Arbe Robotics Ltd. (ARBE)

A pre-revenue 4D imaging-radar chip company with a genuine technology lead — and a 2026 production target it has promised, and missed, before.

Investment Research · Photoncap-style deep dive · v1 of "Arbe Robotics" · May 14, 2026

What Arbe physically does

Arbe designs the silicon and software for 4D imaging radar. Conventional automotive radar — the kind already in millions of cars for adaptive cruise control and blind-spot warning — produces a coarse picture: it tells you something is out there and roughly how fast it is closing, but it cannot resolve fine angular detail, so it cannot reliably tell a stationary car from an overhead sign from a manhole cover. That ambiguity is why legacy radar gets "filtered" — software throws away static returns to avoid false braking — and why radar alone has not been trusted for higher levels of autonomy. Arbe's pitch is a radar with enough channels and processing to produce a genuinely high-resolution 4D point cloud: range, azimuth, elevation, and Doppler velocity, at a resolution Arbe claims is an order of magnitude higher than competing radar. "4D" here means the same thing it does in lidar — three spatial dimensions plus per-point velocity — but radar gets it through a different physics, and radar has the structural advantages of working in rain, fog, snow and darkness, and of being far cheaper than lidar.

The product is a chipset, not a finished sensor: Arbe sells a transmitter chip, a receiver chip, and a radar processor chip, plus the perception software that turns the raw returns into a usable point cloud. Tier-1 suppliers and OEMs integrate that chipset into radar modules. Arbe's two reference module designs are Phoenix — a forward-facing radar for mapping and tracking what is ahead — and Lynx, a surround/corner radar for the rear and side positions, which together are pitched as a 360-degree imaging-radar suite. The binding-constraint argument for Arbe is that L2+ and L3 autonomy need a sensor that delivers lidar-like environmental understanding at radar-like cost and radar-like all-weather reliability — and that high-resolution imaging radar is that sensor. The expansion thesis that puts Arbe in the Robotics theme: the same all-weather, low-cost, velocity-aware perception is useful for off-highway machines, defense platforms and robots, and Arbe has begun explicitly marketing into those adjacencies — including an HD imaging radar variant for off-highway applications.

Product roadmap

Arbe's product nomenclature centers on the chipset and the two module reference designs. The chipset — transmitter, receiver and processor chips — is the actual product Arbe sells; Phoenix (forward-facing imaging radar) and Lynx (surround/corner imaging radar, introduced in 2022 as an "industry-first surround imaging radar") are the reference modules built around it. The critical dated event: Arbe plans the transition of its radar chip from development to production for sale to customers during 2026 — a target the company itself hedges, and one it has set and missed in prior years, which is the single most important piece of context in the entire file. Per Arbe's 2025 annual report (20-F), the company "can give no assurance that this timetable will be met," and one industry account puts Lynx and Phoenix module production at customers beginning as late as late 2027. Arbe has also recently introduced an HD imaging radar variant aimed at off-highway and industrial applications, the clearest product evidence of the move beyond passenger-car automotive into the robotics/heavy-machinery adjacency.

What Arbe does not make: it does not make lidar, cameras or the host compute, and it does not make finished radar modules at scale — it is a chipset-and-software licensor that depends on Tier-1 integrators to build and sell the modules. That dependency is structural: Arbe cannot reach production revenue on its own timeline; it reaches it only when a Tier-1 and an OEM commit a platform.

The financial print

Arbe is, for practical purposes, a pre-revenue company. Full-year 2025 revenue was approximately $1.0 million, with Q4 2025 revenue of about $0.5 million, and the 2025 net loss was $45.2 million on roughly $34.8 million of R&D spend. Management's guidance for 2026 is revenue of $4–6 million and an adjusted EBITDA loss of $28–31 million — so even the bull-case 2026 is a company with mid-single-digit-million revenue burning roughly $30 million. The balance sheet is the existential issue: Arbe held about $45 million in cash as of December 31, 2025, and in January 2026 raised $18.5 million of gross proceeds in an underwritten public offering, while also cutting expenses by roughly 15%. Combining the year-end cash, the January raise and the guided ~$30 million burn, the runway is on the order of roughly two years if everything goes to plan — but "to plan" includes the 2026 production transition actually happening, and any slip both extends the burn and damages the financing terms of the next raise. There is no forward P/E and the conventional valuation framework does not apply; the roughly $145 million market cap is a venture-style bet on optionality, not a multiple on a business. Q1 2026 earnings are scheduled for May 28, 2026 — the next binary, and the first read on whether the 2026 production transition is on track.

Customer mix today

There is no production customer mix to break out, because there is no production. The honest description of Arbe's "customer mix" is a pipeline of engagements at varying maturity. Per Arbe's disclosures, in 2024 the company engaged with roughly 15 OEMs, of which about 11 had advanced to a bid stage and about 8 had entered an "advanced perception project" phase, and Arbe's radar chipset was selected by a top-10 global automaker to support development of that OEM's next-generation imaging-radar systems. Those are real engagements, but the language — "bid stage," "advanced perception project," "support development" — is the language of pre-production qualification, not signed production contracts with unit volumes. The structural shift Arbe wants investors to see is from a 2024 automotive-only pipeline toward a 2026 pipeline that adds defense and robotics/off-highway adjacencies, evidenced by the new HD imaging radar for off-highway applications. But the financial reality is unambiguous: $1.0 million of 2025 revenue means no customer is yet a meaningful revenue contributor, and the entire investment case is that the pipeline converts.

What's actually happening at the Tier-1 / OEM conversion point

The mechanism that matters for Arbe is the conversion of "bid stage" and "advanced perception project" engagements into committed Tier-1-plus-OEM production platforms — and the honest read is that this has not happened on the timeline Arbe has repeatedly projected. Arbe's chipset was selected by a top-10 automaker for next-generation imaging-radar development, and the company says Lynx and Phoenix modules are being integrated into multiple platforms, but the dates keep moving: the chipset production transition is now a 2026 target (hedged), and module production at customers is described in some accounts as beginning as late as 2027. The structural problem is that Arbe does not control its own revenue timeline — it depends on a Tier-1 integrator building the module and an OEM committing a vehicle platform, and OEM platform decisions are slow, conservative and easily delayed. The bull mechanism, if it works: once a single major OEM platform commits to Arbe's chipset, the per-vehicle content and the multi-year platform life create a real, durable revenue base, and the company has spent years and roughly $35 million a year of R&D building a genuine resolution lead to win exactly that decision. The bear mechanism: every quarter without a signed production platform burns cash, and the longer it takes, the more time Tier-1 silicon incumbents have to close the resolution gap.

The competitive threat / TI, NXP, Infineon and Mobileye

Arbe's competitive problem is that it is a sub-$150 million-market-cap company trying to win a market against some of the best-capitalized semiconductor companies on earth. Per industry profiles, Arbe's main rivals include Texas Instruments, NXP Semiconductors, Infineon and Bosch on the radar-silicon side, plus Mobileye, which is pushing its own imaging-radar chipset (positioned to reach OEMs via Tier-1s such as Valeo and Continental). Continental's ARS540 is already one of the first 4D radars in mass production — seen on the Hyundai Ioniq 5 — representing the "safe," already-qualified choice for a Tier-1 integrator even if it trails Arbe in raw point-cloud density. The competitive dynamic is the classic micro-cap-versus-incumbent trap: Arbe may genuinely have the best resolution, but a Tier-1 choosing a radar chip weighs supplier longevity, balance-sheet durability, automotive-grade track record and roadmap support — and on every one of those dimensions TI, NXP, Infineon and Mobileye outscore a cash-strapped micro-cap. Arbe's "two-horse race against Mobileye for L3 imaging radar" framing is the optimistic version; the pessimistic version is that the incumbents close the resolution gap with engineering budgets Arbe cannot match, and Arbe's technology lead expires before its pipeline converts.

The terminal risk

The terminal risk for Arbe is straightforward and severe: it runs out of money before the technology lead converts to production revenue, or the incumbents erase the lead before conversion. Arbe is burning roughly $30 million a year against a cash position — even after the January 2026 raise — of well under $100 million. The bull case requires the 2026 production transition to happen roughly on schedule and a major OEM platform to commit; if instead the transition slips into 2027 (as some accounts already suggest for the modules) and the OEM decisions stay in "bid stage," Arbe faces a dilutive raise from a position of weakness, and the technology window narrows as TI, NXP, Infineon and Mobileye iterate their own 4D radar silicon. The named alternative beneficiaries are precisely those incumbents, plus Continental on the module side. Arbe has a credible roadmap and a real resolution advantage — that is not in dispute — but a credible roadmap is not the same as a financed one, and that is what caps any reasonable position size in this name to a token, venture-style allocation.

Bull / Gap / Optionality (Photoncap framing)

1. The technology lead is real and well-documented. Arbe claims, and industry coverage broadly accepts, an imaging-radar resolution roughly an order of magnitude above competing radar, built over years and roughly $35 million a year of R&D. High-resolution 4D radar at radar cost and all-weather reliability is genuinely the sensor L2+/L3 autonomy wants, and Arbe is the most prominent pure-play.

2. The pipeline exists and includes a top-10 automaker. Arbe disclosed engagement with ~15 OEMs, ~11 at bid stage and ~8 in advanced perception projects, plus a top-10 global automaker that selected its chipset for next-generation imaging-radar development. The funnel is real even if conversion has lagged.

3. The robotics/defense expansion widens the addressable market. The new HD imaging radar for off-highway applications, plus explicit defense and robotics positioning, gives Arbe shots beyond the slow passenger-car platform cycle — all-weather, low-cost, velocity-aware perception is genuinely useful for heavy machinery and robots.

4. The company has bought itself time. The January 2026 raise of $18.5 million plus a roughly 15% expense cut extended the runway to something like two years on the guided burn — enough, if the 2026 production transition lands, to reach the conversion inflection without an immediate emergency raise.

5. The micro-cap structure means a single OEM win re-rates the stock. At a ~$145 million market cap, one signed major-OEM production platform — multi-year, per-vehicle content — would transform the revenue outlook and the equity story. The asymmetry is real; it is just low-probability.

Gap

1. It is effectively pre-revenue with a history of missed production dates. $1.0 million of 2025 revenue, and a 2026 production transition the company itself says it "can give no assurance" of meeting — a target it has set and missed before. The single most important fact in the file is that Arbe's timelines have not been reliable.

2. The runway is finite and the next raise will likely be dilutive. Roughly $30 million annual burn against well under $100 million of cash even post-raise. If the production transition slips, Arbe raises again from weakness, diluting an already micro-cap share base.

3. The competitors are vastly better capitalized. TI, NXP, Infineon, Bosch and Mobileye all build 4D radar silicon and all outscore Arbe on supplier longevity and balance-sheet durability — the exact criteria a Tier-1 weighs. Continental's ARS540 is already in mass production as the "safe" choice.

4. The robotics revenue is entirely aspirational. The off-highway/defense/robotics positioning is roadmap and marketing — there is no disclosed robotics revenue. Within the Robotics theme, ARBE is the purest example of "option value on a loss-making balance sheet" with the option deepest out of the money.

Optionality

The trade

Arbe is a Bucket-C, token-size venture bet — a genuine 4D imaging-radar technology lead attached to a pre-revenue income statement and a runway that depends on a 2026 production target the company has missed before. If taken at all, initiate at $1.12–$1.24 (current price ± ~5%), but note a real price-verification flag: secondary sources showed ARBE closing around $1.01 on May 13, 2026, roughly 14–17% below the $1.18 reference — this is a volatile micro-cap that swings double digits on single news items, so verify the live price before committing capital. Size at no more than roughly 0.4% of risk capital — this is the smallest, most speculative position in the batch, appropriate to a name where the realistic outcome distribution is bimodal — and set a stop at $0.78, accepting that a sub-$1.20 micro-cap can gap through any stop. The named catalyst is Q1 2026 earnings on May 28, 2026, the first read on the production transition; the real binary, though, is whether any quarter in 2026 produces a signed major-OEM production platform rather than another "bid stage" update. If you want imaging-radar exposure without the existential balance-sheet risk, the cleaner expression is to own it through a diversified, profitable name — Mobileye carries an imaging-radar chipset on a fortress balance sheet, and indie Semiconductor (INDI) has radar SoC revenue today. Arbe is the pure lottery ticket; treat the position size accordingly. Conviction: 3 / 10.

Sources referenced inline throughout. Reference v1 of this template format: _Watchlist/hanmi-photoncap-style.md.

\newpage