Foresight in 2025: Commercialization Still Does Not Justify the Dilution

Company Overview

From a distance, Foresight can look like another small ADAS name with a few products, a few collaborations, and a familiar list of large partners around it. That is too shallow a read. In practice, this is a tiny listed company, with a market value of roughly NIS 18 million based on an 8.3 agorot share price and about 216 million tradable shares, sitting on top of two very different technology bets: Foresight Automotive in machine vision, and Eye-Net in cellular V2X. The active bottleneck is not whether the company has technology. It is whether it can convert demonstrations, POCs, and validations into repeatable commercial revenue faster than it dilutes shareholders.

What is working now is the breadth of the proof layer. In 2025 and early 2026, the company showed a wider commercialization map than before: Elbit in defense, SUNWAY in autonomous logistics, KONEC and GINT in autonomous tractors, StreamRail in urban rail, Big Bang in drones, and Eye-Net with SoftBank, Software Republique, a European car manufacturer, and Japanese automotive counterparties. This is not an empty headline stack. There are real threads here, some signed agreements, some ongoing validations, and some actual revenue, but still nowhere near a level that can support the cost base.

What can mislead a first read is the mere presence of revenue. Revenue fell to $398 thousand in 2025 from $436 thousand in 2024, while net loss widened to $12.1 million. The second misleading layer is Eye-Net’s funding optics. The subsidiary raised capital at $45 million and then $55 million pre-money valuations in 2025, which creates the appearance of value creation. But that value is not the same thing as clean, accessible value for Foresight shareholders. It sits inside a loss-making subsidiary, under anti-dilution protections, with parent-level warrants, while Foresight’s ownership fell from 100% to 89.36%.

The active bottleneck is therefore economic, not technological: time to commercialization. Management says existing cash should fund operations through August 2026. That means the market is no longer looking for another presentation about potential. It wants to see whether at least one or two of the more mature threads can move from promise to order, and from order to recurring revenue. Until that happens, the capital market remains Foresight’s main financing source.

Foresight’s economic map currently looks like this:

Another useful anchor is headcount. At year-end 2025 the company had 52 full-time employees and 7 part-time employees, down from 63 and 9 a year earlier. Forty-two employees were still in R&D. That is normal for a company at an early commercialization stage, but it is also an implicit admission that this is still a development-heavy organization, not yet a scaled commercial delivery machine.

The actionability constraint also shows up in trading. The latest TASE trading volume was just NIS 5,556, and short interest is effectively absent, with the latest short float reading at 0.00%. That matters. The market’s skepticism is not currently being expressed through an active short base. It is showing up through weak liquidity, a very small market cap, and a history of issuance, reverse split mechanics, and ADS ratio changes.

Events and Triggers

At Foresight, the events of 2025 matter more than the annual numbers themselves, because they show how management tried to buy time for commercialization.

First trigger: Eye-Net raised capital in March 2025 at a $45 million pre-money valuation, and the parent’s ownership fell from 100% to 94.2%. A second round came in December 2025 at a $55 million pre-money valuation, reducing ownership further to 89.36%. Together, the two transactions brought $5.75 million of gross proceeds into Eye-Net. On one hand, that is proof that outside investors are willing to fund the subsidiary. On the other, it is a reminder that the current path to financing runs through sharing future upside with new investors rather than through organic growth in revenue.

Second trigger: The parent also kept using the market sale facility. During 2025 it sold 125,183 ADSs at an average price of $34.38 per ADS and received about $4.165 million net. Then, after year-end, it sold another roughly 223,850 ADSs in early 2026 at an average price of just $3.99 per ADS for net proceeds of $866 thousand. That gap is not a technical detail. It shows that capital market access is still open, but on much weaker pricing.

Third trigger: On the commercialization side, management is spreading its bets across shorter-cycle end markets. In defense, there is an exclusive agreement with Elbit to commercialize image-processing software, with potential revenue of up to $4 million over five years, and Foresight already recorded $162 thousand of revenue from that thread in 2025. In heavy equipment and agriculture, the KONEC and GINT agreement carries disclosed potential revenue of up to $35 million through 2030, with initial sales anticipated as early as 2026. In urban rail, the StreamRail agreement carries disclosed potential revenue of up to $12 million through 2029, with commercialization also expected to begin in 2026. These are stronger anchors than generic pilot announcements, but they are still not the same thing as a proven order flow.

Fourth trigger: Eye-Net continues to progress, but the story is still not economically closed. The company points to advanced collaboration with SoftBank, work with a Japanese vehicle manufacturer and a European car maker, an agreement with Software Republique, and the Bordeaux live trial that followed a prior phase with a reported 99% detection rate. This reads well, but it remains clearly distant from Foresight’s income statement. What is still missing is a commercial deployment agreement that shows up as booked revenue rather than another validation step.

Fifth trigger: Listing pressure already became real. The company received Nasdaq minimum bid notices in September 2024 and March 2025, regained compliance through a 1-for-7 reverse split in August 2025, and then changed the ADS ratio in February 2026 from 30 to 90 ordinary shares per ADS. This is not theoretical. It means the market has already seen genuine listing stress, and management addressed it through capital-structure mechanics, not through operating proof.

Efficiency, Profitability and Competition

The main point here is not that operating loss improved a bit. It is that the business economics still have not really moved. Revenue fell 8.7% to $398 thousand, gross profit was $258 thousand, and R&D, sales and marketing, and G&A together still totaled $12.4 million. Even after some cost reduction, the company is still running an operating cost base that is more than 30 times annual revenue.

There is partial efficiency progress. R&D expense fell 5.6% to $8.629 million, and G&A fell 2.4% to $2.613 million, which fits the lower headcount. But sales and marketing expense rose 6.9% to $1.197 million while revenue declined. That is not automatically wrong, because the company may be trying to push commercialization in several verticals at the same time, but it does show that the commercial machine is not yet generating operating leverage.

It is also important not to overread the operating loss improvement. Operating loss narrowed from $12.676 million to $12.181 million. But net loss actually widened to $12.103 million from $11.138 million, mainly because 2024 benefited from an unusually strong financial income line of $1.538 million, driven largely by the Rail Vision sale and revaluation. In 2025, net financial income dropped to just $78 thousand. In other words, the 2024 accounting noise faded and the underlying weakness of 2025 became easier to see.

Revenue quality matters as well. Management says 2025 revenue came mainly from Elbit commercialization at $162 thousand, from completion of a POC with a Japanese traffic control and road safety company at $40 thousand, and from successful Eye-Net POC projects with a leading European Tier 1 supplier and a leading European car manufacturer. That is the key point: much of the revenue still comes from milestones, testing, integration work, and one-off project phases rather than from a recurring deployment model.

The geographic revenue split does show breadth, Israel at $162 thousand, France at $74 thousand, Japan at $63 thousand, the United States at $32 thousand, South Korea at $7 thousand, and other at $60 thousand. But this is thin breadth, not depth. The largest customer accounted for $162 thousand, about 41% of annual revenue. That reinforces the point that the company has not yet built a broad recurring customer base, even if the list of names around the pipeline looks stronger than before.

On competition, management is clearly trying to play where the sales cycle is shorter. That is stated almost directly in the business overview: heavy industry, agriculture, drones, rail, and defense are meant to offer faster commercialization than the long automotive passenger-vehicle race. That makes strategic sense for a small company, because it avoids the part of the market where large OEMs and Tier 1 suppliers compress everyone on price and timing. But as of year-end 2025, that strategic pivot had not yet translated into clean unit economics. The company is still competing to prove commercialization, not to expand margins in an already scaled business.

Cash Flow, Debt and Capital Structure

This is the center of the story. Foresight does not really have a classic debt problem. There is no large bank debt stack and no covenant edge that is about to be breached. But that is exactly the point. The capital market is Foresight’s bank. That is why the right framework for 2025 is all-in cash flexibility, how much cash is left after actual cash uses.

The numbers are straightforward. The company entered 2025 with $7.182 million of cash, restricted cash, and similar balances. During the year it used $10.453 million in operating cash flow, spent another $25 thousand on fixed assets, and received $9.476 million from financing activities. After a $109 thousand foreign exchange effect, it ended the year with $6.289 million. Without financing, the year would not have ended with a cash balance. It would have ended with a hole.

Management states this directly as well: existing cash is expected to fund operations through August 2026, while the financial statements still carry a going concern warning because of dependence on external funding and the lack of significant recurring revenue. The window is therefore short. The report was filed on March 25, 2026, which means management itself is effectively describing only a few months of runway absent a sharp revenue step-up or another financing source.

The capital-structure picture is even sharper. Issued and outstanding ordinary shares jumped from 72.7 million at the end of 2024 to 140.6 million at the end of 2025. That is almost a doubling in one year, even after the reverse-split adjustments. The statement of changes in equity makes the mechanics visible: share issuance, warrants, RSUs, and subsidiary-level share issuance all played a role.

This dilution did not only happen at the parent level. It also happened inside Eye-Net, where management chose to fund the bridge period by bringing in outside investors at higher subsidiary valuations. On one hand, that is better than fully funding the subsidiary from the parent balance sheet. On the other, it shifts part of the future upside to new investors under anti-dilution protections and reduces the public shareholder’s claim on the asset that is supposed to sit at the heart of the value thesis.

Another underappreciated point is the non-controlling interest balance. It was positive $664 thousand at the end of 2024. By the end of 2025 it had moved to negative $1.122 million. That does not mean the Eye-Net financing rounds were worthless. But it does show that, on a consolidated basis, the rounds have not yet solved the subsidiary’s loss profile. In other words, value may have been created on paper at the round-valuation level, but it has not yet turned into clean surplus equity at the group level.

The company also carries lease liabilities of $1.512 million, with future lease payments totaling $1.775 million. That is not a balance-sheet breaker, but it is a reminder that in a company this small every operating dollar matters. Lease cash payments of $476 thousand in 2025 are part of the real cash picture, not a footnote.

Outlook and Forward View

Before going into detail, here are the five most important takeaways for 2026:

• The real economic event of 2025 was not revenue growth. It was capital-structure change. Eye-Net raised capital, the parent raised capital, and shareholders bought time through dilution.
• $398 thousand of revenue still does not support a business running more than $12 million of annual operating expense. Even after cost cuts, the company remains far from proving its unit economics.
• Much of the bullish story still sits in round valuations, POCs, and contractual potential, not in reported commercialization.
• 2026 is not a breakout year by default. It is a proof year with a short deadline. Management itself points to runway only through August 2026.
• The central risk is not that the technology is uninteresting. It is that commercialization may not happen fast enough to reduce the need for another round of dilution.

What has to move from proof to revenue

Foresight has several threads that could matter, but they are not all equal.

Elbit is the cleanest thread right now. Not because it is big, $162 thousand of revenue is still small, but because it has already crossed the line from statement to commercialization. If Elbit remains around the same scale in 2026, the market will read that as evidence that even a relatively advanced thread does not automatically become a revenue engine. If the amount grows meaningfully, that would be the first real sign that the company can move up a level.

KONEC and GINT are a different type of thread. Here the disclosed potential is up to $35 million through 2030, with first sales anticipated in 2026. That headline number attracts attention, but it is long-dated and conditional. What matters more than the $35 million is whether 2026 brings the first order, first delivery, and the beginning of revenue recognition tied to a production line rather than a development phase.

StreamRail sits in a similar category. Again, there is disclosed potential, up to $12 million through 2029, and commercialization is expected to begin in 2026. If the company can show that the product is moving through development and into actual delivery, that would add a new engine outside road vehicles, potentially in a vertical that is more manageable for a small company because the operating environment is more structured.

At Eye-Net, the core issue is stage quality. There are many milestones, SoftBank, Bordeaux, UTAC, a European OEM, a Japanese OEM, Continental through co-pace, but there is still very little money visible on the income statement. That is why the market does not need another validation on accuracy or compatibility in 2026. It needs to know when a collaboration becomes a commercial agreement, and whether that agreement creates booked revenue rather than another phase.

Why 2026 is a proof year, not a dream year

Management talks about expected revenue growth in the coming years. That is fair enough. But the right way to read it is different: this is a proof year under capital pressure, not a breakout year resting on an already proven engine. The company has several promising commercial seeds, a limited cash balance, and a clear history of financing itself through the market. Put simply, it has to show something real before investors are asked to pay for another equity bridge.

Two signals will matter most over the next few quarters. The first is revenue quality, whether the company starts to move from POC income into deployment, licensing, system supply, or service revenue that repeats. The second is financing quality, whether the company can get through the year without another raise that brings more dilution, repricing, or another redistribution of upside.

What the market may miss on a first read is that Foresight does not need dozens of deals to change the screen. At this market value, even one meaningful commercial order could shift the interpretation materially. But the opposite is also true: another quarter or two of technology headlines without meaningful revenue could push the whole discussion back to the old question, how much time is left until the next raise.

Created value versus accessible value

This is where it makes sense to separate two kinds of value. One is reputational and technological value, strong partners, higher Eye-Net round valuations, attractive contract potential, and external validation. The other is accessible value for Foresight shareholders. That second category still looks much harder.

For Foresight shareholders to truly benefit from what is being built at Eye-Net, several things need to happen together. Eye-Net has to generate revenue that reduces the need for more rounds. Foresight has to preserve enough ownership for the economics to matter. And the public parent has to get through 2026 without further large dilution at weak prices. Until those things happen together, the subsidiary valuations are mostly optionality, not cash flowing back up the chain.

Risks

The first and most obvious risk is financing risk. Not in the sense of bond refinancing or bank covenant pressure, but in the sense that the company still lives on equity funding. The going concern warning, the runway into August 2026, the ADS sale in early 2026, and the post-balance-sheet full exercise of the Series A warrants all point in the same direction: the company still depends on the capital markets.

The second risk is commercialization risk. The report is full of names, collaborations, tests, and plans. But annual revenue is still only $398 thousand. As long as that remains true, the market may continue to read many of the announcements as proof-of-capability marketing rather than as evidence of a scalable revenue base.

The third risk is dilution. This is no longer a future worry. The share count nearly doubled in one year, Foresight’s stake in Eye-Net fell to 89.36%, and both Eye-Net rounds came with anti-dilution protections subject to a $30 million valuation floor. That means that if the next round, if one comes, is weaker, the price may not just be additional dilution. It may also involve another redistribution of economics toward investors who are already inside.

The fourth risk is listing and liquidity risk. The company already had to solve a Nasdaq minimum bid issue through a reverse split and then changed the ADS ratio. At the same time, TASE liquidity remains very weak. So even if the operating story improves, shareholders still face a real actionability constraint.

The fifth risk is FX exposure. The company explicitly states that it does not hedge its currency exposure, and that a 5% or 10% move in the U.S. dollar against the shekel would have changed 2025 operating expenses by 4.38% or 8.76%, respectively. In a company with such a small revenue base, that is not a marginal issue.

Conclusions

Foresight entered 2026 with more technological proof, more partners, and more time than it had a year earlier. But it did not enter 2026 with a proven commercial model. What supports the thesis right now is the possibility that one of the more advanced threads turns into real revenue in time. What blocks a cleaner thesis is that the company is still buying that time through dilution.

Current thesis: Foresight did a better job in 2025 of buying time and building commercialization options, but it still has not shown that commercialization is moving fast enough to justify the cash burn and the dilution.

What changed versus the prior read? There are now more assets that look monetizable, especially inside Eye-Net and in shorter-cycle industrial verticals, but there is also less direct ownership over part of that upside and stronger evidence that capital markets remain an integral part of the operating model.

Counter-thesis: the market may be deeply underpricing the option value here. If Elbit, KONEC and GINT, StreamRail, and one of the Eye-Net tracks mature in parallel, a market value of roughly NIS 18 million could look far too low relative to the opportunity set.

What could change market interpretation in the short to medium term? A clearly visible first commercial order from one of the larger threads, meaningful growth above the $162 thousand already coming from Elbit, or, on the negative side, another financing round at weak pricing that reminds the market the distance to real revenue is still too wide.

Why does this matter? Because Foresight is no longer just an early technology story. It is a classic test of whether a very small listed company can turn a strong string of technical validations into accessible shareholder value before its capital structure erodes the upside.

What must happen over the next 2 to 4 quarters for the thesis to strengthen? At least one commercialization thread has to move from POC into recurring revenue, and the company has to get through the year without another major dilutive round. What would weaken it? Another sequence of technology announcements without a clear revenue step-up, or more financing that comes at the cost of another redistribution of value.