Aeva Technologies, Inc.
Real FMCW programs, engineering-heavy revenue, and a Sivers link nobody has confirmed

Research date: 23 October 2026. Company: Aeva Technologies, Inc. Listing: Nasdaq, AEVA. ISIN: US00835Q2021 (CUSIP 00835Q202 since the 1-for-5 reverse split of 18 March 2024). Reporting currency: USD (US GAAP). Latest financial period reviewed: quarter ended 30 June 2026 (10-Q, August 2026). Material announcements reviewed through 23 October 2026. Capture window 22 October 19:00 to 23 October 09:05 Europe/Luxembourg.

Tags used throughout: confirmed = company filing, release or primary source; est. = our arithmetic from confirmed inputs; speculated = sell-side, vendor report or social post. Every figure carries the source's own date; broker rows carry a capture timestamp as well.

Idea origination: why we are researching Serenity's AEVA thesis

This idea came onto our research agenda through Serenity (@aleabitoreddit), the same account whose Sivers and X-FAB theses we examined. Aeva first appears in his posts in February 2026 and became one of his most-mentioned names again on 20 October 2026, the day Sivers Semiconductors announced LiDAR production orders worth USD 53 to 138M over the program lifecycle (CW DFB lasers and optical amplifiers, shipments from Q4 2026) for a customer it did not name. Serenity's reading is that the customer is Aeva and that FMCW LiDAR for trucks, cars and physical-AI machines is the next volume market for those lasers. Mention counts and dates come from independent archives (Serenity Stock Tracker, Serenity Tracker universe); x.com returns 402 to fetch tools, so post text is secondary. Serenity gets credit for the idea; filings set the baseline; this memo supplies the evidence checks. We do not assert an audited performance record or his current position.

Investment memo: our current view

Program watchlist; the price already sits on our base case. Aeva has real customers: Daimler Truck and Torc for series-production autonomous trucks, a top-10 European passenger-car maker with start of production in 2028, Bendix, SICK and Nikon, with Jabil, Fabrinet and LG Innotek building the hardware. What it does not yet have is product revenue at scale. In the first half of 2026, 60% of revenue was engineering services (USD 7.4M of 12.4M), and product revenue fell 39% year on year in Q2 to USD 2.5M. The company burned USD 57M of operating cash in the half and still loses about USD 35M a quarter at the operating line.

The Sivers link that brought the stock back into focus is speculated. No Aeva filing, release or proxy mentions Sivers, Sivers names no customer, and Aeva integrates its optics on its own silicon-photonics module assembled by Fabrinet. Our money ladder also shows a scale mismatch: a full Daimler-scale truck program would bring a laser supplier only USD 1.5 to 6M a year, well short of a USD 53 to 138M order unless volumes or content per sensor are much larger than one truck program.

Governance needs watching. Founders and the departing CFO sold about USD 20M of stock between June and September 2026 under 10b5-1 plans adopted three months before the USD 22.25 follow-on offering; there were no insider buys in 38 Form 4 filings since November 2025. The CFO left on 5 September 2026, a month after the raise. A short report (Fugazi Research, 21 August 2026) argued the valuation rests on adjacency; we could not retrieve a primary copy.

At USD 9.88 (22 October 2026 close) the enterprise value is about USD 611M, or 19× 2026 consensus sales. Our scenario values today are USD 1.7 bust, USD 9.2 base, USD 28.3 boom (15% discount rate, four years), so the market prices roughly the base case. External anchors sit above the price: Morningstar quantitative 3 stars with a fair value of USD 14.10 and "consider looking elsewhere"; LSEG consensus Buy on 5 analysts with a mean target of USD 29, which equals our boom case. The balance sheet is adequate for about two years at the current burn, with USD 177.9M of cash and securities, USD 100M of convertible notes and an undrawn USD 125M Sylebra facility that expires on 8 November 2026.

The next hard test is the Q3 2026 report on 4 November 2026: revenue against the USD 17.6 to 23.6M the second half needs to hit the USD 30 to 36M full-year range (given in February, not repeated in August), the product versus services mix, a Daimler series-production date, the permanent CFO, and the Sylebra facility decision.

Thesis underwriting: from source idea to investable conclusion

Listing and timestamps

What the business actually sells

Aeva makes frequency-modulated continuous-wave (FMCW) LiDAR. Each point carries distance plus instant radial velocity ("4D"). The core is Aeva's own silicon-photonics chip, the "LiDAR-on-chip", which combines the optical transmit and receive paths. Assembly is outsourced. (confirmed, 10-K 2025 · 22 Oct 19:04)

Customers and partners named in filings or releases: Daimler Truck with Torc (series-production autonomous trucks, selected Jan 2024), a top-10 European passenger OEM, Bendix (Class 8 ADAS, Aug 2026), SICK, Nikon, NVIDIA DRIVE Hyperion, Airbus and defence work, LG Innotek (investor and manufacturer), Jabil and Fabrinet (manufacturing). Three customers were 64% of 2025 revenue. (confirmed, 10-K 2025 and Q2 release · 22 Oct 19:20)

Sivers link. Sivers' 11 March and 20 October 2026 releases name only a "strategic LiDAR customer". No Aeva filing names a laser supplier, and Aeva integrates its optics on its own silicon-photonics module assembled by Fabrinet. The Sivers-to-Aeva link is speculated (Serenity; RevoAIution Substack 12 Aug 2026 · 22 Oct 19:20). A CW DFB laser plus optical amplifier is a plausible external light source for an FMCW chip, but other FMCW makers (Aurora in-house, SiLC, Scantinel) also exist.

Where the laser sits: one truck program and the money that reaches Aeva

One reference program: Daimler Truck autonomous Class 8 fleet, 50,000 trucks a year at maturity. This is an illustrative volume, not company guidance; Daimler has given no SOP date or volume (est.).

Money ladder for that program, per year at maturity (all est.):

1. Trucks: 50,000 × USD 275k ≈ USD 13.8B fleet spend.
2. Autonomy hardware: 50,000 × ~USD 100k ≈ USD 5B.
3. LiDAR: 150,000 sensors × USD 750 ≈ USD 112M to Aeva.
4. Light source: 150,000 × USD 10 to 40 ≈ USD 1.5 to 6M to the laser supplier.

Closing comparison. One Daimler-scale program fully won ≈ USD 112M a year ≈ 6 years of 2025 sales (USD 18.1M) per year of production. The same program gives a laser supplier only USD 1.5 to 6M a year, so Sivers' USD 53 to 138M lifecycle order implies far more sensors or more content per unit than one truck program. That is another reason not to assume the Sivers customer is Aeva alone. (est.)

The financial starting point

Source: Q2 2026 release (8-K ex-99.1, 5 Aug 2026) and 10-Q Q2 2026; Q1 2026 derived as H1 minus Q2. Captured 23 Oct 08:50.

Balance sheet 30 Jun 2026 (confirmed, 10-Q): cash, equivalents and marketable securities USD 177.9M (marketable securities 134.7M); convertible senior notes USD 100M (Apollo, 4.375%, due 2032, conversion price USD 15.86); warrant liabilities USD 74.0M; total liabilities USD 200.3M. Undrawn Sylebra standby preferred facility USD 125M, expiring 8 Nov 2026. 2026 revenue guidance USD 30 to 36M, of which 12.4M delivered in H1.

Where Aeva stands today: inputs reused in the sections below

Competitors and market share (23 October 2026)

Automotive and industrial LiDAR sensor makers

Reading: the volume market is Chinese ToF at USD 200 to 300 a unit. Aeva's merchant FMCW peers are pre-revenue start-ups, and the two large FMCW programmes (Mobileye, Aurora) are gone or captive. Aeva's 2025 sales are about 4% of Hesai's and a third of Innoviz's.

Addressable market under two anchors

Anchor 1, published. Yole Automotive LiDAR 2025: USD 861M (2024) to USD 3.8B (2030), 28% CAGR (Yole report page; press release variant USD 859M to 3.56B, 24% CAGR). Sivers' own 20 Oct 2026 release quotes the same Yole figure. Industrial LiDAR: no 2025 to 2030 Yole dollar figure found; the 2023 Yole deck gave 27% shipment CAGR. We carry industrial at USD 1.0 to 1.5B in 2030 (speculated, from Ouster USD 169M + Hesai/RoboSense robotics ≈ USD 150M in 2025 growing 25 to 30%/yr). Combined 2030 TAM anchor 1: ≈ USD 5B (confirmed automotive + speculated industrial). cap. 19:17.

Anchor 2, derived. 2024 automotive units 1.6M (Yole); Hesai alone guides 3 to 3.5M for 2026, so 2026 industry ≈ 5 to 6M units (est.). At 25%/yr unit growth to 2030 ≈ 12 to 15M units at a falling blended ASP of USD 200 to 250 ⇒ USD 2.4 to 3.8B automotive (est.), consistent with Yole. Western long-range FMCW units carry USD 500 to 1,000 (est., Aeva has never disclosed pricing; Luminar's Iris was ≈ USD 1,000 and Aeva's Atlas is priced to the same slot). Industrial sensors (Ouster, Eve-class) USD 1,000 to 6,000.

Share needed. For Aeva to reach the scenarios below: base USD 300M in 2030 ≈ 6% of anchor 1 ≈ 300 to 600k units at USD 500 to 1,000; boom USD 700M ≈ 14% of anchor 1 ≈ 0.7 to 1.4M units. Chinese suppliers held 93% of passenger-car LiDAR in 2024 (Yole), so Aeva's addressable slice is the ex-China Western OEM, truck, industrial and defence pool, which we put at USD 1.2 to 1.8B in 2030 (speculated); base case is then 17 to 25% of that pool, boom 40 to 60%.

Scenarios: 2030 revenue under bust, base and boom (23 October 2026)

Consensus (StockAnalysis / S&P Global, 5 analysts, 21 Oct 2026, cap. 19:19): 2025A USD 18.1M, 2026e 32.4M (31.2 to 33.8), 2027e 72.7M, 2028e not public (locked). Yahoo (4 analysts): 2026e 32.3M, 2027e 73.9M (57.7 to 92M). StocksGuide (7): 2026e 33.6M, 2027e 73.2M. Named houses in the public record: Oppenheimer Outperform, target raised to USD 42 from 33; Morgan Stanley Equal-weight, 27 from 18.50; Roth Buy 17 (older); consensus Buy (3 Strong Buy, 2 Hold), average target USD 29, median 27, high 42, low 15. Cantor, Canaccord, Craig-Hallum and Goldman notes were not found on the web (terminal/IBKR needed). MarketScreener's AEVA page resolved to the wrong instrument and gave nothing. Fintel not found.

All three are est.; unit prices are our assumptions, not company data. The bust case alone already requires 50%/yr growth, which is why the stock trades at ≈ 19× 2026e sales (EV basis; corrected 2026-10-23 from ≈ 16×).

Valuation: market data and scenario values as of 23 October 2026

Price USD 9.88 (close 22 Oct 2026, StockAnalysis, captured 23 Oct 08:52). IBKR shows 9.83 for the same close; the 0.5% vendor gap is unreconciled and does not change any conclusion.

Consensus: 5 analysts, 2026e USD 32.4M, 2027e 72.7M (S&P Global via StockAnalysis/Yahoo); 2028e not public. Targets: LSEG mean USD 29 (5 analysts, median 27, range 15 to 42). Latest per broker: Oppenheimer 42 (6 Aug), Craig-Hallum 35, Morgan Stanley 27 Equal-weight (raised from 18.50 of 2 Mar), Canaccord 26 (6 Aug), Goldman Neutral 15 (cut from 20 at the 24 Aug initiation, ~6 Oct per TipRanks). Morningstar quantitative: 3 stars, fair value USD 14.10 (19 Oct 2026), no moat, very high uncertainty (IBKR, captured 22 Oct). (corrected 2026-10-23: Goldman 20 → 15, Craig-Hallum added at 35; Morningstar added) (confirmed, 22 Oct 19:16)

H2 2026 cash burn at the H1 rate (~USD 57M) would leave about USD 120M at year end, roughly two years of runway before the standby facility. (est.)

Dilution history

Shares rose 10.7% from 31 Mar to 30 Jul 2026. There is no ATM program. (confirmed)

Scenario values per share today

Revenue paths from the scenarios section: bust USD 150M, base 300M, boom 700M in 2030. Discount rate 15% over four years (higher than the 12% used for Sivers and X-FAB, because Aeva is pre-production and loss-making). Net cash in 2030 and share counts reflect the funding each path needs. All est.

At USD 9.88 the market prices roughly the base case: Daimler in production from 2027, the European OEM from 2028, and industrial volumes. The analyst average of ~USD 29 sits at the boom value. (est.)

Broker and ratings data, IBKR Mobile, captured 2026-10-22 and 2026-10-23

Two capture runs on emulator-5554, IBKR Mobile read-only: 22 Oct 2026 19:05 to 19:09 (stopped mid-run) and 23 Oct 2026 08:54 to 09:05 Europe/Luxembourg (resume). The Run column shows which. Screenshots are in this folder.

Gaps

• Holder-level ownership table and insider trade log not captured
• Annual revenue forecast not offered by LSEG for AEVA (use StockAnalysis/S&P consensus)
• Short interest as % of float not shown
• News between 10 and 16 Oct partly skipped by scrolling; no material company item spotted at either edge
• Morningstar full PDF not downloaded

Leadership and governance: founders, insiders and the letters (23 October 2026)

Who runs it and who owns it

Beneficial ownership at 31 March 2026 (before the June offering; confirmed, DEF 14A):

Sylebra also holds a Series A warrant for 3.0M shares at USD 5.00 (capped at 19.9% ownership) and the USD 125M standby preferred facility to 8 Nov 2026 (fees USD 3.8M paid). Sylebra filed 13D Amendment 4 referencing a Contribution Agreement and a Termination Agreement dated 30 Jun 2026 (terminating a 5 Nov 2025 forward purchase agreement); post-amendment share count not captured (secondary, EDGAR index, captured 19:30). Free float est. ~55% of 69.7M after the offering, excluding the 25.8% Sylebra block, the two founders and LG Innotek; the Canaan and Lux stakes are venture holders that can sell at any time (est.).

Prior outcomes and controversies. The founders have one prior outcome: Aeva itself, taken public via the InterPrivate SPAC in March 2021 at about USD 2B, which fell to about USD 0.80 split-adjusted by March 2024 (secondary, Fugazi 21 Aug 2026). A USD 14.0M settlement of the Delaware Chancery SPAC class action (Smith and Katz, 2024-0221) received final approval 12 Sep 2025; Aeva paid under indemnification to the former InterPrivate directors with USD 2.5M insurance recovery expected (confirmed, FY2024 10-K; KTMC case page). Neither founder was a named defendant.

Pay 2025 (confirmed, DEF 14A): Salehian USD 8.79M (salary 0.60M, bonus 1.0M, stock 7.18M); Rezk USD 8.90M (includes USD 360k relocation stipend); Sinha USD 2.66M. CEO pay up from about USD 1.5M in 2024 (secondary). Compensation actually paid to the CEO in 2025 was USD 31.7M on a net loss of USD 145M (secondary, gloom.sh proxy summary; verify in the pay-versus-performance table).

Promises versus delivery, last four letters

Pattern: product-revenue promises slip and go undated, financing promises close on time.

Revenue quality, adapted for a pre-production sensor maker

Insider register (Form 4, EDGAR; captured 19:25 to 19:33)

Founders and CFO sold about USD 20M between 16 June and 2 September 2026 (est. from the rows above), all within four weeks of the June raise and the 5 August optical announcement. Sales were plan-based, but the plans were adopted in March 2026, three months before the offering. Equity grants: evergreen adds of 3,079,007 shares each to the 2021 plan and ESPP in January 2026 (about 4.9% of shares, no vote needed).

Capital raised in 12 months (confirmed): LG Innotek USD 32.5M at 9.26 (Aug 2025); Apollo USD 100M converts at 15.86 (Nov 2025); follow-on 5,168,539 shares at USD 22.25, gross USD 115M, net 108.8M (5 Jun 2026, Morgan Stanley, Goldman, Oppenheimer); Sylebra USD 125M standby preferred, undrawn, expires 8 Nov 2026. No ATM programme found. Liquidity 30 Jun 2026: cash and securities USD 177.9M plus the undrawn 125M = USD 302.9M.

Red flags

• Going concern: none in the Q2 10-Q or 10-K; the risk-factor language is the generic "may need additional financing". Burn of about USD 110M a year against USD 178M of cash gives roughly 18 months before the Sylebra facility (expiring November 2026) or another raise matters (est.).
• Short report: Fugazi Research, 21 Aug 2026, "Ridiculous valuation for a bleeding startup masquerading as AI": 72× trailing sales, USD 447M losses over three years, "valuation by adjacency" to NVIDIA and Daimler, USD 11.9M insider sales. Shares fell on the day; no company rebuttal found. The report itself has internal inconsistencies (USD 18.1M vs 21M revenue, 39× vs 47× guidance multiple, unsourced "internal control weakness" claim).
• Restatements: none found. Audit fees +59% in 2025 without explanation.
• Delayed filings: none.
• CFO exit one month after a USD 115M raise and the data-centre pivot.
• Share-price path: SPAC close March 2021 about USD 2B; USD 0.80 split-adjusted low March 2024; 1-for-5 reverse split March 2024; USD 2.50 Sep 2024 to USD 38.80 June 2025 (first retail run, +431% in a year, Oppenheimer target hikes, LG deal); USD 14.66 Sep 2025; offering at 22.25 Jun 2026; 52-week high USD 31.30 (date not captured) after the 5 Aug optical announcement (after-hours +17.8% to 23.10, 25.16 on 12 Aug); USD 15.63 on 28 Aug after Fugazi; about USD 10.03 to 10.60 on 16 to 22 Oct 2026, 52-week low 8.83, year-to-date about −20% (secondary: Morningstar 16 Oct, MarketScreener 22 Oct 09:37 EDT 10.26). Market cap est. USD 0.72B at 10.30 × 69.7M.
• Meme and retail: WallStreetBets and Stocktwits attention in the June 2025 run ("retail remains bullish", Stocktwits 26 Jun 2025); Serenity's mentions (22 since Feb 2026, latest 20 Oct 2026 per serenitystock.com) sit on the Sivers-to-Aeva laser inference, which no Aeva document supports: Aeva's 4D LiDAR integrates the laser on its own silicon-photonics chip module made by Fabrinet (Aeva 2021 release), and no Aeva filing names Sivers.
• Analysts: Goldman initiated Neutral, USD 20, on 24 Aug 2026 ("transportation revenue largely in the multiple") and later cut to 15 (~6 Oct, TipRanks via IBKR; corrected 2026-10-23); Morgan Stanley 27, Canaccord 26, Craig-Hallum 35; aggregator consensus USD 28.75 (range 20 to 42) on 5 analysts, i.e. targets set at two to three times the current price before the October slide (secondary).

Next dated hard test

Q3 2026 results, Wednesday 4 November 2026 after the close (company release 14 Oct 2026). What settles the letters: Q3 revenue against the USD 17.6 to 23.6M the full-year range needs in H2; whether the USD 30 to 36M guidance is kept, cut or quietly dropped; product revenue versus NRE; a Daimler series-production date; any committed volume in the hyperscaler JDA; the permanent CFO; Sylebra facility renewal before 8 November.

Serenity's claims versus the evidence

Bottom line for the parent: the Sivers-to-Aeva link is an inference from product fit and a customer quote, with no primary confirmation on either side. Aeva's own numbers are small (USD 18M 2025), 60% services in H1 2026, and the stock already prices the bust case at about 3× 2030 revenue. The first dated tests are Aeva's Q3 on 4 Nov 2026 (does product revenue recover, any supplier or Daimler SOP disclosure) and Sivers' Q3 on 26 Nov 2026 (first LiDAR shipments and whether the customer is described further).

Fact audit: corrections made on 23 October 2026

Adversarial check of the three section files against the 10-K 2025, 10-Q Q1 and Q2 2026, 8-K exhibits, DEF 14A 2026, the June 2026 424B5, company releases and IBKR captures. Capture 23 Oct 2026, Europe/Luxembourg.

Corrections applied 2026-10-23

• METADATA.json: ISIN corrected from US00835Q1031 to US00835Q2021.
• SECTION_business_valuation.md: listing row now carries CUSIP and ISIN; price note adds IBKR 9.83; analyst line corrected (Goldman 20 → 15, Craig-Hallum 35 added, Morgan Stanley history noted) and Morningstar 3 stars, FV 14.10 added.
• SECTION_competitors_scenarios.md: EV corrected from USD 510M to 611M; EV/sales on 2030 revenue from 3.4/1.7/0.7× to 4.1/2.0/0.9×; "≈ 16× 2026e sales" to ≈ 19×; guidance row notes it was not repeated on 5 Aug.
• SECTION_leadership.md: Goldman cut to 15 and Craig-Hallum 35 added.

What is still open

• Aeva is the Sivers LiDAR customer (speculated; no primary source on either side).
• Sensor ASP and per-sensor laser content (est.; never disclosed).
• Daimler Truck SOP date and volume (undisclosed; slipped from "by 2026").
• RSU and option overhang not added to the fully diluted count.

• Capacity of the LG Innotek and Jabil lines (undisclosed).

• Fugazi report primary copy. Form 4 by Form 4 reconciliation of the ~USD 20M insider total. Sensor ASP and laser content per sensor. Daimler SOP date. 2028 consensus.

Sylebra 13D/A No. 4 post-30 June holdings; the 10-Q Items 1 and 5 (legal, 10b5-1 changes) not captured; FY2025 revenue by customer and geography from the 10-K; pay-versus-performance table; exact date and level of the USD 31.30 high; whether the Q2 call reaffirmed the 2026 range verbatim (transcript paywalled); Rezk's late-2025 and early-2026 sales that Investing.com references (not itemised here).

Checkpoints that can change the view

• 4 November 2026, Q3 2026 results: revenue against the USD 17.6 to 23.6M the second half needs for the USD 30 to 36M range; whether the range is kept; product revenue versus engineering services; operating cash burn.
• 8 November 2026, Sylebra standby facility expiry: renewal, draw or lapse of the USD 125M preferred facility.
• Daimler Truck series production: a dated SOP and committed volumes would move the base case from assumption to schedule.
• CFO: a permanent appointment after the 5 September departure.
• Sivers: any named customer in Sivers' Q3 report on 26 November 2026, or a supplier named in Aeva's 10-K.
• Insiders: further 10b5-1 sales or a first open-market buy.

Information dates and future updates

Price and broker data are stamped 22 and 23 October 2026 (Europe/Luxembourg). Filings: 10-K 2025, DEF 14A 2026, 10-Q Q1 and Q2 2026, Q2 2026 release (5 August 2026), June 2026 424B5, Form 4 filings through September 2026, Q3 results notice (14 October 2026). The review date for every record is 4 November 2026. Sources and retrieval hashes are listed in the published source manifest.

Sources read for the competitor and scenario sections

Sivers release 11 Mar 2026 (sivers-semiconductors.com/press/strategic-lidar-customer-ramps-...); Sivers release 20 Oct 2026 (prnewswire 302913506; inelectronics.co.uk caveat); Aeva 8-K ex-99.1 26 Feb 2026 and 5 Aug 2026 (sec.gov 0001193125-26-076620, -335102); Aeva 10-Q 30 Jun 2026 (sec.gov 0001193125-26-338179); Fool transcript Q2 2026; Aeva Business Wire 14 Oct 2026 (finviz 402257); StockAnalysis AEVA forecast; Yahoo AEVA analysis; StocksGuide; Hesai 6-K FY2025 (sec.gov 1861737); CnEVPost 25 Mar 2026; Innoviz FY2025 release; Ouster FY2025 release 2 Mar 2026; Luminar 8-K and chapter11cases.com; LidarNews 13 Oct 2026; Yole Automotive LiDAR 2025 page and press release; optics.org Yole items; Forbes 5 Jan 2026 (Mobileye FMCW abandoned); RevoAIution Substack 12 Aug 2026; siliconmatter Substack; LG Innotek Reuters 29 Jul 2025; Zacks (100k units/yr); dcf-analysis.com (200k/yr, unverified).