Foresight: How Much of the Pipeline Is Real Commercialization
Foresight’s 20-F reads like a dense commercialization map, but 2025 itself was still narrow: $398,000 of revenue, only $162,000 explicitly tied to Elbit, and most of the rest coming from POCs and development work. The pipeline is real, but most of it still sits at the proof stage rather than the deployment stage.
The main article argued that Foresight’s commercialization progress still does not justify the dilution. This follow-up isolates the reason: there is still a wide gap between the long list of partnership headlines and the actual commercial maturity visible in 2025 revenue. On first read, the 20-F looks like a crowded commercialization map. Once it is read through the company’s own ladder, only one thread has clearly crossed into identifiable revenue, while most of the rest still live in POCs, development work, or commercial discussions.
Foresight itself explains what its commercialization path looks like. First come roadshows and demos. Then come paid POCs, typically lasting three to six months and generating $25,000 to $100,000 of revenue. After that come co-development projects, with typical budgets of $50,000 to $250,000 and the potential to reach a few hundred thousand dollars. Only the final stage is transfer to production, with tens of thousands to hundreds of thousands of units or software licenses per year over eight to ten years. That is the right lens. Once the 2025 numbers are put on that ladder, Foresight’s revenue still looks much closer to stage two and stage three economics than to true production-stage economics.
The numbers make that clear immediately. Revenue fell to $398,000 in 2025. The company explicitly ties $162,000 to commercialization with Elbit, $40,000 to a successful POC with a leading Japanese traffic control and road-safety company, and the remainder to successful Eye-Net POCs with a European Tier 1 supplier and a European car manufacturer. In the main-customer note, only two 2025 customers are individually visible: $162,000 and $40,000. That is the key message: there is real commercial interest around the technology, but the actual revenue base is still narrow, concentrated, and driven mostly by paid proof work rather than broad deployment.
Elbit Has Crossed Into Revenue, but Not Yet Into Scale
If there is one thread the filing genuinely presents as working commercialization, it is Elbit. In July 2023, Foresight signed an exclusive agreement to commercialize its image-processing software, with potential revenue of up to $4 million over five years. In April 2024, Elbit commercially deployed the software to a defense-industry customer for autonomous vehicles, and in 2025 the company recorded $162,000 from this thread. Both the geography note and the main-customer note line up with that: Israel contributed $162,000, and Customer A contributed $162,000.
That is a real anchor point, but it should not be stretched too far. Even Foresight’s most mature thread is still small relative to the group’s cost base, and the filing itself shows the work is not finished. One of management’s 2026 goals is to initiate software-license sales to Elbit’s end customers. So Elbit is no longer an experiment, but it is not yet a scale engine either. It is better understood as first commercial proof than as a broad revenue machine.
KONEC, GINT, and StreamRail Sound Commercial, but the Money Still Sits Ahead
This is where it is easiest to get carried away. The language is sharp: KONEC and GINT are framed around mass-production preparation and commercial deployment, StreamRail is already described through a development-and-commercialization agreement, and the company attaches potential revenue figures that look very large relative to its current size.
In the Korean case, the sequence clearly moved forward: a June 2023 collaboration with KONEC, a December 2024 commercial cooperation agreement, and then an August 2025 three-way agreement with KONEC and GINT to develop an autonomous control system for tractors, with potential revenue of up to $35 million by 2030 and initial sales anticipated as early as 2026. But in the revenue-by-geography note, South Korea contributed only $7,000 in 2025, and the filing does not explicitly link that amount to KONEC or GINT. So what exists today is a commercialization thread with a timeline and a revenue target, not a proven revenue stream.
The same pattern appears in StreamRail. In May 2025, Foresight signed a development and commercialization agreement with the Chinese rail-technology provider, with initial commercialization expected in 2026 and revenue potential of up to $12 million through 2029. The filing even lays out the next step: after development would come manufacturing, marketing, and distribution. Precisely because of that, StreamRail still sits on the pre-revenue side of the line. The agreement exists, the potential exists, and commercialization is already written into the contract language, but the 2025 filing still does not tie any recognized revenue to it explicitly.
That chart captures the gap. Israel already produces $162,000, Japan and France are visible on the map, but South Korea, despite all the commercialization language around KONEC and GINT, barely appears in 2025 revenue. That is not a contradiction. It is a reminder that Foresight’s most advanced pipeline threads are still ahead of meaningful revenue recognition.
Eye-Net in Europe and Japan Is Still Proving Capability, Not Deployment
With Eye-Net, it is especially easy to confuse proof of interest with commercialization. The Japan thread with SoftBank and the Japanese vehicle manufacturer has progressed: there was a paid POC, then an additional development project in March 2024, then technology validation with SoftBank, and by January 2026 the filing was already talking about another validation stage in Japan. But the wording matters. The filing talks about commercial-validation efforts, the exploration of commercial opportunities, and the possibility of integration into infotainment and ADAS. It does not talk about a production order, broad commercial rollout, or a recurring revenue line.
It is also important not to blend different Japanese threads together. The operating-results section ties $40,000 of 2025 revenue to a Japanese company in traffic control and road safety. That is not the same as Eye-Net’s SoftBank and Japanese vehicle-manufacturer thread. In other words, the fact that Japan contributed $63,000 of revenue in 2025 does not mean Eye-Net’s V2X path in Japan has already crossed into commercialization.
Europe carries the same lesson. Eye-Net was selected into the Software Republique program in 2023. In November 2024, it signed an agreement to integrate and conduct real-time testing around Bordeaux. In November 2025, a live trial was launched in Bordeaux together with Renault and Orange, after a prior phase had achieved a 99% detection rate in urban interactions. In February 2026, the live trial was completed successfully, and the parties entered commercial discussions for Bordeaux and additional cities in France. That sounds promising, but the story is still told in the language of trial, validation, and commercial discussions. It is still not a city deployment contract.
The same middle ground applies to the European OEM thread. In July 2025, Eye-Net completed the first phase of a paid POC with a top European vehicle manufacturer to test future commercial integration. The operating-results section confirms that 2025 revenue included successful Eye-Net POCs with a European car manufacturer and a European Tier 1 supplier. That matters because it shows customers are willing to pay for validation. But the filing still describes those paths as successful POCs followed by discussions about next steps, not as serial deployment.
| Thread | What the filing really says | Explicit 2025 revenue | What is still missing |
|---|---|---|---|
| Elbit | Exclusive agreement, commercial deployment to a defense customer, global offer through Elbit | $162,000 | Broader end-customer sales at scale |
| KONEC and GINT | Commercial agreement and mass-production preparation, first sales targeted for 2026 | No explicit tie | A purchase order or identifiable revenue |
| StreamRail | Development and commercialization agreement, initial commercialization expected in 2026 | No explicit tie | Development completion and first sales |
| SoftBank and the Japanese manufacturer | Paid POC, paid development, commercial validation, additional testing | No explicit tie | Real commercial integration |
| Bordeaux and SWR | Integration, live trial, 99% detection, commercial discussions in February 2026 | No explicit tie | A city deployment contract |
| European car maker and European Tier 1 supplier | Successful paid POCs and real revenue | Revenue exists, but no split between the threads | A move from POC to order or rollout |
What the Market Needs to See Now
The key test for the next two to four quarters is already clear. The market does not need another list of collaborations. It needs one additional thread, beyond Elbit, to cross in a measurable way from POC into commercial revenue.
In Korea, that means the promised first sales in 2026 need to become revenue that is actually attributable to KONEC and GINT. In StreamRail, it means the joint development phase has to end and turn into a first sales line. In Bordeaux, it means the move from live trial to commercial discussion must end in an actual deployment agreement. And for SoftBank and the Japanese vehicle manufacturer, it means validation has to give way to real product integration rather than another test phase.
Until that happens, Foresight’s pipeline is more accurately read as a pipeline of commercialization options than as a proven revenue base. That is not a dismissal of the technology. On the contrary, the filing shows the company has several serious threads. But it also shows, in its own language and in its own numbers, that most of those threads still sit between proof and paid proof or early commercialization, not at the point where they can be called a recurring revenue engine.
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.