October 29, 2026 17 min read

Mobileye (MBLY) — Q3 FY2026 Analyst Q&A Breakdown

Participants on the answer side: CEO Amnon Shashua (programs, technology, robotaxi, Mentee), CFO Moran Shemesh (financial mechanics and guidance), EVP of Corporate Strategy Nimrod Nehushtan (competitive positioning, product strategy), with IR head Dan Galves moderating. Nine sell-side firms asked 18 question units, including separate follow-up turns and three multi-part asks. Figures analysts inquired about are labeled as analyst-provided throughout; the per-question entries carry only what management added beyond the prepared remarks.

Q1 — Chris McNally, Evercore ISI

The Question: (a) Status of the VW relationship, split between the production-car decisions — “the three big ones over 2027, 2028” — and (b) the robotaxi relationship with Moia, referenced against German press reports of CARIAD taking on new partners (Wayve in, possibly Bosch out) and Moia not going as financially planned.
Management’s Response: A structured four-program map of the VW Group portfolio: base ADAS and Surround ADAS across VW brands, SuperVision led by Porsche, Chauffeur led by Audi, and the ID. Buzz program with Moia — all now running on production hardware and production-intent vehicle platforms, with software hardening as the primary focus toward 2027–2028 launches across multiple models. On CARIAD: it has been inside VW’s autonomy plan since the 2022–2023 program announcements, is now in “a phase of replacing their partner” (previously Porsche) — and Mobileye cannot be that partner, because CARIAD exists precisely to be a second source for VW’s autonomy ambitions. On Moia: the focus is executing toward driver-out on the roughly 100 ID. Buzz prototypes deployed across multiple geographies, with steady progress on data accumulation for validation and key performance metrics; management declined to engage the partner-internal speculation.
The “Analyst’s Take”: Direct and unusually complete — the cleanest public articulation of the VW structure this cycle. The load-bearing boundary line: “That someone else cannot be Mobileye because that is the whole point of CARIAD.” Tone stayed unruffled by the press angle.

Q2 — Chris McNally, Evercore ISI

The Question: Paraphrase check: if CARIAD continues (shrunken over time), would anything CARIAD builds internally for VW compete for production awards in coming years — more in the 2030s launch timeframe, winning on its own merits?
Management’s Response: Essentially a confirmation (“That’s correct”). Amnon’s addition: it does not reflect on Mobileye’s nominated programs, which are slated for production. Nimrod closes the loop: normal future-architecture nominations will run in parallel, CARIAD was always in that picture, and nothing is new here other than CARIAD selecting a new partner.
The “Analyst’s Take”: Low-information by design — but valuable as a categorical clarification: any CARIAD-internal VW win is a 2030s-era phenomenon and does not encroach on the nominated 2027–2028 programs.

Q3 — Chris McNally, Evercore ISI

The Question: On the dual-track go-it-alone strategy — is it fair to say Mobileye would look for partners to carry the physical infrastructure side (depots, vehicle ownership), as other AV players do?
Management’s Response: Yes — partners take the operational layer: “warehouses, parking lots, all the low level and mid-level maintenance work that relates to the fleets themselves.” Current focus is sequenced: put the vehicle platform together with the self-driving system as a ready platform in 2027, start in one city, then scale gradually over the following years as the ecosystem matures.
The “Analyst’s Take”: Concise and direct — really a sequencing statement: control the autonomy stack in-house, rent the operations layer from partners.

Q4 — Dan Levy, Barclays

The Question: China export-share durability — Mobileye’s share-holding has surprised; competitors like Horizon and Momenta claim they can serve exports through various JVs; why does management expect to hold a good export share?
Management’s Response: Nimrod: the claims are “not new” — the company has faced versions of them “at least for 4 years now.” Share holds on two pillars: regulation compliance, and the performance standard European consumers expect — by his claim, most driver-assist features available in Europe are provided by Mobileye. Chinese OEMs building share in Europe need the most credible driver-assist offering available; the logic extends to Asia-Pacific, South America, and India. Confidence cites repeated selection across multiple decision cycles by the big Chinese OEMs.
The “Analyst’s Take”: Assertive but numbers-free. Management argues the moat from regulation and performance credibility rather than price, and pivots away from the JV-mechanism question — the direct competitive rebuttal (Horizon/Momenta by name) never lands.

Q5 — Dan Levy, Barclays

The Question: The two new Cloud-Enhanced ADAS wins following Stellantis — how wide can adoption get, and are they gateways to further advanced-content wins with those automakers?
Management’s Response: Nimrod details the mechanics: these are incremental upgrades to existing EyeQ6-like base-ADAS programs that are mature and proven in production — same chip, same hardware, “mostly software validation work” — delivering better features at affordable cost. The wins improve the long-term relationship, showcase more of the stack, and are positioned as the deliberate bridge: “the next logical step is the Surround ADAS with EyeQ6H.”
The “Analyst’s Take”: Genuinely new — this clarifies why Cloud-Enhanced wins are cheap to execute but strategically load-bearing: they are OEM on-ramps to the higher-content programs, not endpoints.

Q6 — Edison Yu, Deutsche Bank

The Question: (a) On the internal robotaxi effort — the big milestones for next year; (b) the envisioned vehicle form factor, given the variety of vehicles now on the road even in places like L.A.
Management’s Response: Amnon frames the vertical-integration rationale around control — better product definition, faster software updates, faster hardware decisions — and concretely around a cost experiment: the standard lidar/radar/camera robotaxi is expensive, and with its imaging radars (claimed best on the planet for resolution, dynamic range, and side-lobe detection) Mobileye wants to build a genuinely cost-optimized platform it could not pilot inside an OEM program. Milestones: in 2027, deploy 100 vehicles while owning the full cycle — self-driving system, vehicle platform, upfitting, operations, fleet management, customer-facing app, teleoperation, back office — “all of that on 100 vehicles in one city in 2027. We are thinking of the Orlando area, Miami next.” In parallel, Mobileye is working with a number of OEMs to define a level-4-ready vehicle platform that can be bought at scale for expansion from 2028 and beyond.
The “Analyst’s Take”: Milestone-dense and direct on (a). But (b) was never answered — the form factor did not get a word, despite being half the question.

Q7 — Edison Yu, Deutsche Bank

The Question: (a) Mentee Robotics is in final stages of design — should we expect the next version ready by CES? (b) On AMD’s acquisition of World Labs — thoughts on the importance of in-house world-model or spatial-model capability.
Management’s Response: Amnon: the plan is to bring V3.2 on stage at CES performing useful end-to-end tasks — “not just locomotion.” V4 “should be built around the end of Q1 2027,” with a plan to build a few hundred robots for production in 2028. On world models: Mobileye is developing one as “part of our stack” — implying the capability question the acquisition prompted is already handled in-house.
The “Analyst’s Take”: Net-new and specific on the Mentee timeline; the World Labs angle was defused in a single sentence.

Q8 — Joseph Spak, UBS

The Question: Broadening from VW: there feels like a sea change in OEMs thinking about solutions beyond the current generation — the lay of the land in next-generation customer conversations, and the implications for Mobileye beyond the current awards.
Management’s Response: Amnon’s most strategic answer of the call. The period’s strategy was deliberate: put all resources into bringing four new platforms to production simultaneously with the lead customer (VW Group) rather than building “demoware” for business development, as some competitors do. He names the trade-off himself: “the fewer new nominations in the near term and more durable, sustainable growth in the long term.” SuperVision with Porsche reaches maturity in early 2027, which unlocks business development during 2027 with real production systems — exactly what OEMs discussing the next advanced L2++ and Level 3 programs will want to evaluate, while demoware competitors still carry demo-to-production risk that Mobileye has already paid down.
The “Analyst’s Take”: Candid to a fault — the self-critical framing (“One can argue whether it was the right strategy or not”) reads as a deliberate tone reset. The operative forward signal: the BD window reopens in 2027 on production-grade evidence.

Q9 — Joseph Spak, UBS

The Question: The full-year guidance was raised essentially by the Q3 beat, yet Q3 included the memory pass-through worth real revenue — with a similar pass-through expected in Q4, did something underlying change, or was memory always in the prior outlook?
Management’s Response: Moran works through the composition: memory was in the prior outlook, and of the $7M total Q3 pass-through, “probably half of it or 60% of it was already baked into the forecast” — so only part of the Q3 upside was new money. Q4 will not carry a like amount because SuperVision volume drops sharply into the year end (the roughly 60,000 run-rate to date rising to about 65,000 for the year means the Q3 pull-forward made Q4 smaller). EyeQ stays on its original volume track; no change was made to it.
The “Analyst’s Take”: A precise concession that de-dramatizes the beat: part of the upside was zero-margin memory revenue already partially forecast, confirming the guidance raise is SuperVision-shaped rather than a pricing-power shift. One of the call’s highest-signal financial answers.

Q10 — Shreyas Patil, Wolfe Research

The Question: With memory prices industry-wide expected to increase even more in 2027 and a bigger SuperVision shipment step-up planned, how should the pass-through and pricing evolve into next year?
Management’s Response: Amnon: “We are in the process of working with our customers on the prices for 2027” — negotiations mid-flight, full picture promised at the next earnings call. He frames memory inflation as industry-wide, not Mobileye-specific, and cites precedent: “successful occasions of working with our customers on these macro-driven cost increases” — the same approach applies now.
The “Analyst’s Take”: Procedurally vague but date-anchored; there is no commitment that future pass-through matches cost one-for-one. The January call is the test of whether pass-through keeps pace with a rising bill of materials during SuperVision’s scale-up.

Q11 — Shreyas Patil, Wolfe Research

The Question: Chinese-OEM exports are going primarily to rest-of-world markets, not so much Europe — so how should the European picture be weighed, as European OEMs cede share and Chinese automakers exporting (or eventually localizing in-region) deploy their own locally developed ADAS or hit regulatory challenges?
Management’s Response: Nimrod: today, the India market is where the Mobileye solution is being adopted. Europe is different: regulation is very strict and comprehensive, and Mobileye products have always scored highest in those regulatory tests — which is part of why Chinese OEMs keep choosing Mobileye. He says not to expect a change under local manufacturing: “if anything, it will help us to improve our competitiveness further in the European market.”
The “Analyst’s Take”: Confident restatement rather than new evidence — the localization-era assumption (that regulation-anchored advantage survives in-region production) is asserted, not demonstrated on this call.

Q12 — Shreyas Patil, Wolfe Research

The Question: A quick one on the Q4 EyeQ shipment split — international OEMs versus Chinese OEMs, given the usual seasonal uptick in Chinese volume in Q4.
Management’s Response: Moran declines the split (“we usually do not give specific volume”) but delivers an aggregate: Mobileye is delivering more than 4.5 million chips to China, roughly equivalent between the quarters, with Q2/Q3 slightly higher and no significant Q4 drop in China-OEM demand.
The “Analyst’s Take”: Granularity refused, aggregate granted — defensible disclosure posture, but it leaves the seasonal China-uptick thesis untested into the print.

Q13 — Mark Delaney, Goldman Sachs

The Question: Contextualizing the ASP opportunity: with Mobileye’s outlook for the L2+ segment at about a 20% CAGR through the end of the decade (analyst-provided figure), plus the Surround ADAS awards and growing REM momentum — where does ASP per EyeQ chip trend over the next two to three years?
Management’s Response: The call’s densest data exchange. Moran on near-term ASP: two forces pull in opposite directions — geography/mix (China OEM and emerging-market second-chip programs carrying low software content) versus Cloud-Enhanced ADAS and advanced-program bundles with higher software; Q4 should see some expected EyeQ ASP increase on demand for higher bundles; “Cloud-Enhanced ADAS plus REM, that obviously can add a few tens of dollars per EyeQ,” potentially at least offsetting the mix drag depending on volume and take rate. Nimrod on the two-to-three-year ladder: today’s base-ADAS ASP runs roughly $47–$50; Surround ADAS prices at roughly twice base (“2X”) with similar gross margin; Cloud-Enhanced ADAS contributes roughly single-digit dollars per car per year, structured as either multi-year lump sums at day one or annual licenses; SuperVision prices above $1,000 per car with gross margin “maybe closer to the 40s”; Chauffeur at about $2,500 ASP with gross margin also closer to the 40s. The blended ASP depends on volumes, which management says are too early to predict — though the VW Surround ADAS program covers a double-digit number of vehicle models selling in the millions per year, and Mobileye holds three programs of similar meaningfulness. Dan Galves closes with the framing: L2+ is the fastest-growing ADAS segment in North America and Europe on Mobileye’s own rising projections; single-chip ASPs have eroded for a couple of years, and the new wins — Surround ADAS at 2–3× the base ASP plus Cloud-Enhanced ADAS with gross profit per unit similar to Surround’s — are the turnaround mechanism.
The “Analyst’s Take”: Unusually generous with model inputs. Carry away the tension, though: the Cloud-Enhanced monetization was quoted on two unreconciled bases (per-EyeQ versus per-car-year), and the blended ASP path stays unprojectable until volume weights exist.

Q14 — George Gianarikas, Canaccord Genuity

The Question: Against Tesla’s FSD milestone announcements and (analyst-provided) take rates hitting 75% in North America — has Mobileye observed acceleration in customer inbound interest for advanced autonomous solutions?
Management’s Response: Amnon: internal testing and driving expeditions on the current SuperVision program — FSD-like — show performance that is “super good,” and the production version should land “not… that far from a Tesla FSD performance without the parking.” The gating factor is demonstration readiness: software hardening, burning down tickets, and sensor maturity must complete first. Tesla’s 75% take rate is called a tailwind for OEM attention once Mobileye can show a production-grade demo.
The “Analyst’s Take”: A rare performance claim placed on the record — and an implicit acknowledgment that demo-ability, not capability, is currently the bottleneck (consistent with the demoware trade-off management defended in Q8).

Q15 — George Gianarikas, Canaccord Genuity

The Question: As the robotaxi fleet launches next year, how does Mobileye model network-operating-center headcount and vehicle-to-operator ratios, and how should that ratio descend over the next several years?
Management’s Response: Amnon: “It always starts with one operator per one vehicle at the beginning. Then later it starts reducing.” Vision-language-model technologies to reduce teleoperator attention were shown at earlier CES events and will be integrated when the time comes.
The “Analyst’s Take”: Terse. Vehicle-to-operator ratios are cost-critical to robotaxi unit economics, and the company gave the floor ratio without a path or an intermediate milestone.

Q16 — Colin Rusch, Oppenheimer & Co.

The Question: The thought process behind the Beep investment, and how the public-transport market evolves as a demand creator for self-driving vehicles.
Management’s Response: Amnon: Beep is established as the partner for robotaxi operations and day-to-day fleet management and maintenance, built on multiple years of collaboration and a common approach between the two companies’ technologies — Beep will be the partner for the coming years of the U.S. robotaxi launch. On demand creation: Beep’s activity around public transit and micro transit is an interesting angle for autonomous mobility use cases, and Mobileye is evaluating that in parallel with the launch plan.
The “Analyst’s Take”: Substantive — it pins Beep as the operator-of-record for the U.S. rollout rather than a technology deal, and surfaces an optional second demand channel that the prepared remarks underplayed.

Q17 — Colin Rusch, Oppenheimer & Co.

The Question: With the introduction of color LiDAR into the market, is there a real opportunity to evolve the sensor suite and simplify it?
Management’s Response: Amnon does not grade the incoming technology and restates the company’s architecture bet: cameras are first, and seven years of imaging-radar development produced systems capable of carrying the redundancy load — long term, “you would not need the lidars. It is only cameras and imaging radars,” with the timing horizon still unknown.
The “Analyst’s Take”: A pivot, not an answer — the specific technology framing went unaddressed while the answer reasserted the roadmap.

Q18 — Andrew Percoco, Morgan Stanley

The Question: (a) On the cadence of the ASP re-rating: with Surround ADAS and Cloud-Enhanced ADAS effectively near-term-2028 stories, should 2027 device pricing follow 2026’s trend before the 2028–2029 benefit arrives? (b) On investment needs into 2027: a notable R&D or broader-opex uptick to support Mentee and the robotaxi business?
Management’s Response: Nimrod confirms the launch timing first: both Surround ADAS and Cloud-Enhanced product launches are 2028 events. Moran quantifies the interim path: 2026 carries the near-$2 ASP decline (China OEM mix plus second-chip programs not sold in 2025); for 2027 “we expect to be more flat,” with possible moderate decrease from software-content and bundle-mix dynamics — but the 2026 headwinds should not repeat. On (b), answered late — the operator had begun closing the queue and Dan Galves reopened it to catch the second ask — Amnon: there will be some investment in 2027, “not something major,” with plans including buying GPU farms, details to come when 2027 guidance is discussed.
The “Analyst’s Take”: Part (a) is well-answered and genuinely new — the flat-2027-ASP guide is the cleanest single-half-year pricing forecast on offer. Part (b) arrived after a visible queue wobble with no numbers; the GPU-farm line is the only new thread.

Synthesis

Clusters

Concern	Analysts (Q#s)	Uniform management answer	Signal
China OEM share on exports	Levy (Q4), Patil (Q11, Q12)	Regulation compliance plus European-consumer performance credibility; repeated selections across decision cycles; sustained China demand (Q4 included)	The standing bear case gets a consistent moat answer anchored to Europe — and never a direct engagement with the JV mechanism
Robotaxi vertical-integration cost and ops	McNally (Q3), Yu (Q6), Gianarikas (Q15), Rusch (Q16)	Partner-led operations (Beep and others), single-city 2027 launch, one-teleoperator-per-vehicle start, cameras-plus-imaging-radar cost play	The reps probe capital and operating burden; management answers with optionality and partners — and no cost or ratio targets
ASP trajectory and mix	Delaney (Q13), Percoco (Q18a)	A product ladder — Cloud-Enhanced plus REM at a few tens of dollars per EyeQ, Surround ADAS around 2× base, SuperVision above $1,000, Chauffeur near $2,500 — against a 2027 expected flat from 2026’s ~$2 slide	Single-chip ASP erosion is the modeling frontier; management frames the 2028 launches as the fix and declines blended forecasts
Guidance integrity vs memory pass-through	Spak (Q9), Patil (Q10)	Memory was partially in the prior outlook (about half–60%), Q4 pass-through shrinks with SuperVision volume, and 2027 customer pricing is still under negotiation	The beat-quality check: the raise is SuperVision-shaped, not a pricing-regime change

Deflections

Topic (Q#)	What was asked	Deflection technique
VW partner politics (Q1)	Moia financials; CARIAD’s rumored new partners	Refused to engage media reports; reframed CARIAD as a structural second source whose next partner cannot be Mobileye
Robotaxi form factor (Q6)	The envisioned vehicle type	Answered strategy and milestones; the question’s half about the vehicle never came back
2027 investment scale (Q18b)	R&D/opex uptick for Mentee and robotaxi	“Not something major” — deferred to the 2027 guidance conversation; GPU farms named without dollars
Color LiDAR (Q17)	Whether incoming tech changes the sensor suite	Pivoted to the imaging-radar roadmap; graded nothing
Q4 China-vs-Western split (Q12)	Volume split by OEM type	“Usually do not give specific volume” — provided the aggregate China figure and its evenness instead

Contradictions

Contradiction	Prepared-remarks claim	Q&A walk-back
Robotaxi vehicle-build timing	“We expect vehicle build to kick off at the beginning of 2026”	The Q&A timeline (ready platform in 2027, 100-vehicle one-city launch, Orlando first) puts build and launch in 2027 — the prepared-remark date reads a year early and is most plausibly a transcription slip
Guidance-raise attribution	The raise is SuperVision upside, with revenue-to-income conversion “consistent with our expectations”	Q9: roughly half to 60% of Q3’s memory pass-through was already in the forecast, and Q4 cannot repeat it on lower SuperVision volume — the “newness” of the raise is smaller than the beat
Lidar-exit urgency	The robotaxi stack relies on imaging radars to “reduce lidar content in anticipation of the next cost reduction wave”	Q17 frames lidar elimination as a long-term bet with an “unknown” horizon — near-term remained hedged
Cloud-Enhanced monetization	Moran in Q13: “can add a few tens of dollars per EyeQ”	Nimrod, same exchange: “roughly single digits in dollars per car per year” — per-chip versus per-car-year bases were never reconciled on the call

Unanswered

Q6 (Yu) — the robotaxi vehicle form factor was never described.
Q13 (Delaney) — a blended, volume-weighted ASP path is “a little bit early” to predict, so the headline modeling question stays open.
Q15 (Gianarikas) — no vehicle-to-operator target beyond the 1:1 launch ratio; improvement logic only.
Q17 (Rusch) — color LiDAR was never directly engaged.
Q18 (Percoco, part b) — the 2027 investment scale remains unquantified beyond “not something major” and GPU-farm plans.