Arbe Robotics: What In The Pipeline Is Really Contracted, And What Is Still Only Evaluation
Arbe's commercialization pipeline looks busy, but its contractual layer is still narrow. Magna sits on a conditional framework agreement, Hirain provides the only clearly disclosed binding anchor, and the $257 thousand of remaining performance obligations shows how little has already converted into near-locked revenue.
What This Follow-Up Is Isolating
The main article made a simple point: Arbe bought time, but 2026 still has to prove commercialization. This follow-up narrows that argument into a more important question: which parts of Arbe's pipeline already sit on a contract, purchase order or measurable commitment, and which parts still rest on evaluation language, future-selection language or partnerships that do not yet create revenue certainty.
That distinction matters especially here because Arbe says it expects to offer its chipset solution in commercial quantities in 2026, but ties that timetable explicitly to finishing product development and obtaining orders for commercial quantities. In the same filing, it says that in order to achieve profitability it needs orders and binding commitments from its current evaluation projects. Even in management's own wording, interest, evaluation and selection are not the endpoint. They are only the setup.
That matters even more when the visible dollar layer is still so thin. 2025 revenue was $1.026 million, cost of revenue was $1.828 million, and gross margin remained negative at 78.2%. At the same time, remaining performance obligations at year end were only $257 thousand, all expected to be recognized in 2026. That is the key read: the pipeline can look busy, while the revenue already sitting on hard contractual footing is still very small.
The Pipeline Map: What Is Contracted, What Is Conditional, What Is Still Evaluation
| Channel | What is actually disclosed | Economic certainty level | What is still missing |
|---|---|---|---|
| Magna | A Supply Agreement signed in November 2025 that will regulate the production phase when Magna has an OEM contract for imaging radar | A real framework agreement, but still conditional on an outside event | No disclosed OEM contract, no volume, no shipment timeline, no backlog |
| Hirain | A set of agreements for development and manufacturing of radar products based on Arbe's chipset, an older preliminary order for 340 thousand chipsets to secure capacity, and a separate binding PO for thousands of chipsets to ship in 2026 | This is the clearest disclosed binding anchor | No disclosed dollar value, no named end program, and the 340 thousand units are still not binding backlog |
| Sensrad | A close collaboration, alongside several Sensrad customer wins in off-road vehicles, smart infrastructure and other fields | Real traction outside passenger cars, but still one layer above revenue visibility | No order size, no delivery cadence, no disclosed revenue contribution |
| OEM projects | Multiple active RFIs, RFQs and advanced evaluations, eight selected projects, two with top-five global OEMs, and more than six large OEMs in Europe, the U.S. and Japan in evaluation during 2026 | Broad commercial interest, but mostly pre-contract | The decisions, design wins and binding commitments are still ahead |
| Hard-dollar contracted layer | Remaining performance obligations of $257 thousand at year end, expected to be recognized in 2026. Two customers accounted for 72.85% and 16.03% of revenue | This is the hard contractual layer visible at year end | It is too small to qualify as proof of scale |
That table is why it is misleading to talk about "the pipeline" as one block. Part of it is a conditional framework, part is a real but still small binding order, part is evidence that adjacent markets are alive, and a large part still sits at the stage where customers are evaluating, comparing or moving toward a future decision. Once those layers are blended together, revenue quality looks better than it really is.
Magna And Hirain Are Not The Same Rung
Magna is a gate, not a win
The Magna disclosure looks stronger than it is on a fast read because the wording is narrow for a reason. Arbe says Magna Electronics and Arbe signed a Supply Agreement that will regulate the production phase when Magna has an OEM contract for imaging radar. That is the whole point. Arbe is not saying Magna already has the OEM production program. It is not saying there is a purchase order. It is not even giving a revenue timeline. It is saying the legal and operational framework for production is in place, but the economic trigger still depends on Magna winning at the OEM level.
That still matters. It suggests Arbe is building real production architecture with a named Tier 1 rather than only circulating presentations. But it is not yet evidence of near-term contracted revenue. Until the OEM contract exists, Magna is proof of commercial preparation, not proof of signed demand.
Hirain is the clearest contracted signal, but not yet proof of scale
The Hirain disclosure actually contains two very different layers, and this is where the commercialization story is easiest to over-read. The first layer is the preliminary order issued in the fourth quarter of 2022 for 340 thousand radar chipsets to secure capacity for the anticipated production ramp-up. The company says explicitly that because of the preliminary nature of that order, it is not included in backlog. That makes it a signal of intended scale, not a piece of binding revenue that can already be counted.
The second layer is the binding purchase order that Hirain has already issued for thousands of chipsets to be shipped in 2026 in support of the initial ramp to production. That is a different level of disclosure. This is not only intent. It is a binding order. That makes Hirain the only place in the filing where Arbe clearly crosses from evaluation language into order language.
Even here, though, scale still matters. Thousands of chipsets for an initial production ramp are not the same thing as proven serial production. They show that the path has moved forward in a meaningful way, not that the distance has already been covered. Once that is read alongside the filing's separate statement that GlobalFoundries' AEC-Q100 qualification process is scheduled to be completed in 2026, the right conclusion is narrower: Hirain is a real anchor, but it still sits inside a transition year in which both technical readiness and commercial certainty need to widen.
Sensrad And The Adjacent-Market Problem
Sensrad matters because it prevents the lazy conclusion that everything outside the Western passenger-car path is still just a story. The filing is more concrete than that. Sensrad delivered its first radar series powered by Arbe's chipset for a defense-sector autonomous off-road vehicle application and for smart infrastructure projects. It also announced several customer wins in off-road vehicle applications, traffic infrastructure and other fields.
That means Arbe already has evidence that its technology is relevant outside the long automotive OEM decision cycle. That is important, especially because the company also says these adjacent-market initiatives are expected to begin contributing revenue in 2026.
But the economic layer disclosed here is still thin. There is no contract size, no delivery schedule, no revenue contribution, and no way to tell from the filing whether these wins are isolated projects or the early shape of a repeatable engine. So Sensrad supports the view that Arbe's technology is commercially relevant outside passenger cars, but it still does not justify treating adjacent markets as a contracted revenue engine.
The Real Bottleneck Is The Gap Between Selection Language And Dollar Language
This is where the filing is actually unusually clear. On one side, the commercial-interest layer is real. Tier 1s have already submitted radar systems based on Arbe's chipset into multiple RFIs, RFQs and advanced solution evaluations that are active with OEMs, robotaxis, delivery robots and autonomous trucks. Arbe says it has been selected for eight imaging-radar-based perception projects, two of them with top-five global OEMs, with selection dates now in 2026 and 2027 after slipping from the earlier 2024 target. It also says its radar is in constant field trials with the top-15 global automakers by sales, and that more than six large automotive OEMs in Europe, the U.S. and Japan are planning or are currently evaluating solutions using Arbe's chipset during 2026.
Those are not empty phrases. They do indicate real commercial interest. But the same filing also says something much more important: in order to achieve profitability, the company needs orders and binding commitments from its current evaluation projects. In other words, even by management's own wording, the decisive stage has not been completed yet.
That is why the $257 thousand of remaining performance obligations is analytically heavier than it first looks. RPO is not the whole pipeline, and it is not supposed to capture every future opportunity. But it does show the layer that is already sitting on signed, not-yet-recognized work. When that number is this small, while two customers generate almost 89% of annual revenue, the reasonable conclusion is that the pipeline exists, but the part of it that has already become high-quality revenue is still very limited.
So the debate around Arbe is not whether there is interest in the technology. There is. The debate is how quickly interest, trials, selections and framework agreements actually convert into orders, backlog and revenue that can carry the story. Until that layer thickens, the company remains closer to proof mode than to proven commercialization.
What Would Count As Real Proof From Here
The next step does not require a broader pipeline narrative. It requires movement between stages. First, Hirain has to widen from a ramp-stage order for thousands of chipsets into a broader commercial order, or part of the 340 thousand preliminary reservation has to move into binding order or backlog status. Without that, the most advanced disclosed commercial path still remains too small.
Second, Magna or one of the eight selected projects has to mature into an OEM-linked contract that creates a higher-certainty revenue layer. As long as Magna still sits at "when Magna has an OEM contract", and as long as the two top-five OEM programs still sit at future selection dates in 2026 and 2027, the market will continue, reasonably, to treat much of the value as optional rather than locked.
Third, adjacent markets have to show up in dollars, not only in names. If Sensrad and the non-automotive channels are indeed expected to contribute revenue in 2026, then that layer has to start appearing in reported revenue, in RPO, or at least in a broader customer base. Otherwise it remains an encouraging story, not commercialization proof.
That is the follow-up conclusion: Arbe's pipeline is not hollow, but as of year-end 2025 most of its value still sits between commercial interest and revenue-bearing contract. Hirain is no longer only a promise. Magna is still conditional. Sensrad is already relevant, but not yet measurable. Everything else still has to pass the only test that really matters, whether evaluation language starts turning into dollar language.
Disclosure: Deep TASE analyses are general informational, research, and commentary content only. They do not constitute investment advice, investment marketing, a recommendation, or an offer to buy, sell, or hold any security, and are not tailored to any reader's personal circumstances.
The author, site owner, or related parties may hold, buy, sell, or otherwise trade securities or financial instruments related to the companies discussed, before or after publication, without prior notice and without any obligation to update the analysis. Publication of an analysis should not be read as a statement that any position does or does not exist.
The analysis may contain errors, omissions, or information that changes after publication. Readers should review official filings and primary sources before making decisions.