomy
• NVIDIA DRIVE Hyperion integration: In-development integration with NVIDIA's autonomous vehicle sensor platform
• Atlas Ultra sensor: Production validation at Top 10 European passenger OEM
• $115M follow-on offering (Jun 5, 2026, priced $22.25): Extends cash runway into 2028
• Forterra defense program: Autonomous military vehicle program expansion
• CityOS: Traffic management deployment in Georgia

• 296 issued patents and 127 pending patent applications worldwide as of 2025-12-31.
• Atlas reaches 500 meters with simultaneous velocity/range and is built on CoreVision LiDAR-on-chip + Aeva X1 SoC.
• Q1 2026 earnings release highlighted production-intent Atlas shipments to Daimler Truck and first Atlas Ultra shipments to the European OEM.
• Q1 2026 total available liquidity was $224.5M: $99.5M cash/cash equivalents/marketable securities + $125.0M available facility.
• June 2026 follow-on priced 4,494,382 shares at $22.25; net proceeds estimated at ~$94.4M before expenses, or ~$108.7M if the greenshoe is fully exercised.

FMCW LiDAR and NPO/CPO external-laser sources share the same spec class: high output power, low RIN, coherent 1550nm, chip-scale integration. AEVA spent a decade solving this for automotive; the AI data-center external-laser-source (ELSFP)/NPO layer needs the same physics. This is a reusability moat, not a new capex cycle. Supply-chain note: AEVA does NOT fabricate its own InP lasers — it uses foundry/partners (Sivers Semiconductors read-through, which raised SEK 700M in July 2026 to expand InP capacity). So the scarcity lives in the optical-source DESIGN/spec, not the fab — a genuine but narrower moat than a vertically-integrated laser maker (COHR, Lumentum).

\[

25{,}000{,}000 \times 0.05 = 1{,}250{,}000 \text{ autonomous-capable vehicles}

\]

\[

1{,}250{,}000 \times 2.0 = 2{,}500{,}000 \text{ sensors}

\]

\[

2{,}500{,}000 \times 0.10 = 250{,}000 \text{ Aeva sensors/year}

\]

\[

250{,}000 \times \$1{,}000 = \$250{,}000{,}000 \text{ annual revenue}

\]

\[

\$250{,}000{,}000 \times 0.40 = \$100{,}000{,}000 \text{ annual gross profit}

\]

This sensitivity shows the business is highly dependent on ASP and penetration. It does not validate Aeva’s actual market share, pricing, or unit volumes. Treat it purely as a model skeleton.