Delek Group: Isracard and ESH, a new financial engine or just more complexity?

The main article argued that Delek's financial leg is now real, but still unproven at the parent level. This follow-up isolates Isracard and ESH because that is where the gap between the subsidiary's economics and what Delek actually reports is the sharpest.

That is the core tension. Isracard itself finished 2025 with a growing business, a larger credit book and positive adjusted earnings. Delek, by contrast, printed a loss in its financial segment. That is not a contradiction. It is what an acquisition year looks like when partial-period consolidation, one-off deal costs and purchase accounting all hit at once.

That is also why ESH matters as more than a strategy headline. The disclosed structure already says something about the kind of financial engine Delek is trying to build. Not a free option, but a platform that can work only if it keeps producing clean earnings, preserves capital headroom and does not turn the parent into a recurring capital provider.

Isracard itself looks better than the line Delek gets

Strip out the deal noise and Isracard does not look like an asset bought too early. It finished 2025 with ILS 3.522 billion of revenue, adjusted net profit of ILS 305 million and an adjusted return on average equity of 9.8%. Transaction volume on cards issued by the company reached ILS 256 billion, while the total credit portfolio climbed to ILS 11.8 billion. This is no longer just a mature card-processing franchise. It is a scaled payments-and-credit platform entering 2026 with real weight.

The growth quality also looks respectable. Non-accruing balances fell to 0.76% of receivables from 0.80% a year earlier. Credit-loss expense fell to 1.30% of average receivables from 1.75% in 2024. In other words, Isracard did not buy growth in the credit book through an obvious deterioration in asset quality. Credit risk is still central here, but the operating engine Delek bought is clearly functioning.

The final quarter of the year was constructive as well. Fourth-quarter revenue rose to ILS 892 million from ILS 825 million, while adjusted net profit improved to ILS 66 million from ILS 54 million. That matters because it shows 2025 was not only a transaction year. The business kept improving even before the ESH angle enters the story.

The credit book explains why Delek sees Isracard as more than a payments brand. Consumer credit on the company's own balance sheet rose to ILS 8.583 billion and commercial credit to ILS 3.259 billion. Together with 4.75 million active cards, including 1.378 million non-bank cards, that is already a platform on which a broader financial product set could be built.

This also fits the management framing. Isracard described a multi-year transition from a card-processing organization into a broader payments-and-credit platform, and said it was examining options for future banking activity. ESH therefore is not a random jump. It sits on top of a strategy that had already been signaled before the MOU, and management later framed it as part of a path toward a broader "Bank Isracard" proposition.

Where the accounting breaks the picture

At Delek, the same financial activity looks very different. In Delek's segment note, the financial segment contributed only ILS 1.823 billion of revenue in 2025, a pretax loss of ILS 261 million and an attributable loss of ILS 68 million. Read on its own, that line can make the acquisition look weak. That is the wrong read.

The first issue is measurement basis. Delek consolidates Isracard only from the date control was acquired, so the ILS 1.823 billion is not Isracard's full 2025. It is only the post-acquisition period. That already makes the view incomplete. On top of that come the one-offs of the deal year, and on top of those sits purchase accounting.

The one-offs at Isracard were substantial. In 2025 the company booked about ILS 62 million of cancellation fees tied to the terminated Menora deal, an increase of about ILS 290 million in the VAT-assessment provision for 2012-2024, about ILS 15 million of prior-year tax expenses, plus costs tied directly to the Delek transaction: a sale bonus to employees of about ILS 45 million, about ILS 15 million of Run-Off insurance and about ILS 13 million of advisory expenses. That is enough to explain how a business with decent underlying earnings can still print a reported loss in the transaction year.

But Delek adds another layer. On acquisition, ILS 865 million of identifiable intangible assets were recognized at Isracard, including ILS 534 million of business customer relationships, ILS 63 million of private customer relationships and ILS 268 million of brand value. In addition, goodwill of ILS 634 million was recorded. Those intangibles are amortized over 10 years, and by year-end ILS 43 million of amortization had already been booked.

This is not a footnote. It is the reason Isracard's economics and Delek's accounting will not look alike in year one. What Delek bought is a business that can earn money, but from the moment it bought control it also loaded itself with amortization, goodwill and transaction costs. So 2025 does not say Isracard is weak. It says the bridge from acquisition accounting to a clean operating print is not complete yet.

ESH is not just a strategic option, it is also a capital transaction

This is where ESH comes in, and the structure matters far more than the headline. The memorandum of understanding is non-binding. It still requires due diligence, binding agreements, board and shareholder approvals, regulatory approvals, amendments to existing agreements between the technology company and the bank, and third-party consents. It includes a 60-day No-Shop period that Isracard may extend by another 30 days. So even on timing, this is still a negotiating framework, not a completed transaction.

On economics, Isracard is supposed to buy 100% of ESH Bank at an implied value of ILS 400 million, not through a clean cash acquisition but mainly through newly issued Isracard shares priced at ILS 15.96 per share. Of that amount, shares worth ILS 250 million would be issued at closing, while shares worth another ILS 150 million would be issued to a trustee and released against milestones tied mainly to deposit balances built at the bank. Beyond that, the bank sellers may become entitled to up to ILS 100 million of cash over five years, based on events to be set in the binding agreement and focused mainly on growth in Isracard's own market value.

And that is before the technology leg. Alongside the bank acquisition, Isracard is expected to buy 25% of ESH OS via a $40 million investment, at a $120 million pre-money valuation. That part is real cash, not Isracard paper.

The implication is straightforward. One clear inference from the disclosed structure is this: the bank leg is funded mainly with dilution, while the technology leg consumes cash. That softens the immediate liquidity hit compared with a full cash bank acquisition, but it does not make the move cheap. It simply moves the price into other forms: dilution, milestones, contingent payments and future capital consumption.

The MOU also contains a small clause with major implications for Delek. Alongside the share issuance to the bank sellers, Isracard may carry out a private placement of up to roughly 5% of its own shares to Delek at the same issue price, in order to allow the controlling shareholder to comply with the Bank of Israel control permit. At that point the parent is already in the story directly. If that clause is activated, Delek is not merely holding a new financial engine. It may also have to write a cheque to preserve the control structure around it.

What this means for Delek

The good news is that Isracard is not entering this discussion from a weak capital base. At year-end 2025, equity attributable to shareholders stood at ILS 3.202 billion. Common equity Tier 1 stood at 10.66% versus an 8.0% regulatory minimum. Total capital stood at 12.4% versus an 11.5% minimum. The leverage ratio was 8.5% versus a 4.5% minimum.

But that is exactly Delek's test. Isracard has room, not endless room. If ESH turns into a binding transaction, the real question will not be only whether the strategic logic is sound. The question will be whether Isracard can widen its product set, absorb a banking layer and still preserve enough capital headroom, rather than swallowing its earnings into regulation, capital build and integration work.

That matters especially for Delek because the parent-level question is not whether Isracard can become a good company. It is whether Isracard can become a good company that also releases value upward. Even Isracard's short-term funding documents are written around capital ratios and distribution discipline, not around open-ended growth. So if the banking move starts consuming capital, the first problem may show up exactly where the parent cares most: in upstream value, not only in the accounting line.

That leads to a simple 2026 test. If Isracard can keep showing reasonable adjusted earnings, preserve asset quality and capital, and turn ESH from a statement of intent into a binding deal without forcing a fast capital response at Delek, then the market can start calling this a new financial engine. If instead the path runs through more dilution, more amortization and, above all, more capital absorbed before any value flows upward, then this is still an extra layer of complexity rather than a clean new engine.

Conclusions

Isracard looks like a better asset in 2025 than Delek's segment line suggests. The business itself grew, credit quality did not show an obvious deterioration, and adjusted earnings remained positive. The problem is that what Delek sees today is filtered through an acquisition year, partial-period consolidation and purchase accounting.

ESH may yet turn Isracard into a broader and stronger franchise. But the disclosed structure already shows that the first price is paid in shares, milestones, capital and regulatory complexity. For Delek, this is not yet "Bank Isracard". It is a test of whether a growing financial platform can move from acquisition-year noise and accounting drag to clean value that actually reaches the parent.