Unitronics: Are Data Centers Really Offsetting US Weakness, or Only Changing the Mix

What This Follow-Up Is Actually Testing

The main article already marked data centers as a new option for Unitronics. This follow-up isolates a narrower, more commercial question: are data centers truly offsetting weaker US demand, or are they mostly changing product mix and channel mix for now.

The good news is real. Until 2024, sales into data centers were negligible. In 2025, that activity moved above 10% of total company sales. At the same time, direct sales kept expanding, reaching 26% of total revenue versus just 13% in 2022. In the US market itself, direct sales rose from 16% of sales in 2022 to 37% in 2025.

But this still does not read like full compensation for broader weakness. Annual revenue fell to NIS 154.8 million from NIS 192.2 million in 2024. The board ties the decline mainly to a US slowdown driven by market uncertainty around the US tariff program starting in the second quarter, and says data-center growth only partly offset that pressure. That is the core debate: there is a real new engine, but in the 2025 report it is not yet strong enough to erase the broader friction in the US market.

So the right read is not "Unitronics found a new growth driver, therefore the US issue is solved," but it is also not "nothing changed." What changed is the quality of the mix and the proximity to the customer. What is still unproven is whether that shift is now large enough to stabilize revenue and margins under tariffs, customer hesitation, and a shorter order cycle.

Data Centers Are Real, but They Are Not Yet a Full Replacement

The most important datapoint in the filing is that data centers are no longer an experiment. The company invested in targeted marketing and in dedicated controller capabilities for cooling and energy-management systems used in data centers, and during 2025 it expanded support for features such as RedFish, IPv6, and additional functionality tailored to the stricter requirements of Hyperscale customers. This is not cosmetic. It means Unitronics adapted its product to a very specific demand pocket that is growing globally.

The problem is that the same report frames this activity as only a partial offset, not a full substitute. If data centers were already covering the broader weakness in the US market, the top line should show it. Instead, revenue fell by about NIS 37.5 million and gross profit dropped by about NIS 21.5 million. In other words, the new activity is already meaningful enough to change the mix, but not yet broad enough to neutralize the pressure on the company’s older core demand buckets.

That distinction matters because it changes how the story should be read. For now, data centers are a commercial wedge, not a full-system replacement. They prove that Unitronics can penetrate a niche with strong demand. They do not yet prove that the broader US customer base is buying again at a healthy pace.

The rise in the direct channel strengthens that reading. When the company moves from distributor-led sales toward direct sales, it is not just selling more on its own. It is changing the point of contact with the customer, improving its ability to tailor the solution, and deepening its grip on strategic accounts. That fits especially well with data-center customers, where technical fit and tighter commercial interaction matter more.

But it also means that part of what looks like "offset" is really a change in channel quality, not necessarily a full recovery in end demand. If more sales shift into the direct model, Unitronics gains more control, but it also becomes more directly exposed to customer hesitation, pricing friction, and tariffs that affect the purchase decision itself.

The Direct-Sales Push Is Working, but It Also Exposes the Weakness More Clearly

The company works through roughly 180 distributors, including about 105 in the US and North America, while also using an internal sales structure and independent sales representatives in the US and Italy. That setup lets it combine broad reach with deeper direct penetration into selected accounts. By 2025, the direction is clear: direct sales are growing faster than distributor sales, and in the US they already account for 37% of sales.

At first glance, that looks entirely positive, and to a large degree it is. Direct customer access is an asset. It allows Unitronics to offer a fuller solution rather than just sell a controller off the shelf. It also fits strategic customers that require more customization, support, and tighter implementation, exactly the kind of demand seen in cooling and energy-management systems for data centers.

But this is also where the yellow flag sits. When the shift toward direct sales happens in a market where customers are hesitating because tariffs and import conditions keep changing, the company is closer to the customer but less insulated by the distributor layer. Put differently, commercial strength is improving, but direct exposure to weak demand is rising with it.

The annual report says this explicitly. Revenue in 2025 fell mainly because of a US slowdown caused by market uncertainty around the American tariff program starting in the second quarter. That same US market is also the one where Unitronics is pushing its direct-sales strategy most aggressively. So this is not two separate stories. It is one story: Unitronics is deepening its US exposure at the exact moment the US market has become harder to read and harder to serve.

Tariffs Are Eroding the Quality of the Offset

The tariff note makes the point sharper. In April 2025, the US imposed a 10% tariff on imports from Israel. Beginning on August 7, 2025, that tariff rose to 15%. On February 20, 2026, the US Supreme Court ruled that the president had exceeded his authority in implementing the broad tariff program, but the administration then announced a global 10% tariff that would be added to the tariffs that remained in force. The company says it still cannot estimate the full impact of the new policy at this stage.

That means the question is not only whether there is demand in data centers, but at what price and with what commercial friction that demand arrives. The company says it updated selling prices in the US to reduce the direct profitability impact. That is a rational move, but it does not solve the hesitation problem. A price increase can protect margin percentage, yet it can also delay a purchase decision for a customer that is already cautious because of shifting trade rules.

The operating discussion shows that this friction is already costing money. Selling and marketing expense fell by only about NIS 0.9 million in 2025, and that reduction was partly offset by tariff costs on exports to the US starting in the second quarter. So even on the cost side, the report shows that data-center growth and direct-sales expansion are not a free gift. The company is working against a market that offers stronger demand in one niche, but also more commercial cost and more uncertainty.

That also helps explain why gross margin did not hold up. Gross margin fell to 49.6% from 51.1% in 2024. The direct explanation is that fixed-cost items became heavier as a percentage of a lower sales base, partly offset by lower raw-material prices. But for the thesis here, the implication is simple: even with data-center contribution, the company did not yet create enough volume or enough pricing power to neutralize the broader pressure.

Backlog No Longer Carries the Visibility It Did During the Shortage Years

One risk in reading 2025 is to treat backlog as if it were still a primary visibility indicator. The company itself says that is no longer the case. During 2022 and 2023, when global component shortages stretched lead times, customers placed orders several months ahead. Starting in 2024, Unitronics returned to shorter delivery times, and customers largely returned to ordering for immediate delivery.

That means the year-end 2025 backlog of NIS 17.956 million is not directly comparable to the year-end 2024 backlog of NIS 23.065 million in the way investors were used to reading backlog during the abnormal shortage period. As of February 24, 2026, backlog had already risen to NIS 20.249 million, and the company adds that differences between periods mainly reflect FX changes. So anyone looking for backlog to prove that data centers have fully replaced weak US demand is likely reading the number too literally.

The important point is not that backlog is "weak." It is that it is less informative than it was before. In a shortage period, long backlog meant visibility. In a fast-delivery market, a larger part of demand shifts back to orders placed closer to shipment. So if the goal is to test whether data centers are genuinely offsetting US weakness, the focus should be more on reported revenue, margins, and the continued rise in direct sales, and less on backlog as if it were still the main dashboard.

The Same Growth Driver Is Also Creating Procurement Pressure

There is one more detail in the filing that complicates the story. Late in 2025, global lead times for memory components began to lengthen, mainly because of a sharp rise in demand tied to data centers, and in some cases prices also increased. To protect supply continuity, the company began increasing safety stock for certain components.

This looks like a small footnote, but it is critical for assessing demand quality. The same trend that supports sales can also make the business more inventory-heavy and more cost-sensitive. If data centers are pushing the broader memory-component market back into tighter conditions, Unitronics can enjoy stronger end demand while also needing to carry more inventory and absorb more cost volatility. In that sense, data centers are not only a growth engine. They are also a growth engine that demands more discipline in procurement, pricing, and inventory management.

That is another reason not to stop at the headline that the activity is now above 10% of sales. For this story to become a true offset to weak US demand, the company will need to show that the new niche can keep growing without overloading safety stock, procurement costs, and margins.

Bottom Line

Data centers are a real offset, but in 2025 they were still not a full replacement. They improved the business mix, strengthened the direct-sales channel, and proved that Unitronics can penetrate a fast-growing niche. At the same time, the report clearly shows that weakness in the broader US market remained large enough to pressure revenue, gross profit, and the cost structure.

So the right read of 2025 is that data centers changed the quality of the mix before they changed the level of total demand. If over the next 2 to 4 quarters the company shows stable US revenue, continued growth in direct sales, and better margin resilience despite tariffs and component pressure, then the new engine will be doing more than just reshaping the mix. In this report, it is still mostly offsetting part of the weakness, not erasing it.