ALFA, when the data center moves from paper uplift to a real NOI engine
At the end of 2025 ALFA sits on Levinstein’s books at only ILS 70.6 million, with 0% signed leases and only 7% financial completion, while the March 2026 presentation already points to expected NOI of ILS 103 million and a completed value of ILS 1.079 billion. So the real issue is not whether there is potential, but how fast and how cleanly the story moves from an appraisal model to actual NOI.
What This Follow-up Is Isolating
The main article framed ALFA as one of the assets that could change Levinstein’s scale over the coming years. This follow-up isolates a narrower question: when does the data center stop being paper uplift on land and start becoming a real NOI engine.
The short answer is that ALFA has already moved past the idea stage, but it is still far from the income-producing stage. At the end of 2025 the company’s share in the asset is carried at fair value of ILS 70.6 million, after an annual revaluation gain of ILS 5.8 million, with 7% financial completion and 0% of the space under binding lease agreements. At the same time, the March 2026 presentation already frames phase 1 at 18 to 21 MW IT, expected NOI of ILS 103 million, and completed value of ILS 1.079 billion for the full project. The gap between those two states is the whole story.
What can mislead a superficial reading is the mixing of three very different layers: the value sitting on the balance sheet today, the as-complete valuation model, and the NOI that should exist only after completion and operation. As long as those three numbers are read as if they belong to the same layer, ALFA looks closer than it really is.
Four points matter from the start:
- ILS 70.6 million is the current balance-sheet position for the company share, not an as-complete value.
- ILS 103 million of NOI and ILS 1.079 billion of completed value are 100% project numbers, not the company share.
- Phase 1 has moved into permitting and construction, but as of the end of 2025 there were still no binding lease agreements.
- Phase 2 is not part of the locked model yet. It is still in zoning approvals, and it may also be replaced by commercial and employment uses. So the two routes should not be counted together.
What Has Already Been Proven, And What Has Not
The positive point is that ALFA is no longer raw land. In November 2022 the company signed an agreement to sell an undivided 50% interest in the site to an unrelated third party owned by an international group with experience in building and operating data centers, for about ILS 43.75 million. The completion conditions were met in June 2024, and at the same time the parties signed a 50-50 joint-venture and cooperation agreement. In the March 2026 presentation the partner is identified as NED from the UK-based KAO DATA group.
There is also real execution progress now. The note in the annual report says that in December 2025 the project received a building permit and construction work began. That matters, because it moves ALFA from pure planning into a stage where real project capital is already being deployed.
But this is exactly where the reading still needs discipline. In the annual report the same asset is still presented as one whose construction is expected to be completed in stages during 2027 through 2029. The March 2026 presentation already narrows phase 1 to an expected completion in the fourth quarter of 2026. That is not just a schedule update. It is a shift from a long-dated development narrative to a narrative of an asset that is supposed to begin facing proof points almost immediately. Once management tightens the framing like that, the burden of proof rises with it.
One more detail matters. The presentation itself switches between two measurement bases. In the short-term development pages ALFA is shown as 10,000 square meters because that is the company share. In the dedicated ALFA pages phase 1 is shown at about 20,000 built square meters because that is the 100% project basis. That is not a mistake, but it is a reading trap. Anyone who misses the measurement basis can very easily double-count both area and value.
How The Value Model Is Built, And Where The Reader Can Slip
The material-under-construction table gives the end-2025 position: the company’s share in ALFA sits at fair value of ILS 70.6 million, carrying cost of ILS 29.0 million, annual revaluation gain of ILS 5.8 million, 7% financial completion, and 0% of space under binding lease agreements. In plain language, year-end 2025 still reflects early land uplift and the first stage of execution, not an income-producing asset.
The March 2026 presentation, by contrast, already lays out the full phase-1 model: expected total cost of ILS 774 million including land, expected NOI of ILS 103 million, completed value of ILS 1.079 billion, expected developer profit of ILS 305 million, and a 39% profit-on-cost margin. One small footnote carries a big implication: the completed value is presented as a present value discounted at 7.65% over 21 years. So even the ILS 1.079 billion headline is not rent already locked in by signed leases. It is a discounted as-complete model of an asset that still has to be built and ramped.
This is where the distinction between paper uplift and a real NOI engine becomes critical. Until the project is completed, leased, and operating, most of the model remains an appraisal and execution formula rather than a reported operating result.
Translating the model into the company share makes that even clearer. On a 50% basis, the implied completed value for the company share is about ILS 539.5 million, implied NOI is about ILS 51.5 million, and implied developer profit is about ILS 152.5 million. Those are very large numbers relative to the group. The implied company-share NOI alone equals more than half of the group’s consolidated 2025 NOI, which was ILS 92.8 million.
That is exactly why ALFA matters so much to the Levinstein read. If phase 1 reaches operation on anything close to the presentation model, it changes the group’s NOI mix. If it does not, it stays for longer as an attractive value block that remains inaccessible.
The simplest way to see the distance still left is to compare what sits on the balance sheet today with the implied completed value for the company share.
That gap is not an error. It simply means that most of ALFA’s value story still lies ahead, not behind.
Where The Funding Friction Sits
ALFA is not just a future-NOI story. It is also a heavy-investment story on the way there. The annual report speaks about phase-1 investment of about ILS 780 million including land, with the company share at about ILS 390 million. The March 2026 presentation updates the total cost to ILS 774 million. The difference is small in thesis terms. The message stays the same: this is still a capital-intensive project.
This is where an easy-to-miss point matters. In the company’s projected 2026 cash-flow statement, only ILS 19 million of equity investment is included for the data-center project. That does not necessarily contradict the much larger total company share in phase-1 investment. But it does mean the ALFA path will not be funded out of the listed parent’s cash box alone, and certainly not all at once. It depends on staged investment, project financing structure, and the pace at which the company and its partner inject equity and raise debt.
In other words, it is not enough to read the ILS 103 million NOI headline without also reading the roughly ILS 390 million company-share investment burden and the ILS 19 million 2026 projected outlay. Only those numbers together tell the right story: large potential, but along a funding path that is not fully closed yet.
There is also a wider group context. In the real-estate activity section, the company itself says that with full occupancy of Be’er Sheva, Kfar Saba, Pardesia, and the data center, group NOI could rise by about 90%. That explains why ALFA sits so close to the center of the thesis. But inside that asset bundle, ALFA is still the least proven piece at end-2025: no NOI yet, no binding leases yet, and no shift into result language before construction and operation begin to replace model language.
What Will Separate Paper Uplift From A Real NOI Engine
The distance between end-2025 and an income-producing asset is not abstract. It has a fairly clear checklist.
| Checkpoint | End-2025 status | What needs to change for the read to improve |
|---|---|---|
| Commercialization | 0% of the space under binding lease agreements | A move into binding contracts or at least a clearly committed commercial path that translates into revenue |
| Execution | Only 7% financial completion | Construction progress that actually supports the new fourth-quarter 2026 phase-1 target |
| Funding | Company share in investment of about ILS 390 million, versus only ILS 19 million in the 2026 projected cash flow | Better visibility on the debt and equity layers that will fund the remaining investment jump |
| Rights scope | Phase 1 is defined, while phase 2 is still only in zoning approvals or may be replaced by commercial and office rights | A clean distinction between what is relatively closed in phase 1 and what remains longer-dated optional upside |
The first trigger the market should look for is not another valuation headline, but evidence that the operating path is starting to close. In a project like this that means build progress against the tighter schedule, a move from zero binding leases into signed commitments, and later on reported NOI. As long as one of those steps is missing, ALFA remains mostly a potential story.
Conclusion
ALFA can become one of the assets that changes Levinstein’s scale, and certainly one of the assets that changes its NOI mix. Even on the company-share basis alone, the implied phase-1 NOI is large enough to move the group. So this is not a side asset. It is a real thesis node.
But as of the end of 2025 ALFA is still more model than engine. What sits on the balance sheet is fair value of ILS 70.6 million, after a building permit and the start of construction, but with no binding leases and only 7% financial completion. The real transition from an appraisal number to an NOI engine will happen only if phase 1 holds the tighter new schedule, closes commercially, and proves that the heavy capital spend is in fact rolling into an operating asset rather than only into another valuation step-up.
That is exactly why ALFA matters so much to the current Levinstein read: it already deserves a separate analytical lens, but it still does not justify reading 2025 as if the NOI engine is already here.
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.