Delek Group: Debt, dividends and buybacks, how much real flexibility is left at the parent?
On paper, Delek ended 2025 with NIS 9.579 billion of distributable profits. In practice, the parent finished the year with NIS 1.427 billion of cash, about NIS 5.18 billion of external debt, a control acquisition at Isracard, and another reminder that the real test is not the legal ability to distribute, but the cost of distributing aggressively.
The main article argued that the core issue at Delek is no longer whether value is being created in the assets, but how much of that value is truly accessible at the parent. This follow-up isolates the parent layer only, because this is where the key valves sit. A strong year at the group level can still translate into two very different realities for shareholders once it reaches the holdco.
The short answer is that Delek still has real parent-level flexibility, but it is much narrower than the headline distributable-profits number suggests. At the end of 2025 the company had roughly NIS 9.579 billion of distributable profits. That is a very large legal room number. But the same year also included about NIS 1.035 billion of dividends paid to shareholders, roughly NIS 34 million of buybacks executed, about NIS 391 million of cash interest paid, and about NIS 1.655 billion invested in a held company, in the same year in which the Isracard control deal was completed.
What is working now should be stated clearly. In the separate parent statements, Delek ended 2025 with about NIS 1.427 billion of cash and another roughly NIS 14 million of short-term investments, with equity of about NIS 10.083 billion, positive working capital of about NIS 423 million, and a solo equity-to-assets ratio of about 64%. This is not a stressed parent balance sheet. In addition, the company actively refinanced through 2025, pushed maturities out, and by the annual-report approval date also had bank facilities totaling roughly NIS 500 million that had not yet been drawn.
That is precisely why March 2026 matters. In that month the board approved a new NIS 100 million buyback plan, while the company also moved again toward a fresh bond issue, with Midroog assigning an A2.il rating, stable outlook, to up to NIS 500 million par value in new Series MB and Series MG bonds. That combination says something important. Delek is not acting out of immediate pressure, but it is also not behaving as if accounting distributable profits are the same thing as truly free parent cash. Debt-market access remains part of capital-allocation policy, not just a treasury footnote.
The company is not short of legal room, it is short of fully free surplus
The first distinction to make is between distributable profits and free cash at the parent. They are not the same thing, and at Delek the gap is especially visible.
| Parent-level metric | 2025 | Why it matters |
|---|---|---|
| Cash and cash equivalents | NIS 1,427 million | The immediate liquidity layer at the parent |
| Short-term investments | NIS 14 million | A small add-on to immediate liquidity |
| Current external bank and bond debt | NIS 902 million | The next 12 months of parent obligations |
| Long-term external bank and bond debt | NIS 4,276 million | The main debt stack still sits at the parent |
| Dividends received | NIS 1,636 million | The main source of cash moving up from below |
| Cash flow from operations | NIS 1,408 million | The incoming cash picture before the big capital-allocation uses |
| Dividends paid to shareholders | NIS 1,035 million | A major return of capital already completed |
| Buyback executed | About NIS 34 million | The signal exists, but the actual spend was modest |
That chart is not a distress story. It is a proportions story. The parent does have a meaningful liquidity cushion, but it sits against a debt pile that is still several times larger. So the real question is not whether a one-off distribution is possible. The real question is whether the company can keep doing dividends, keep a buyback option open, fund new investments, and refinance debt without becoming too dependent on whatever continues to flow up from the subsidiaries.
This is where the cash framing matters. On a normalized basis, the parent generated about NIS 1.408 billion of operating cash flow in 2025, with that number supported by about NIS 1.636 billion of dividends received. In other words, even at the parent, the cash engine is not some rich standalone operating platform. It is cash coming up from below. On an all-in basis, after shareholder distributions, buybacks, and cash interest, there was hardly any real surplus left. Once the investment in a held company is added, the gap becomes large.
This is the core of the story. The chart is not trying to replicate every cash-flow line. It is isolating the parent-level capital-allocation bridge. The conclusion is clear: aggressive distributions are possible at Delek, but they are not being funded out of a wide, comfortable surplus left over after everything else. They coexist with refinancing, with cash coming up from subsidiaries, and with conscious decisions about how much liquidity the parent wants to keep on hand.
That is also why distributable profits are important, but not decisive. They open the legal door. They do not finance the walk through it.
Where the real distribution constraints actually sit
It is too easy to read Delek only through the bond indentures. In practice there are several layers of constraint, and they do not all sit in the same place.
| Layer | Threshold or restriction | What it means in practice |
|---|---|---|
| Bond indentures, distribution gate | Equity of at least NIS 3.5 billion and solo equity-to-assets of at least 22%, with no breach and no acceleration event | A distribution is allowed only if the parent remains strong after paying it |
| Bond indentures, ongoing maintenance | Minimum equity of NIS 3.25 billion and minimum solo equity-to-assets of 18% | This is the creditors' hard protection floor |
| Bond indentures, coupon step-up | Equity below NIS 3.35 billion or solo ratio below 20% | There is an economic cost even before a covenant is actually broken |
| Isracard control permit | Solo equity-to-assets of at least 50% and consolidated ratio of at least 35%, with cure zones at 45% and 30% | Isracard adds a regulatory layer to the parent, not just another asset |
| Isracard pledge limits | At least 20% of the control means in Isracard must remain unencumbered | Even if management wants to lever the new asset, there is a structural limit |
| Isracard upstream distributions | Subject to capital, liquidity, leverage, and bank-supervision approval or non-objection when required | The new financial asset is not an automatic cash pipe to the parent |
What matters here is not whether Delek is close to the wall. It is not. At year-end 2025 the company explicitly said it was in compliance with all indenture conditions, with equity of about NIS 10.083 billion and a solo equity-to-assets ratio of about 64%. It also stated that it was in compliance with the Isracard control-permit conditions. So the active bottleneck is not covenant proximity. It is capital-allocation discipline.
Still, Isracard changes the quality of the parent's flexibility. In July 2025 Delek invested about NIS 1.358 billion in Isracard in exchange for the newly issued control stake. Isracard's special dividend of about NIS 1.2466 billion was paid in August 2025, but Delek's share of that dividend, based on the stake it held before the deal closed, was only about NIS 60 million. Put differently, the new asset required a large cash outlay at the parent, but did not create an offsetting upstream cash stream of similar scale at the same moment.
That matters even more because of the nature of the asset. In March 2026 Isracard reapproved a dividend policy of up to 40% of half-year net profit, but explicitly subordinated it to legal requirements, regulatory approval or non-objection when needed, and its own capital and leverage targets. In practice, because of the loss recorded in the first half and in full-year 2025, Isracard did not distribute a dividend out of 2025 profits. So Delek acquired an asset that may become an upstream cash source in the future, but for now it also adds a regulatory layer and has not yet contributed an ongoing distribution stream from the acquired year.
That means the group may look more diversified, but the parent's access to value still depends mainly on the older channels through which cash already knows how to move upward.
Buybacks are more of an option than an engine
If one looks only at the headlines, it is easy to think Delek is running an aggressive buyback policy. If one looks at execution, the picture is much more measured.
The June 2024 buyback plan authorized up to NIS 200 million, but roughly NIS 95 million was actually used, or about 47.42%. The March 2025 plan authorized up to NIS 105 million, but only about NIS 34 million was executed, or about 32.71%. Then on March 24, 2026, the board approved a new NIS 100 million plan through year-end 2026, while explicitly stating that the funding could come from the company's own resources and or bank credit, if obtained.
That is a material point. At Delek, buybacks do not function like an automatic capital-return conveyor belt. They are better understood as an optional tool the board likes to keep open. This allows management to signal that the shares look attractive, without committing to a rigid pace of execution and without giving up the ability to slow down, stop, or redirect the cash if needed.
The wording of the March 2026 immediate report reinforces that interpretation. The board did state that the legal distribution tests were met, with distributable profits of about NIS 9.329 billion at the approval date, but management was given very broad discretion over timing, price, mode of execution, and which group entity would actually buy the shares. Anyone looking for a fixed, mechanical capital-reduction program will struggle to find one here. Anyone looking for another flexible capital-management lever will find exactly that.
That is also where the risk sits. A buyback looks shareholder-friendly when it rests on unquestioned surplus cash. It looks less convincing when it sits alongside a desire to preserve liquidity, fund new strategic moves, and remain a frequent issuer in the bond market.
Refinancing is not a footnote
In 2025 Delek raised roughly NIS 3.298 billion through bond issuance and repaid about NIS 2.404 billion of bonds. Series M was issued in February and expanded in June. Series MA was issued in September and expanded again in December to fund the early redemption of Series LH. This is a classic maturity-management and funding-cost story. But it also means shareholder returns are being managed inside a rolling refinancing framework, not above it.
There is also a less-discussed layer. In August 2025 a foreign bank extended a NIS 238 million loan to the company, secured by all of the InPlay shares it held. In addition, in August and October 2025 the company signed bank facilities totaling about NIS 500 million, secured by part of its NewMed units, and as of the annual-report approval date those facilities still had not been drawn. In other words, even when the balance sheet looks calm, Delek continues to build financing layers against specific assets.
March 2026 completed the picture. Midroog assigned an A2.il rating, stable outlook, to up to NIS 500 million par value in new bond series, and the company announced its intention to hold an institutional tender that same day for two new bond series, with proceeds earmarked for ongoing operations. That does not read like distress. It does read like a management team that does not want shareholder returns to come at the expense of continuous access to debt markets.
This is a key point in understanding Delek as a holding company. Financial flexibility is not measured only by how much cash sits in the parent account on the reporting date. It is also measured by the parent's ability to keep rolling debt, preserve a liquidity cushion, and stay capable of funding both a new strategic move and a distribution. At Delek, as of now, the positive answer still depends on debt markets remaining open and on subsidiaries continuing to move cash up.
Bottom line
Delek still has real room to maneuver. The covenants are not close, liquidity is reasonable, and the company showed a clear ability through 2025 to refinance and improve the debt stack. But that is not the same thing as saying there is a very large pool of truly free capital sitting at the parent waiting to be distributed.
The misleading number is NIS 9.579 billion. The truer numbers are closer to NIS 1.408 billion of operating cash flow, NIS 1.636 billion of dividends coming up from below, and the gap that opens once the parent's actual cash uses are added back into the picture. That is why the main risk here is not a single acute funding event. The main risk is distributing too aggressively and gradually making the parent story more dependent on continued upstream dividends and on a permanently open bond market.
For investors, the implication is straightforward. Delek can keep paying dividends, and it can probably keep approving buybacks as well. But every such decision should be tested not against the legal distributable-profits line, but against three other questions: how much cash actually moved up to the parent, how much of it is already committed to be sent back out, and how much flexibility remains after debt, interest, and the next round of investment are taken into account.
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.