Ashkol Between Power and Real Estate: How Real Is the Data-Center Option

The main Dalia Energy article argued that the existing fleet is funding the expansion phase, but that Ashkol’s value is still not truly accessible. This follow-up isolates the exact angle that can distort that read: the data-center option. On one side, this is not a detached fantasy. There is a site with unusual physical advantages, an initial commercial framework that has already been signed, and explicit language around size, land and rent. On the other side, the gap between an interesting option and monetized value is still wide.

That matters because the value is already in the numbers. As of December 31, 2025, Ashkol real estate is carried at a fair value of NIS 1.736 billion on a 100% basis, split between NIS 449.5 million of leased investment property and NIS 1.286 billion of investment property under development. So this is no longer a vacant lot waiting for a dream. But it is also not yet external cash flow that justifies reading the whole value as if it has already turned into accessible cash.

The right question therefore is not whether there is a data-center option. There is. The right question is which part is already embedded in the books, which part still rests on an MOU and initial estimates, and what still has to happen before that option stops looking like an attractive real-estate layer in the accounts and starts looking like a real monetization path.

That chart puts the ground under the debate in order. Most of the value still sits in the development layer, not in contracted rent. So even if the data-center thread advances, the question is not only about one more tenant. It is about whether part of the value that is still being measured as development potential can become contractual income.

What Is Already Embedded In The Site

The first point to clarify is that the company already shows real economics at Ashkol, just not yet data-center economics. The leased part of the site is reported at 100% occupancy, with 115,600 square meters actually leased, annual income of NIS 31.361 million, and monthly NOI of NIS 2.613 million. The appraisal explicitly refers to an implied net yield of about 7% on current NOI and states that the rest of the land is not leased.

An important detail sits in the methodology. The valuer used nine comparable industrial and logistics properties around Ashdod, Kannot, Be'er Tuvia, Gedera and Ramla, with rent of about NIS 270 per square meter per year. That is the heart of the story. The current value at Ashkol already captures real estate with development options, but it is still not based on a signed data-center transaction or on the economics of an operating AI campus.

The investor presentation adds the current lease layer. Ashkol Yitzur currently pays NIS 30 million per year, and after the steam units close that amount is expected to fall to NIS 15 million. Ashkol Avshal is expected to pay NIS 20 million per year after the option is exercised, from the start of construction. Those are important numbers, but they need to be read correctly: the annual report explicitly classifies these leases as part of intercompany arrangements among the Ashkol entities. In other words, they support site economics at the Ashkol Energies level, but they are not the same thing as a new outside tenant bringing in external cash.

That is exactly the difference between site value and accessible value. At the Ashkol Energies layer, internal rent can support valuation, LTV and the financing structure. At the listed-company layer, it is not equivalent to a new outside agreement that brings in a new partner, a new tenant, and income that does not depend on Ashkol’s own internal structure.

Why The Data-Center Option Looks Real

This is exactly where the read can go wrong in two opposite directions. It is too easy to dismiss the MOU because it is not binding, and just as easy to get carried away by it as if it were already a signed deal. Neither extreme is right. In infrastructure terms, this is already a relatively advanced early-stage framework: exclusive negotiations for up to six months, a minimum of 130 IT MW, about 50 dunams of land, a 24-year and 11-month lease term, initial annual rent estimated at 2026 prices in the NIS 30 million to NIS 50 million range, electricity to be supplied by one of the Dalia group’s licensed suppliers at a discount to tariff, and an option for Ashkol Energies to hold up to 30% of the joint venture.

The interesting point is not only the rent range, but rent density. The currently leased area at the site reflects average annual rent of NIS 271.25 per square meter. By contrast, the data-center MOU refers to about 50 dunams, or roughly 50,000 square meters. Dividing the initial rent range by that footprint yields roughly NIS 600 to NIS 1,000 per square meter per year. That is a very different land-use economics.

That chart does not prove that fair value should be marked up tomorrow morning. It does explain why the data-center path is a real option rather than just a glossy slide. If the deal actually matures, it could change the quality of rent on part of the site, not merely add one more user.

That is reinforced by the physical advantages of the location, and here too the company is no longer speaking in generic terms. The presentation points to seawater cooling, energy redundancy thanks to multiple generating units at the same complex, and potential connection to an international submarine communications cable landing at Ashkol. Beyond that, the company says it has already received a Ministry of Communications license for a submarine optical cable landing station at the site. So this is not only open land. It is a site trying to position itself at the intersection of power, cooling and connectivity.

There is also real movement in the planning layer. The company writes that during the coming year it intends to advance building permits on existing rights of about 100,000 square meters for data-center and AI-center uses, while also promoting a new zoning plan aimed at expanding and maximizing the site’s building rights. That is still not a permit in hand, but it is no longer the language of merely “examining options.”

What Still Stands Between The Option And Accessible Cash

The problem is that the land at Ashkol does not arrive clean, cheap and unencumbered. The company itself estimates roughly NIS 80 million of soil treatment and rehabilitation for site development, plus about NIS 65 million of demolition and evacuation costs, including removal of long-idled turbines. In other words, before the discussion even reaches whether the MOU turns into a binding agreement, there is already an initial site-prep threshold of about NIS 145 million.

And that is before the more detailed environmental layer. The soil and water surveys performed before the site was delivered identified areas with potential soil contamination and groundwater contamination. The company says it is operating under the Ministry of Environmental Protection’s instructions, and that in the section intended for the new generating unit, the remaining soil-rehabilitation work is planned to be completed by the end of the second quarter of 2026. In addition, the groundwater survey identified PFAS material, and the company notes that it has proposed a solution to the Water Authority, with full cost disclosure still dependent on regulatory approval. That matters. The land may have strong potential, but it is not a ready pad where the value can simply be switched on.

The second constraint is financing, and it is not minor. Ashkol Energies’ recycled real-estate loan stands at NIS 780 million, plus a NIS 50 million facility. The presentation refers to outstanding debt of NIS 770 million as of December 31, 2025, floating-rate, with final maturity in December 2030. The annual report makes clear that the land rights and related rights are pledged under a first-ranking fixed charge, including a mortgage, and that distributions are prohibited until the loans are repaid.

As of year-end 2025, Ashkol Energies’ LTV stands at about 47%, versus an 80% breach threshold, so there is no immediate covenant squeeze here. But that is exactly the point: even if the asset looks reasonably safe from a financing perspective, value created at the site is not automatically available to the listed company. It first serves the asset layer, the debt that sits on it, and the restrictions that come with it.

The third constraint sits at the public-company layer. The company states explicitly that it depends on cash flows from its subsidiaries, and that its ability to receive those cash flows may be limited, among other things, by distribution restrictions under financing agreements. So even if the data-center thread becomes a signed external deal, the next question is still who benefits from that cash flow first, when, and through which route. That is the answer to the gap between optional value and monetized value.

So How Real Is The Option

The sharp conclusion is that the option is real, but still not monetized. It is real because there is an unusual site, clear physical and operational advantages, a supportive communications license, an initial commercial framework with identified partners, and a rent range that looks materially above the average rent per square meter on the currently leased area. It is still not monetized because, as of now, it remains a non-binding MOU on land that still needs preparation, inside a financing structure that restricts distributions, and inside a listed issuer that in any case depends on subsidiaries’ ability to move cash upward.

Put differently, Ashkol is no longer just a power station, and it is no longer just land either. But anyone jumping from here straight to “the data center justifies the whole site value” is skipping three critical steps: a binding agreement, actual site-prep spending, and the ability to convert asset-level value at Ashkol Energies into accessible value at Dalia Energy.

The good news is that the market does not need to wait for the final stage to get an answer. Over the next two to four quarters, three proof points are enough to change the read materially: a binding agreement instead of an MOU, planning and permitting progress on existing rights, and evidence that the new economics remains attractive even after site-prep cost and financing restrictions. Until then, the data-center thread is an option that should not be dismissed, but it is too early to count it as cash.