Amnon Shashua has expanded Mobileye's definition of mobility just as its balance sheet delivered a severe verdict on the past. In February 2026, Mobileye completed the $900 million acquisition of Mentee Robotics, a humanoid-robotics company co-founded and chaired by Shashua. Two months later, Mobileye recorded a $3.79 billion non-cash goodwill impairment connected to Intel's 2017 acquisition of the driving-technology group. The write-down did not consume cash, but it reduced the tolerance for another promise whose value lies years ahead.
The strategic case is that autonomous vehicles and humanoid robots share a difficult core problem: machines must perceive an unpredictable physical environment, infer intent, plan actions and operate safely. Mobileye has spent more than two decades combining computer vision, specialised chips, mapping and driving policy under severe automotive cost and reliability constraints. Mentee adds a vertically integrated robot and a team focused on general physical artificial intelligence. Shashua describes the combination as Mobileye 3.0.
The commercial case is less mature. Mentee expected its first customer proof-of-concept deployments in 2026 and targets series production in 2028. Mobileye's automotive products already sit in more than 230 million vehicles and generated $1.89 billion of revenue in 2025. One business has scaled qualification, customers and supply; the other is attempting to prove that a humanoid can perform useful work autonomously and economically. Similar technology does not make those markets equally ready.
A deal that demands governance
Mobileye paid roughly $612 million in cash and issued about 26.3 million Class A shares for Mentee. Shashua was its chairman, co-founder and a significant shareholder; Mobileye's chief technology officer, Shai Shalev-Shwartz, was also a co-founder and significant holder. Shashua recused himself from the Mobileye board's consideration. A committee of disinterested directors recommended the transaction, the audit committee approved it under the related-person policy and Intel, Mobileye's controlling Class B shareholder, also consented.
Those procedures are necessary, but they do not erase the conflict. Mobileye's proxy disclosed that Shashua's son and son-in-law were Mentee employees with options that received consideration. The buyer's two most important technical leaders had financial interests in the target. Independent directors therefore had to decide not only whether humanoid robotics fit Mobileye, but whether this particular company justified $900 million and whether the terms matched alternatives available to an arm's-length buyer.
Public shareholders must assess the result without a long operating history or disclosed revenue base for Mentee. Mobileye said the acquired unit would modestly increase 2026 operating expenses by a low-single-digit percentage and operate independently. The share component spreads risk but dilutes existing owners. The board authorised up to $250 million of repurchases partly to offset dilution from stock compensation and the Mentee transaction, effectively using more capital to manage the ownership impact.
Related-party acquisitions are not inherently value-destructive. Founders often create adjacent companies because they see opportunities before a mature organisation can pursue them. Shashua has built several applied-AI businesses and possesses unusual technical knowledge. The standard must nevertheless be higher: clear milestones, transparent capital use, independent oversight and a willingness to stop funding if customer economics do not develop.
The automotive engine is recovering
Mobileye's first-quarter performance provides some capacity for the experiment. Revenue rose 27 per cent to $558 million as EyeQ system-on-chip volumes increased 28 per cent. Adjusted operating income reached $95 million, a 17 per cent margin, compared with $59 million and 13 per cent a year earlier. Operating cash flow was $75 million even after transaction costs and accelerated option payments associated with Mentee. The company raised the midpoint of its 2026 revenue guidance by 2 per cent to a range of $1.94 billion to $2.02 billion.
The headline GAAP loss of $3.82 billion was dominated by the goodwill impairment. That asset originated when Intel acquired Mobileye in 2017 and was pushed down to Mobileye's accounts around its 2022 public listing. A decline in market capitalisation and greater macroeconomic and geopolitical uncertainty triggered a new test. The charge says more about the valuation embedded in Intel's purchase than the quarter's product demand, but it is still relevant. Public markets valued Mobileye's expected cash flows far below the accounting assumptions carried from the previous transaction.
The advanced automotive pipeline is substantial but must convert. Mobileye estimated future expected automotive revenue of $24.5 billion over eight years at the end of 2025, up 42 per cent from its prior update three years earlier. Design wins covered base driver assistance, a new Surround ADAS category, SuperVision, Chauffeur and robotaxi systems. Such estimates are based on vehicle-production projections and expected prices; they are not orders that guarantee revenue.
Execution with Volkswagen is particularly important. More than 100 autonomous ID.Buzz vehicles using Mobileye Drive were testing on public roads in six cities by the first quarter, with Oslo expected to join. Volkswagen and MOIA began pre-series production, and US testing involved Uber and other partners. EyeQ6 High-based SuperVision was running in pre-production vehicles, while Chauffeur aims at eyes-off conditional autonomy. Delays in vehicle programmes can move revenue by years, and safety failures can affect the entire platform.
India widens the map
A new design win with Mahindra gives the Asia strategy greater balance. The Indian carmaker became Mobileye's third Surround ADAS customer and second customer for the next generation of SuperVision. Mobileye expects advanced driving functions to help differentiate mid-range and premium vehicles rather than remain restricted to luxury models. India offers high vehicle volumes, difficult roads and price-sensitive buyers: a demanding test of whether Mobileye can deliver sophisticated perception with economical hardware.
Success would matter beyond one customer. The company's traditional model placed a relatively low-cost EyeQ chip and software in large numbers of vehicles. Higher-content products increase revenue per car, but they also face more competition from carmakers developing systems internally and from Chinese suppliers integrating chips, sensors and software. Indian manufacturers can choose among global and regional platforms and will press for local adaptation, cost control and ownership of data.
China remains important to EyeQ volumes and first-generation SuperVision deployments, yet local competition has intensified. Chinese electric-vehicle groups iterate rapidly and increasingly favour domestic silicon or internally developed driving stacks. Export restrictions and data rules can constrain mapping and model development. Mobileye must demonstrate that its independent architecture delivers better safety and lower total system cost than vertical integration by a carmaker.
Japan and South Korea present another route through established global manufacturers and tier-one suppliers, but qualification cycles are long. Across Asia, the transition from basic assistance to hands-off and eventually eyes-off functions will be governed differently. Mobileye's modular progression is an advantage only if regulators and consumers trust the boundaries between each level. A system marketed as hands-off but requiring eyes on the road depends on effective driver monitoring and unambiguous responsibility.
Robots are not cars without wheels
The shared technology between driving and humanoid robotics can be overstated. A car operates mainly on roads under a constrained set of rules. A general-purpose robot encounters hands, tools, stairs, deformable objects and close human contact. It needs manipulation, balance and physical interaction in addition to perception and planning. Hardware reliability, battery life and maintenance determine economics as much as model intelligence.
Mobileye does contribute valuable disciplines. Automotive systems must run on limited compute, pass rigorous validation and handle rare safety events. Its mapping and sensor experience can support localisation. Its relationships with manufacturers may help Mentee establish supply and production. Training infrastructure can be shared, and improvements in context-aware reasoning may benefit both domains. Mentee's independent unit can preserve focus while drawing on those assets.
The danger is capital and managerial distraction. Mobileye is simultaneously ramping new chips, several levels of driver assistance, robotaxi deployments and a new robotics business. Research and development expense reached $323 million in the first quarter, more than half of revenue. Cash fell after the acquisition, although the company retained roughly $1.21 billion of cash and another $133 million of marketable securities and deposits at quarter-end. A prolonged robotics development cycle could compete with automotive launches that already carry the $24.5 billion pipeline.
Intel's continuing majority ownership adds another layer. Intel has its own capital constraints and strategic priorities, while Mobileye public shareholders bear the effects of dilution and investment. The controlling shareholder approved Mentee, but its interests and time horizon may not always align perfectly with minority owners. Strong independent oversight is therefore an operating requirement, not a formality.
Milestones must replace vision
Shashua's entrepreneurial record makes the physical-AI thesis credible enough to test. Mobileye itself grew from academic computer vision into a mass-produced safety platform and survived acquisition, relisting and a severe automotive inventory correction. The first quarter showed renewed volume growth, better adjusted operating leverage and important design wins. The robotaxi programme also moved from demonstrations towards pre-series vehicles.
But Mentee should be judged with a venture-stage scorecard inside a public company. Proof-of-concept deployments need named operating tasks, uptime, human-intervention rates and customer willingness to pay. The 2028 production target needs manufacturing partners, bill-of-materials progress and safety evidence. Any technology transfers or resource sharing between Mobileye and a company formerly owned by insiders require clear disclosure.
The goodwill charge has removed an old valuation from the balance sheet; it should also sharpen the culture around new capital. Mobileye does not need Mentee to become a large revenue contributor immediately. It does need evidence that the acquisition accelerates capability more efficiently than internal development, a partnership or a smaller investment would have done.
Shashua is attempting to extend a proven autonomy stack into a much larger market before the economics of that market exist. If automotive launches stay on schedule and Mentee reaches autonomous customer deployments, the deal could join two rare assets in physical artificial intelligence. If milestones slip, the related-party structure will dominate the strategic explanation. After a $3.8 billion reminder of how optimistic valuations age, technical ambition must be matched by unusually disciplined proof.
