Matrix after Magic: What actually changes in the group's economics

The main article already argued that Matrix's 2026 story would be judged in two places: the Magic integration and cash quality. This continuation isolates only the first half of that equation. Now that the merger actually closed on February 24, 2026, the question is no longer whether the deal will happen. It is what the deal really does to the group's economics.

The answer is sharper than the headline. Matrix is changing, but less radically than one pro forma slide might suggest. The group does become larger, more international, and more software-heavy. At the same time, it does not stop being a company whose economic core is still built around IT services, projects, cloud, and integration. The immediate economic cost is less about cash and more about dilution, higher customer concentration inside the Magic layer, and a heavier execution burden from day one.

Four points organize the story:

• The closing itself hardly burns cash. At closing, Magic shareholders received 28,861,564 Matrix shares, based on an exchange ratio of 0.5878202 Matrix shares for each Magic share. This is a share deal, not a cash deal.
• The scale jump is real. On a 2025 pro forma basis, revenue rises from NIS 6.239 billion to NIS 8.387 billion, operating profit from NIS 522.1 million to NIS 749.4 million, and EBITDA from NIS 721.2 million to NIS 1.022 billion.
• The mix shifts, but disclosure is still incomplete. Before closing, Matrix's software-products segment was only 6.9% of revenue. After Magic, management is already talking about 14 proprietary software systems and solutions, but there is still no combined segment table showing how much of the new group is truly software versus services.
• Concentration and execution risk do not disappear inside the 6,000-customer headline. Magic itself disclosed that its top two customers were 13.3% of 2025 revenue and its top five were 21%, while Matrix's domestic core operated across more than 3,500 clients with no dependence on any single customer.

What actually closed on February 24

The first thing to put in the right place is the transaction structure. On the closing date, Magic received the merger certificate, all conditions precedent were satisfied, the company was delisted from Nasdaq and TASE, and it became a wholly owned private subsidiary of Matrix. The immediate economics of that step are straightforward: Matrix did not write a large cash check to buy Magic. It paid in shares.

That matters because it means the first-order cost is not cash depletion. It is a change in ownership and in mix. The annual report already framed the issuance as representing about 31.125% of Matrix's issued and outstanding share capital on a fully diluted basis. So anyone reading the deal mainly through the cash line is missing the point. The first risk is not liquidity. It is whether that dilution truly buys a stronger layer of earnings, software exposure, and international activity.

The financing layer was also pre-positioned. In early February 2026, Matrix issued roughly NIS 300 million of Series 2 convertible bonds for five years at a 0.5% annual coupon, with most of the proceeds intended to repay more expensive financial debt at Magic. That reinforces the read that Matrix tried to pay for the transaction mainly with shares while softening the funding cost of the acquired layer.

Against that question, the pro forma only gives a partial answer. On paper, the new group looks stronger: revenue rises 34.4%, operating profit 43.5%, and EBITDA 41.6%. Operating margin also moves from 8.4% to 8.9%. But this should not be read as a clean run-rate view. The pro forma strips out roughly NIS 20 million of merger-related costs recognized at Magic and aligns accounting policies and estimates to Matrix's treatment. It is analytically useful, but cleaner than real life.

That is also why the first 2026 test matters so much. If the first combined numbers come in weaker than the pro forma, that will not automatically mean the deal was wrong. It may simply mean the slide already cleaned up part of the dust while the real integration work is still being done.

More international and more software-heavy, but still services-led

Before Magic, Matrix was already a broad platform, but its mix was still fairly clear. In 2025, 60.1% of revenue came from Israeli IT solutions, consulting and management, 25.8% from cloud and infrastructure, 7.3% from U.S. IT activity, and only 6.9% from software-product marketing and support. Geographically, it also remained highly Israeli: only NIS 571.3 million, 9.2% of revenue, came from outside Israel.

After Magic, the headline changes quickly. The presentation shows a group with more than 17,000 employees, more than 6,000 customers, operations in more than 50 countries, 22% of revenue from outside Israel, and 17% from the U.S. That is a real change. International exposure is no longer a thin layer built mostly around GRC and professional services in the United States. It becomes part of the group's identity.

The software layer also thickens. Magic is described as operating in two main areas: professional IT services on one side, and software solutions on the other, including Low-Code platforms and vertical software solutions. Management already translates that into a marketing message of 14 proprietary software systems and solutions. That matters, because it means Matrix is buying not just more headcount and more projects, but also more product assets.

But this is exactly where discipline matters. There is still no disclosure showing how much of the combined group is actually software. There is no new segment table linking Magic into Matrix's reporting structure. So it is still impossible to claim that the group has become a product company, or even that software has reached a specific weight in the post-close mix. What can be said is that the direction has shifted, and that international and software exposure are both bigger than before, while the service-heavy core remains materially larger than any product layer disclosed so far.

In other words, Magic does not cancel Matrix's DNA. It stretches it. If the market starts pricing the group as if it moved overnight from IT services into proprietary software, that would be a stronger reading than the current disclosure supports.

Concentration did not disappear. It moved inward

The 6,000-customer headline is easy to digest, but it can also mislead. Matrix's domestic core entered the deal from a very diversified position: more than 3,500 customers in Israeli IT solutions and services, mostly large and mid-sized organizations, with no dependence on any single customer. Even in U.S. GRC, the company explicitly says it has no dependence on a particular client, despite serving large financial-sector customers.

Magic brings a different picture. It explicitly says it depends to a large extent on recurring software sales and services to its existing customers. Beyond that, it discloses a visibly higher concentration profile: the top two customers were 13.3% of 2025 revenue, and the top five were 21%. That is not extreme single-customer dependence, but it is clearly a more concentrated customer base than what Matrix usually disclosed in most of its core engines.

This is the number the market may miss on a first read. The promise of Magic is more software, more international reach, and more vertical depth. But layers like that often come with a less dispersed customer base. So Matrix's new economics are not just upgraded. They are also less homogeneous. The group will now contain engines that distribute risk differently: a broader and more diversified domestic layer, and an acquired layer where each large customer matters more.

The good news is that this concentration sits inside a larger group. The less comfortable news is that it will shape how revenue quality should be read. If Magic is meant to be one of the engines that pulls Matrix further toward software and the U.S., then the stability of Magic's customer layer will matter much more to how the whole group gets read.

Accounting will smooth the comparison, but the execution load starts immediately

The least obvious point sits in the accounting layer. Because this is a business combination under common control, the transaction will be accounted for under the As Pooling approach rather than the purchase method. That means Matrix will not remeasure Magic's assets and liabilities at a fresh transaction value, and it will not create original purchase-price differentials. More than that, comparative numbers will be presented as if Magic had already been consolidated in Matrix's books since it was acquired by the controlling shareholder.

That is a critical point, because it may make the numbers look smoother on paper than the business really is on the ground. The accounting comparison will be cleaner than the operational integration. Readers will get a financial picture that looks like the two companies have lived together for longer, while management is still in the middle of connecting leadership, systems, sales layers, products, development centers, and reporting processes.

And that burden is not theoretical. Matrix ended 2025 with NIS 7.493 billion of backlog, of which NIS 5.479 billion is due within the current year. Magic ended the same point in time with NIS 1.629 billion of backlog, of which NIS 1.506 billion is already scheduled for the current year. That means 92.5% of Magic's backlog is meant to turn into revenue within the year, versus 73.1% at Matrix. The group is not getting a long digestion window. It is getting a workload that needs to move through the system quickly.

The quality of that backlog also deserves caution. Matrix already says that a large part of its backlog, especially in Israeli and U.S. IT services, is based on expert-services contracts and framework agreements that can be reduced or stopped on short notice. Magic adds its own warning: its backlog split is based on estimates, including some orders not yet separately received, and remains subject to customer cancellation rights. So putting the two books together does not automatically create higher certainty. It creates a larger revenue book that still depends on execution and discipline.

This is where the operating dimension comes in. Matrix alone spoke about 12,880 software, hardware, engineering, integration, and training professionals before closing. Magic added 4,070 employees at the end of 2025, and the presentation now speaks about more than 17,000 employees worldwide, 3,500 of them outside Israel and 2,500 in the U.S. In addition, the Magic group alone consisted of about 80 active legal entities at the reporting date. That is no longer a counting exercise. It is a stitching exercise.

So the new bottleneck at Matrix is not whether the company knows how to buy growth. It has already shown that it does. The bottleneck is whether it can turn that growth into one group managed through one set of targets, disclosures, and margin discipline, rather than a group whose accounting looks more integrated than the business really is.

Bottom line

After Magic, Matrix really is a different group. It is larger, more international, and carries a thicker software layer. But that still does not make it a classic product company, and it does not mean the economic uplift will arrive without friction. The first cost is dilution, not cash. The new risk is higher customer concentration inside the Magic layer. And the immediate test is execution under load, not just a neat pro forma presentation.

The right way to read 2026 is not by asking whether Matrix simply bought more revenue. It is by asking whether the new group can produce three things at once: a deeper international and software mix, preserved margin discipline, and disclosure good enough to separate what is already embedded in the economic engine from what still lives mainly in the slide deck. If the first combined reports provide those three answers, the deal will start to look like a real economic expansion. If not, the market may quickly discover that more scale does not automatically mean more clarity.