Gilat Telecom: Does The Bezeq IRU Really Create A Moat, or Mostly A Capacity Commitment

Why Isolate The IRU

The main article already argued that Gilat’s turnaround looks real, but it also flagged one proof point that is still not fully settled: can the Bezeq cost advantage turn into durable profit, or does it mainly give the company a strong strategic narrative. That is exactly why the IRU deserves to be pulled out and read on its own.

The key point is that the IRU is no longer a theoretical story. The original December 2024 agreement covered 18 thousand lines over 15 years in three tranches. By the time the annual report was published, the company was already talking about roughly 22 thousand active lines under the Bezeq agreement, alongside 17,929 private internet subscribers, more than 300 business customers, and more than 1,000 business fiber lines. In other words, actual scale had already moved beyond the original framework, so the November 2025 expansion was not built on a distant dream. It was built on utilization that had already started to outrun the first contract.

That is also where the analytical problem begins. A moat exists only if purchased capacity is filled at a pace that justifies the fixed cost layer. If that pace weakens, the same mechanism that looks like a cost edge starts to look like a financed capacity commitment. So the right question is not whether Bezeq gave Gilat a good price. The right question is whether the structure creates an advantage that survives maintenance, cost spreading, and financing.

What The Company Actually Bought

The original December 2024 agreement gave the subsidiary an irrevocable right of use for 15 years over 18 thousand infrastructure lines, with 8,000 lines starting in early 2025, another 5,000 lines spread over four equal portions from April 2025 to January 2026, and a further 5,000 lines spread over four equal portions from April 2026 to January 2027. Total consideration for that first agreement period was NIS 88.02 million, or about NIS 4,890 per line.

That is not the whole picture. The original agreement also included speed-upgrade rights, a price-protection mechanism from 2028 linked to regulated line pricing, an option for another 10,000 lines on the same terms, and extension options for two additional five-year periods. At the same time, the company made clear that these were rights over non-specific lines and therefore not leases. From an accounting perspective, prepaid amounts and the interest component are recorded as deferred expenses and amortized over the agreement period including the two option periods, meaning 25 years.

That detail matters. The cost of the IRU does not confront the reader as one closed CAPEX item that simply vanished into cash. It is spread forward and enters cost of sales over time. That is exactly why it is easy to read the margin improvement without fully seeing how much future capacity has already been purchased.

In November 2025 the company signed a major supplement to the agreement. This is where the scale jumps. Instead of 18 thousand lines, the framework expands to up to 90 thousand lines in four tranches:

The first takeaway is positive. Unit headline cost actually improved. The first tranche of the November expansion implies about NIS 4,444 per line, around 9% below the original NIS 4,890 per line. The next three tranches imply the same rough level. So Gilat did receive a lower unit cost. That is the heart of the moat argument.

The second takeaway is just as important. This moat is not free. Under both structures, the company pays annual operating and maintenance fees equal to 4% of consideration, in quarterly payments, over all lines for which use rights have already been granted. In addition, each tranche can be spread over ten annual payments with interest tied to Bezeq’s shekel bond yield at the relevant duration. Put differently, Gilat can soften the cash hit. It cannot eliminate the price of the commitment.

When It Starts To Look Like A Real Moat

To understand whether the discount really creates an advantage, three numbers need to be connected even though the report presents them separately: cost per line, maintenance burden, and average revenue per customer.

In private internet, the company reports monthly ARPU of NIS 86.85 for 2025. As a rough heuristic only, that translates into about NIS 1,042 of annual revenue per line. Against that, under the original agreement, annual maintenance alone equals about NIS 196 per line. If the upfront NIS 4,890 line cost is spread across the same 25-year period the company uses for accounting amortization, that adds another roughly NIS 196 per line per year. So the two base items together already amount to about NIS 392 per line each year.

That means that in the simplest heuristic, before customer support, equipment, marketing, credit cost, technical service, and financing, roughly 38% of annual revenue at the disclosed private ARPU is already absorbed by maintenance and accounting amortization of the use right. Under the November expansion, that burden falls to about NIS 356 per line per year, or roughly 34% of the same heuristic. That is a real improvement, but not a world in which line cost becomes trivial.

That is why the IRU only works above a certain utilization threshold. If the company keeps adding customers, preserves reasonable ARPU, and continues to expand the business side of the activity as well, the unit discount can genuinely translate into a cumulative cost advantage. If customer growth slows, or if pricing has to weaken too quickly to fill capacity, the same fixed structure starts eating into what first looked like margin.

There is an important positive point here as well. By the time the report was published, the company said the active line count already stood at about 22 thousand, above the original 18-thousand-line framework. That is the first proof point. Gilat is no longer sitting on IRU rights that exist only on paper. It is already deep into real utilization. So the claim that the Bezeq contract is merely a strategic story no longer holds.

Where It Still Looks More Like A Capacity Commitment

The next step is to see that the “up to 90 thousand lines” headline sounds both heavier and lighter than the underlying economics.

On the lighter side, the report gives the company a release valve. If during the realization period of the prior tranche plus one additional year the relevant growth target is not achieved, Gilat is released from the obligation to purchase the next tranche. More importantly, that test also includes lines purchased from another infrastructure provider. So the 90-thousand-line headline is not one blind and unconditional check.

On the heavier side, that does not make the structure low-risk. The first tranche of the November expansion stands at 36 thousand lines. Against roughly 22 thousand active lines today, the company still needs to add about 14 thousand lines just to fill the first expansion threshold. That is about 64% above the current level. Against the full 90-thousand-line framework, the gap is roughly 68 thousand additional lines.

That is the core risk. The moat already exists at the unit level, but the commitment is still large at the volume level. Even the ability to spread payments over ten years is not a cancellation of that risk. It is a conversion of that risk. Near-term cash pressure becomes lighter, but financing cost tied to Bezeq’s bond yield takes its place. From an equity-holder perspective, this means the cost edge does not depend only on lower line pricing. It also depends on Gilat’s ability to fund the expansion without letting financing absorb the savings.

The report also sharpens another point that is easy to miss. During 2025 the balance sheet already recorded deferred expenses for fiber infrastructure use rights at a cost base of $3.902 million, with a net balance of $2.823 million at year-end, including $1.112 million short term and $1.711 million long term. That is proof that the agreement has already started moving from strategic narrative into a real cost layer in the accounts.

What Has To Be Read Differently In The Next Reports

Anyone who reads Gilat through the simple “90 thousand lines” headline may conclude too quickly that a deep retail moat has already been built. That is too early. The better reading is that the company bought itself a better cost curve, but also signed up to a stricter utilization discipline.

In the next reports, three things will decide whether the IRU starts looking like a real structural advantage:

• growth in active lines, not just in signed frameworks or nominal capacity,
• the ability to keep ARPU and revenue quality stable while filling capacity,
• and clearer evidence that the cost advantage is flowing into cost of sales and direct profit in the internet activity, not just into management’s strategic language.

If those three signals appear together, the Bezeq expansion will look in hindsight like a move that locked in a real cost advantage in a very competitive market. If fill rates slow, or if the market forces Gilat to concede pricing too quickly to fill lines, the same IRU will begin to look like capacity bought too early.

Conclusion

The short conclusion is that the Bezeq IRU can become a moat, but not because it looks cheap on paper. It becomes a moat only if Gilat keeps filling lines at a pace that preserves the unit advantage after maintenance, amortization, and financing.

The report already gives one meaningful proof point in support of the thesis: the company reached roughly 22 thousand active lines under the Bezeq agreement, so the original 18-thousand-line framework is no longer the real story. At the same time, the November expansion does improve unit pricing by around 9%. But the report also gives the warning: the new first tranche still requires roughly 14 thousand more lines beyond current scale, while maintenance remains a fixed burden tied to consideration.

So the more precise reading is neither “moat” nor “risk.” It is a leveraged operating bet on utilization pace. If the pace holds, Gilat benefits from a cheaper line base and from a price-protection mechanism linked to regulated line pricing. If the pace slows, the same asset will mostly reveal itself as capacity bought too early.