Shva: How Much of the Margin Pressure Is Temporary, and How Much Is Structural

The main article argued that Shva's fee per transaction no longer covers its new cost base. This follow-up isolates that cost base itself. That matters because the 2025 filing compresses three different kinds of pressure into one headline: Masav-separation cost that does not read like a passing event, a modernization wave whose cash intensity may ease, and a capitalization layer that defers part of today's economic burden into tomorrow's amortization.

The right way to read 2025 is not as just another year of elevated spending. It is a year in which Shva's cost structure is being rewritten. The separation from Masav is no longer an open dispute, but an agreed framework running through June 30, 2029. At the same time, the company pushed forward core modernization, bought a new core computer, and continued to capitalize a meaningful portion of development effort into intangible assets. A reader who looks only at operating expenses misses part of the story.

Three findings need to be held together. First: most of the separation burden is already in the numbers. The impact on 2025 results was about NIS 16 million, versus an estimated NIS 17 million to NIS 18 million in 2026 and about NIS 19 million per year by the end of the separation process. In other words, the path to 2029 still matters, but the big step-up has largely already happened. Second: the more temporary layer sits in the investment stack. In 2025, Shva added NIS 26.45 million to intangible assets and another NIS 15.02 million to property, plant, and equipment, together about NIS 41.5 million. Third: temporary in cash does not necessarily mean temporary in margin. Depreciation and amortization already rose to NIS 14.78 million from NIS 9.74 million, so even if the spending peak is behind the company, the income statement may not clean up immediately.

The Masav Separation: Structural, Longer, and Already Near the Run Rate

The filing does not describe the Masav separation as a one-off legal or advisory line. Quite the opposite. Shva says explicitly that the 2025 impact, about NIS 16 million, came mainly from higher wages and related costs as services previously provided to Masav were reduced, net of the portion capitalized to intangible assets, and from higher other operating expenses, mainly hardware and software maintenance plus depreciation and amortization, as Masav gradually exited shared infrastructure. That is already the definition of a structural cost line.

The timeline matters as well. The original outline called for a full separation of shared links by the end of 2027. But after Masav requested a change, the Competition Authority approved a postponement, and the final deadline for separating all shared ties was pushed to June 30, 2029. That cuts in two directions. On one hand, the pressure is not disappearing in 2026. On the other hand, if 2025 already carries about NIS 16 million of burden and the steady-state annual cost is expected to be about NIS 19 million, most of the transition is already embedded in the cost base.

The analytical implication is sharp: the separation is more structural than temporary, but also less of a future cliff than a simple "through 2029" headline may imply. What remains ahead is mainly another few million shekels of annual tightening, not a second doubling of the burden. Anyone waiting for a year in which the separation expense simply disappears is probably reading the filing incorrectly. But anyone assuming each year through 2029 will deteriorate materially versus the year before is also likely overstating the slope.

The permanent agreement signed in December 2024 reinforces that reading. It anchors a framework in which each company continues to bear the ongoing costs of each system until its exit date, and for hardware and software remaining with one side there will be no further balancing payment. Put simply, this looks more like a cost that becomes locked into Shva's standalone operating model, and less like an expense that is later reversed.

The quick orientation table looks like this:

Modernization: The Cash Peak May Be Temporary, but Amortization Has Started

If the Masav separation is the structural layer, modernization is the layer where it is easiest to get the read wrong in both directions. On one hand, as of the report date the cost of modernization and related development stood at about NIS 16 million, and the company says significant additional investment will still be required in future years. On the other hand, 2025 already shows clear signs that part of the burden is a spending peak rather than a permanently recurring cash run rate.

That is very visible on the balance sheet. Net property, plant, and equipment rose to NIS 31.06 million from NIS 21.83 million, and the company explains that the change came from the purchase of a new core computer net of depreciation. The property note states that the new core computer cost about NIS 11 million, and total PP&E additions in 2025 came to NIS 15.02 million, versus just NIS 2.01 million in 2024. At the same time, net intangible assets rose to NIS 49.79 million from NIS 29.71 million, with NIS 26.45 million of additions from purchases and capitalized development costs during the year.

On an all-in cash flexibility basis, this is probably the more temporary layer. In 2025 Shva absorbed a heavy wave of capability build-out, infrastructure purchases, and internal development spending. There is no evidence that this exact pace must recur every year. But on a reported-margin basis, that temporary character is only partial. The moment development and infrastructure cost stop being pure cash and become assets is also the moment they start to return gradually through depreciation and amortization.

And that is exactly what already happened in 2025. Depreciation and amortization rose to NIS 14.78 million from NIS 9.74 million. It matters even more to break down where that move came from:

The truly interesting point is that the jump came overwhelmingly from amortization of intangible assets, not from leases and not mainly from ordinary hardware depreciation. In other words, the pressure that began as capitalization of internally developed software has already started to come back into operating profit. That is why a slower investment pace does not automatically mean an immediate margin snap-back.

The Capitalization Layer: Where the Income Statement Still Looks Cleaner Than the Full Economics

This is the most delicate part, and probably the most important one. The company is not hiding costs. It is simply following standard development accounting, where part of the effort is not recorded immediately as expense but as an asset amortized over time. But for anyone reading the margin line, that distinction matters a great deal.

In the separation section, the company states explicitly that the impact on results is presented net of the portion capitalized to intangible assets. In the modernization section it adds that development costs in sections 1.2 through 1.4 were capitalized in 2022 through 2025 into intangible assets in a cumulative amount of about NIS 43 million. That means part of the effort tied to separation and modernization did not hit the 2025 expense line in full, but was deferred forward.

The intangible-asset note shows how material that is. Out of NIS 26.45 million of additions in 2025, NIS 25.56 million came from internally developed software and only NIS 0.889 million from separately purchased software and licenses. This is not a small layer of purchased tools or external licenses. It is internal development effort moved onto the balance sheet. By year-end, the net carrying value of internally developed software stood at NIS 47.69 million, roughly the scale of a full year of operating profit.

That is why the operating-expense line, which rose by NIS 14.34 million to NIS 110.20 million, tells only part of the story. The second part sits on the balance sheet and will flow back through profit and loss over time. That is also why the reading that "2025 was unusual, so 2026 should clean up by itself" is too simplistic. Some of the unusual load does belong to a spending peak. But another part has already been locked into a future amortization tail.

Put differently, 2025 is not just a year of high spending. It is also a year in which Shva pushed part of the price of this structural transition into future periods. Anyone looking only at EBITDA or only at current operating expenses gets too clean a picture of the cost reset.

Leases Also Show What Is Structural and What Is Not

The lease note matters here not because it is the center of the story, but because it sharpens the scale. After Masav left the shared offices, its lease rights in the office were assigned to Shva. In addition, at the beginning of 2025 an addendum was signed for the backup site, such that Shva's share of the rental cost stands at 67.5% starting in December 2024. So even in the physical operating layer one can see how the separation transfers more recurring cost onto Shva.

But the scale still needs discipline. Total lease expense recognized in profit and loss in 2025 was NIS 3.36 million, and total negative cash flow for leases was NIS 3.12 million. Those are meaningful figures, but they are much smaller than the overall separation burden and the technology-investment wave. So leases confirm that structural cost ownership has shifted upward, but they are not the main explanation for margin compression.

That is exactly why they are analytically useful. They show that not every structural component is the same. Some pressure, like leases, looks recurring but relatively contained. Another piece, like the Masav separation itself, is recurring and material. And a third piece, like capitalized development, looks temporarily lighter in the expense line but can turn into a long amortization tail.

So How Much Is Temporary, and How Much Is Structural

The cleanest summary looks like this:

That is why the right reading is neither "the pressure is temporary" nor "everything has become structural." Part of the strain genuinely belongs to a build-out phase, which means cash generation can improve before the income statement fully cleans up. But the new cost floor is already higher even before the most unusual spending disappears, and the next few years will have to absorb both a structurally higher operating base and an amortization tail that has already been built on the balance sheet.

That also leads directly back to the main article's thesis. If a tariff update comes, it is not just there to cover one difficult investment year. It has to cover a new cost base. And if technology investment slows, that may help cash flow before it helps reported margin. Anyone looking for a very fast and perfectly clean recovery in the next filings may miss that distinction.