Innoviz Technologies

Phase A — Understand the business

Lens 1 · Company Overview

Innoviz is an Israeli (Rosh HaAyin) Tier-1 direct supplier of automotive-grade LiDAR sensors plus a complementary perception software stack. Incorporated January 2016; went public April 2021 via SPAC merger with Collective Growth (Perception Capital / Antara), raising ~$370M in gross proceeds.

Business model — two revenue lines, and the mix is the whole story:

1. NRE (non-recurring engineering) — paid development work for OEM programs, recognized at milestones/customer-acceptance. This is the dominant line today: NRE contributed ~$37.2M of FY25's $55.1M revenue, vs ~$18.0M of FY24's $24.3M.
2. LiDAR sensor product sales — recognized point-in-time on delivery. Still small; the volume ramp is ahead, not behind. FY25 revenue was further flattered by a one-time $8.4M sale of machinery to a customer — i.e. ~$45.6M of the $55.1M is NRE + a non-recurring machinery sale, and recurring product revenue is modest.

Architecture: direct-detection Time-of-Flight (ToF) at ~905nm wavelength, using mature silicon detectors — deliberately chosen over 1550nm (more expensive InGaAs, more power, larger form factor) and over FMCW (which Innoviz argues is less automotive-mature). They own a custom signal-processing ASIC. Product ladder: InnovizOne → InnovizTwo (Long-Range + Short-to-Mid) → InnovizThree (3rd-gen, announced Dec 2025, CES Jan 2026, engineering samples only) → InnovizSMART/SMARTer for non-automotive.

Customers / programs (the crown jewels): BMW L3 (first design win 2018, reached series maturity 2024); Volkswagen (direct supplier across several brands incl. ID.Buzz, InnovizTwo); Mobileye (InnovizTwo LR + SMR for Mobileye Drive); Daimler Truck (Sept 2025, future series supplier for SAE L4 Class-8 trucks). A concrete near-term deployment: MOIA's VW ID.Buzz robotaxi on the Uber platform in Los Angeles by YE2026 — 9 InnovizTwo LiDARs per vehicle (3 LR + 6 SMR). Contract structure is automotive design-win economics: long, expensive qualification cycles, then (theoretically) multi-year series-production volume — but with no take-or-pay protection disclosed, and revenue from each win is admittedly hard to predict.

Lens 2 · Supply Chain

Upstream → Innoviz → end customer, named where disclosed:

• Inputs: ~905nm laser diodes, silicon photodetectors (the cost-advantage vs InGaAs), a proprietary in-house signal-processing ASIC, MEMS/scanning optics, and IC packaging. Components are sourced "from a variety of third-party manufacturers"; Innoviz designs critical components in-house rather than buying off-the-shelf modules. Specific upstream chip/laser vendors are n/a — not named in the 20-F.
• Manufacturing: InnovizTwo is built at Innoviz HQ and at contract manufacturers with automotive-grade facilities. The named CM is Fabrinet — the Q1 2026 ramp shipped "a record number of units… about half of what we shipped in all of 2025" in a single quarter. InnovizThree samples are still built in-house.
• Downstream: sells direct to OEMs as Tier-1 (VW, Daimler Truck) and through Tier-1 partners as Tier-2 (e.g. via Mobileye's Drive platform; Torque on the Daimler program). Non-automotive goes through channel partners (security/perimeter — Kela for drone detection; LOXO LOI for L4 delivery).

Chokepoints / single-source risk: the in-house ASIC is both a moat and a single point of dependency. Management states it currently holds "sufficient component inventory" for near-term demand and is building buffer stock. The deeper chokepoint is demand-side, not supply-side: the entire chain is gated on OEM SOP timing, which Innoviz does not control.

Lens 3 · Competitive Advantages (moats)

What's genuinely defensible:

• Automotive-grade certification stack — ISO 26262 (functional safety), IATF 16949, ISO/SAE 21434 (cybersecurity), ASPICE CL1, ISO 17025, ISO 27001. These take years and are a real barrier to new entrants — a credentialing moat more than a technology moat.
• Marquee design wins as proof-of-trust — BMW/VW/Mobileye/Daimler Truck. In automotive, a competitor cannot simply price you out of a won socket mid-program; switching costs within a program are high.
• IP — 81 issued patents + 124 pending as of 2025-12-31, across systems, lasers, scanners, receivers, optics, perception.
• Cost architecture — the 905nm/silicon/ToF choice is a structurally cheaper BOM than 1550nm rivals; each generation has cut cost.

Where the moat is thin (be honest):

• Bargaining power is inverted. Innoviz's own risk factors concede it targets "large companies with substantial negotiating power, exacting product standards and potentially competitive internal solutions". A ~$130M-market-cap supplier negotiating against VW/Mobileye is the weaker party.
• The market leader is pulling away. Hesai shipped its 1,000,000th LiDAR in 2025 and is doubling capacity toward 4M units; it holds ~55% of China long-range auto LiDAR. Scale economics in this industry accrue to the volume leader — and that is not Innoviz.
• No network effects; software stack is real but not a lock-in flywheel yet.

Lens 4 · Segments

Innoviz does not report multiple operating segments — it runs as a single reportable segment, disaggregated only by revenue type (LiDAR products + NRE) and not by product line or geography in the 20-F segment note. segments.csv was empty. What can be sourced:

• By revenue type (FY25): NRE ~$37.2M, machinery sale ~$8.4M, LiDAR sensor sales = the residual ~$9.5M.
• By end-market trajectory: automotive is ~99% of FY25 revenue; non-automotive is guided to rise from ~1% of revenue in 2025 to up to ~10% in 2026 — a deliberate pivot toward shorter-cycle, higher-ASP "Physical AI" applications (security, infrastructure, robotics, aerial) to bridge revenue while auto SOPs slip.
• By geography: substantially all revenue is USD-denominated; cost base is heavily ILS (Israel) — the USD devalued ~12.5% vs ILS in 2025, inflating the USD cost of operations.

Trend read: revenue type mix is the risk. A P&L that is ~83% NRE + one-off machinery is not yet a product company — it is a funded engineering shop whose product ramp is still a forward promise.

---

Phase B — Measure performance

Lens 5 · Earnings Result

FY2025 (audited, 20-F):

On paper FY25 looked like a turnaround: revenue +127%, first-ever positive gross margin, opex down 20%. But the quality is poor — the revenue beat was NRE + a one-time machinery sale, the gross-margin flip was partly from "decreased production inefficiencies of InnovizOne and decreased excess & obsolete inventory" (i.e. fewer write-offs, not operating leverage), and the opex drop was a 23% R&D cut driven by headcount reduction and lower SBC, not efficiency. Cutting R&D 23% at a company whose entire thesis is staying technically ahead is a runway-preservation move, not strength.

Then the print that matters — Q1 2026 (post-20-F):

• Revenue $7.1M, down from $17.4M in Q1 2025 — NRE milestones "pushed into coming quarters… in part due to customers' requests for additional content."
• Gross margin ~ −22% (back negative — low fixed-cost absorption on low unit volume).
• Net loss $26.2M (vs $12.6M Q1 2025); opex $24.9M (+18% YoY).
• Liquidity ~$60.1M, no long-term debt; ~$15.8M quarterly cash + capex use.
• Management reaffirmed FY26 revenue guide of $67–73M (~27% growth at midpoint) and 2–3 new programs. This is the central tension: full-year guidance held while Q1 came in at ~10% of the annual midpoint, implying a very back-half-loaded year. Hitting $67–73M now requires ~$60M+ across the remaining three quarters.

Balance-sheet flags: cash fell from $72.1M (YE25) to $60.1M (Q1 26) — one quarter consumed ~$12M of liquidity. At that pace the disclosed "sufficient for at least 12 months" runway is real but finite, and the ATM is the safety valve.

Lens 6 · Earnings Calls (sentiment trend)

transcripts/ was empty; sourced from web. Tone over the last several quarters has been persistently, structurally optimistic regardless of results — "robust pipeline," "executing on the plan," "renewed momentum in lidar / Physical AI". The recurring phrases: design wins, NRE, SOP (start of production), Physical AI, Fabrinet ramp. What's notably absent from management's voice: any direct address of the sub-$1 share price, the Nasdaq deficiency, a reverse split, or dilution — the Q1 2026 call did not engage them. The sentiment delta over time is subtle: the story (Physical AI, multi-program ramp) has gotten bigger while the near-term numbers have gotten worse — a widening gap between narrative and tape that the market has clearly chosen to price on the tape.

Lens 7 · Comps

Western/Asian LiDAR peer set. Multiples are `` with date or n/a.

The comp read is brutal and clarifying. On TTM revenue (~$55M) Innoviz is roughly mid-pack among Western names, yet its market cap (~$130–161M) is ~5% of Hesai's and ~8–9% of Aeva's despite Aeva having a fraction of Innoviz's revenue. The market is not paying Innoviz an LiDAR-peer multiple — it is pricing in dilution + delisting risk. The two opposite poles (Hesai = profitable scale; Luminar = near-zero) frame the binary outcome space Innoviz sits between.

Lens 8 · Stock-Price Catalysts

Pattern over the last ~2 years:

• Down −47% over the trailing 12 months; −25% in one recent 30-day window.
• March 26, 2025: first Nasdaq sub-$1 deficiency notice (Rule 5450(a)(1)).
• 2025 (mid-year): "optimization of operations" — ~9% headcount cut, ~$12M 2025 cost savings, framed as runway extension.
• Goldman Sachs downgrade to Neutral, PT cut to $0.75 (from $1.25).
• March 25, 2026: second formal Nasdaq notice (Rule 5550(a)(2)), 180-day grace to Sept 21, 2026.
• Institutional ownership reportedly collapsed ~66% as funds exit ahead of the delisting clock.
• May 14, 2026: Q1 2026 miss (revenue $7.1M). What moves this stock: not earnings beats — it reacts to (a) financing/dilution events, (b) the listing clock, and (c) headline design-win announcements. The market treats design wins as call-option premium and treats every quarter of cash burn as theta decay. This is a survival-and-dilution name, not a fundamentals-compounding name — yet.

---

Phase C — Judge people & books

Lens 9 · Management

• CEO Omer Keilaf (46), co-founder, in seat since 2016. Ex-STMicroelectronics, Consumer Physics, IDF elite intel/tech unit; BSc/MSc EE + MBA, Tel Aviv University. Track record: the genuinely impressive achievement is landing BMW's L3 series-production design win in 2018 as a startup, beating older incumbents. Founder-operator archetype, deeply technical.
• CFO Eldar Cegla (since 2017); CR&DO Avishay Moscovici (ex-Motorola/Freescale/NXP); long-tenured insiders — low executive churn, which for a long-cycle automotive supplier is a positive.
• Board pedigree is unusually strong for a microcap: Chairman Amichai Steimberg (ex-CEO Orbotech, acquired by KLA); Stefan Jacoby (ex-CEO Volvo Cars, ex-CEO VW Group of America — directly relevant automotive credibility); Aharon Aharon (ex-CEO Israel Innovation Authority, ex-GM Apple Israel); Dan Falk (NICE, Evogene boards).
• Skin in the game: insiders + directors as a group hold 5.2% (CEO 2.3%, incl. options/RSUs vesting in 60 days). Real but modest; not a high-conviction founder stake.
• Capital-allocation history — the weak spot. Burned through most of the $370M SPAC war chest; diluted from 169.4M → 214.1M shares in 2025 alone (~25%) via a Feb 2025 registered direct ($37.3M), an August 2025 Jefferies ATM (~$13.3M of a $75M facility used), and option exercises. ROE/ROIC are deeply negative (sustained operating losses). To their credit, the FY25 opex discipline (−20%) and the Tier-1 strategic pivot were sound calls. Red flags: in December 2025 the board granted CEO Keilaf 6,256,265 PSUs plus the group's existing 3.78M options (WAEP $7.45) and 9.44M RSUs — a very large equity grant as the stock collapsed toward $0.60, which optically rewards the team during a drawdown even if vesting is share-price-performance-gated. Total FY25 named-exec comp ~$5.1M; CEO ~$1.29M (~62% equity).
• Founder vs professional: clearly founder-led. At this stage that cuts both ways — conviction and technical depth, but also the optimism bias visible on the calls.

Lens 10 · Forensic Red Flags

Acting as a forensic analyst on the 20-F:

• Revenue recognition / quality (highest-priority flag). Revenue is ~83% NRE + a one-off $8.4M machinery sale. NRE is milestone/acceptance-based — inherently lumpy and partly discretionary in timing, which Q1 2026 proved when milestones "pushed out" and revenue fell 59% YoY. The "127% growth" headline is low-quality. The $8.4M machinery sale is non-recurring and should be stripped from any run-rate.
• Cash vs earnings. Operating cash use ($47.9M) was less than net loss ($67.8M) — but the bridge is $16.0M SBC + $5.9M D&A, i.e. losses are flattered by heavy non-cash stock comp, not by genuine cash generation.
• SBC is large relative to the P&L: $16.0M SBC in FY25 (down from $19.7M) on a company this size, and the ongoing 5%/yr evergreen share-reserve increase + Jan 2026 +16.96M-share reserve top-up institutionalize dilution.
• FX exposure: USD revenue / ILS cost base; a swing of the ~$5.0M FX line wiped out nearly all "financial income, net" (FY25 $0.1M vs FY24 $7.3M).
• Going concern: No going-concern qualification. Management explicitly states liquidity (~$72.1M YE25) is "sufficient… for at least the next 12 months". No "substantial doubt" language appears in the filing. This is the reassuring counter to the bear case — bankruptcy is not the near-term risk; dilution and delisting are.
• Remaining performance obligations: only ~$7.6M of contracted RPO to recognize within 12 months — a thin contracted backlog relative to a $67–73M guide, underscoring how much of FY26 is not yet under contract.

Regulatory findings sub-section:

• SEC Litigation Releases: none naming Innoviz (EDGAR EFTS LR, 2021-06-18→2026-06-18).
• SEC AAERs: none.
• Item 3 / Legal Proceedings (company's own disclosure): "not currently a party to any material… legal proceedings". The one disclosed matter is the Sheadrick Richards lawsuit (filed March 28, 2024) — a purported Collective Growth stockholder alleging the SPAC's 2021 business-combination disclosures were "materially incomplete and misleading" and that directors breached fiduciary duties. Innoviz indemnifies certain former Collective Growth directors/officers; it carries run-off insurance but warns coverage "may not cover all claims". This is legacy-SPAC litigation, not an operating-fraud signal.
• Non-SEC web search ("Innoviz" + FTC/DOJ/FDA/CFPB/consent decree/settlement/penalty): no material enforcement hits found.
• Net: No material regulatory or accounting-enforcement findings — verified via SEC EDGAR EFTS (LR, AAER), web search, and 20-F Item 3/Item 8 as of 2026-06-18. The accounting risk is revenue-quality/dilution, not fraud.

---

Phase D — Project & stress-test

Lens 11 · Forward Projection

Bottom-up from FY25 actuals + the reaffirmed FY26 guide. All ``, inputs labelled. EPS will stay deeply negative across the projection window — the relevant variable is not EPS but revenue trajectory, cash burn, and share count.

• FY2026 (current year): revenue $67–73M per reaffirmed guidance — but with Q1 at $7.1M, this is steeply back-half-loaded and carries high miss risk. NRE $20–30M of new payments; non-auto to 10% of revenue. Assume opex roughly flat-to-up ($95–100M run-rate from Q1's $24.9M ×~4) → operating loss ~$(60)–(75)M. Net loss EPS ≈ $(0.40)–(0.50) on ~220–235M wtd shares.
• FY2027: if VW/Mobileye/MOIA SOP volume genuinely ramps, revenue could step to ~$90–120M; but equally a program slip leaves it ~$70–80M. Operating loss narrows only if product (not NRE) gross margin scales. EPS ≈ $(0.25)–(0.40), share count likely +5–15% from continued ATM use.
• FY2028: the year management points to ("numerous L2+/L3/L4 programs decided in 2026, SOP 2028+"). Bull path: revenue $150M+ with positive product gross margin and a path toward operating breakeven; bear path: still sub-$100M and out of independent runway, having done a reverse split. EPS still likely negative.

Runway-to-catalyst (the lens that actually matters here): ~$60.1M liquidity (Q1'26) ÷ ~$50–55M annual cash burn ≈ ~12–14 months of independent runway, extendable via the ~$60M of unused ATM capacity. The cash question and the Nasdaq Sept-21-2026 listing clock converge in H2 2026. Forecast NOT logged (forecast.ts create skipped per --watchlist rules). The scoreable binary I would track: "INVZ reports FY2026 revenue ≥ $67M (low end of guide)" — my subjective p ≈ 0.30 given the Q1 hole.

Lens 12 · Bull vs Bear

Bull case. Innoviz is a real Tier-1 with certified, OEM-validated product and a design-win roster (BMW, VW, Mobileye, Daimler Truck) that competitors cannot quickly replicate — credentials that take years. The Physical-AI/LiDAR secular tailwind is intact and broadening (Mercedes/Hesai L3, NVIDIA Hyperion). Concrete 2026 deployments (MOIA/VW ID.Buzz robotaxi on Uber in LA, 9 sensors/vehicle) convert "design win" into shipped units; Fabrinet ramp is real (1 quarter = half of all 2025 units). Non-auto going 1%→10% adds shorter-cycle revenue. At ~$130M market cap on ~$55M TTM revenue with marquee customers and no debt, the risk/reward is asymmetric IF SOP volume arrives — sell-side median PT ~$2.03 implies ~190% upside. The contrarian read: the market has fully priced delisting and is mispricing the option value of the won sockets.

Bear case (2–3 permanent-impairment risks).

1. Dilution death-spiral + reverse split. Sub-$1, a Sept-21-2026 listing deadline, ~$50M/yr burn, and an active ATM = the most likely path is more shares at lower prices, then a reverse split that "signals desperation" and resets the option lower. Equity holders can be structurally impaired even if the business survives.
2. Revenue is NRE-funded engineering, not product. ~83% NRE + one-off machinery in FY25, then a 59% revenue collapse the moment milestones slipped (Q1'26). If product SOP volume keeps slipping right (automotive's chronic habit), the company funds itself by issuing stock indefinitely.
3. The leader is compounding away. Hesai is GAAP-profitable, shipping >1M units/yr, doubling to 4M, winning Mercedes L3 — scale economics flow to it, not to a sub-scale Israeli supplier whose R&D it just cut 23%.

Pre-mortem (18 months out, thesis broke): It's late 2027. Two of the "2026-decided" programs slipped to 2029 SOP; FY26 came in ~$55M (missed the guide); the company did a 1-for-10 reverse split in Q4 2026 to hold Nasdaq, raised dilutive equity twice more, and the stock is a post-split ~$3 that's already drifting back down. The product works — but volume never showed up fast enough to outrun the cash clock.

Are multiples too high? No — multiples are low (≈5% of Hesai's cap on comparable revenue). The bear case isn't "overvalued," it's "the equity is an option that may expire / be diluted before it pays."

Lens 13 · Devil's Advocate (short-seller)

Dismantling the bull case: The bull thesis rests entirely on future SOP volume the company does not control and has repeatedly failed to time. Revenue is concentrated in a handful of OEM programs (VW, Mobileye, Daimler Truck) — lose or slip any one and the model breaks; Q1 2026 showed a single quarter of milestone slippage halves revenue. The moat bulls cite (certifications, design wins) did not protect the stock from a −47% year because in automotive a "design win" is a multi-year option, not a contract — OEMs can de-scope, delay, or dual-source (note Innoviz's own disclosure that customers are "potentially competitive internal solutions"). The most dangerous competitor bulls underestimate is Hesai, whose profitability + 4M-unit capacity means it can win Western programs on price and scale while Innoviz is cutting R&D to survive. Worst capital-allocation tells: 25% dilution in a single year and a 6.26M-PSU CEO grant as the stock cratered. For today's ~$0.61 to hold, you must believe FY26 hits ~$70M (Q1 says ~$7M), the company avoids a punitive reverse split, and product margin inflects — three things that must all go right. If revenue disappoints 20–30% (i.e. lands ~$50–55M, flat-ish), the cash clock tightens, the ATM runs harder, and a reverse split becomes near-certain — permanently impairing per-share value even if enterprise value holds. The single scenario that permanently impairs: a program cancellation by VW or Mobileye, which would simultaneously gut revenue and the credibility that underwrites every other RFQ. Plausibility: moderate, not negligible.

Lens 14 · Management Questions (ordered by information value)

1. Q1 2026 revenue was $7.1M against a reaffirmed $67–73M full-year guide — quarter by quarter, what specific NRE milestones and product shipments bridge the remaining ~$60M, and which are contracted vs forecast?
2. With ~$60M liquidity, ~$50M annual burn, and a Sept-21-2026 Nasdaq deadline — what is your explicit plan to regain $1 compliance, and under what conditions would you execute a reverse split?
3. How much of FY26 revenue is product (units shipped) vs NRE vs one-off, and when does product revenue exceed NRE for the first time?
4. For VW, Mobileye, and Daimler Truck individually: what is the contracted SOP date, the committed annual unit volume, and the per-vehicle content — and what penalties or take-or-pay protect you if they slip?
5. Of the ~$60M unused ATM capacity, how much do you expect to draw in 2026, and at what point does dilution become value-destructive enough to prefer strategic alternatives?
6. You cut R&D 23% in 2025 — which programs were deprioritized, and how do you stay ahead of Hesai's scale while spending less?
7. What is your realistic path to product gross margin >30%, and at what cumulative unit volume does fixed-cost absorption turn the corner?
8. Hesai is GAAP-profitable at >1M units/yr and won Mercedes L3 — where do you win against it in Western programs, concretely, on something other than "we're automotive-grade too"?
9. The Dec-2025 6.26M-PSU CEO grant: what are the exact share-price/performance hurdles, and how do you defend a grant of that size during a >40% drawdown?
10. Non-auto is guided 1%→10% of revenue — is this a durable second leg or a bridge to fill the auto-SOP gap, and what are the ASPs/margins there?
11. What would a strategic acquirer (a Tier-1 like Continental/Valeo, or an OEM) pay for the certified IP + design-win book, and is a sale a live alternative to standalone dilution?
12. The Fabrinet ramp shipped half of 2025's units in Q1 2026 — what is installed annual capacity, and is it demand- or supply-constrained today?
13. How exposed is the cost base to further USD/ILS moves, and do you intend to hedge beyond the current 3-month window?
14. On the Sheadrick Richards (Collective Growth) litigation — what is the current status, your reserved/expected exposure, and insurance adequacy?
15. If by mid-2027 the 2028-SOP programs have not converted to firm volume, what is Plan B?

---