BESI’s miss is the wrong fight: orders say AI packaging is turning before mainstream assembly does

The print should reset the argument from “is BESI still in a downcycle?” to “which parts of the cycle are already turning?” What was priced in was a cleaner top-line recovery: street revenue was €163.3 million, while actual revenue came in at €154.2 million for a -5.6% surprise. What actually surprised was earnings resilience, with EPS of €0.37 versus €0.31 expected, a +19.4% surprise. That combination is not a generic quality beat; it says the market was too focused on shipment timing and not focused enough on the margin and operating-expense bridge inside a quarter where orders improved ahead of revenue. The bear case can still point to mainstream assembly weakness, but the evidence from this event says the revenue miss is backward-looking while the order book is already reflecting advanced packaging capacity decisions tied to data center and photonics.

The distinction matters because the company’s own reported basis was weaker than the street-comparison revenue figure, and management did not try to make it sound like a broad recovery. Richard Blickman framed Q3 as “within prior guidance in an assembly equipment market showing early signs of recovery,” a phrase that earns attention because it confines the optimism to early signs rather than declaring a full-cycle turn. On the company’s reported basis, revenue decreased by 10.4% versus Q2 ’25 and by 15.3% versus Q3 ’24, with management calling out continued weakness in mobile and automotive applications and lower hybrid bonding revenue. That is the part investors thought they were buying into, and it was worse than the street wanted. But the offset was not financial engineering: operating income was described as at the high end of guidance, driven by higher-than-anticipated gross margins and slightly better operating expense development.

The financial trajectory makes the print look less like demand deterioration and more like a mix trough with a delayed shipment conversion. Revenue has been pinned in a relatively narrow band across the displayed history, but the latest reported quarter in the income-statement series shows the cost of that mix shift: Q3 FY2025 revenue was €132.4 million and gross margin fell to 58.0%. That is why the EPS beat against the street matters. If revenue disappointment had flowed straight through the model, EPS would not have cleared estimates by +19.4%. Instead, the company showed that even in a quarter with a weaker revenue base and adverse mix, expenses and gross margin were not as bad as feared. The market may be mispricing that asymmetry: the income statement still carries cyclical scars, but the orders are already moving in the businesses that should carry higher strategic value into 2026.

The order data are the center of the thesis, because they are specific enough to separate AI packaging from the rest of assembly. Blickman said, “Order levels improved significantly Q3 ’25 with bookings of EUR 174.7 million, increasing by 36.5% and 15.1% versus Q2 ’25 and Q3 ’24, respectively.” The wording matters because it ties the inflection to bookings, not hope, and the magnitude is large enough to change the near-term revenue setup. Management attributed that improvement principally to die attach bookings by Asian subcontractors for mostly 2.5D data center applications and renewed capacity purchases by leading photonics customers. Those are the two read-throughs that matter for the supply chain, even though the data pack names no specific customers or suppliers: Asian subcontractors are adding 2.5D data center capacity now, and photonics customers are renewing purchases now. The implication is not a vague AI halo; it is that €174.7 million of bookings came with identifiable demand sources, while mobile and automotive were still weak.

That split also explains why the Q4 guide is more important than the Q3 revenue miss. Management guided Q4 revenue to increase by approximately 15% to 25% versus Q3 and gross margin to range between 61% and 63%. A guide like that does not erase the Q3 shortfall, but it makes the miss less useful as a run-rate input. If the order recovery converts on schedule, the next debate becomes whether gross margin can climb out of the Q3 trough without waiting for a full mainstream recovery. The near-term answer is partial: management is guiding a margin rebound versus 58.0%, but not back to the prior 64% to 68% target range referenced by Adithya Metuku on the call. That is the key tension. Advanced packaging demand is improving, but foreign exchange, mix, and R&D spending are still holding back the historical margin framework.

The broader market backdrop supports the idea that this is an early-cycle turn rather than a completed one. TechInsights cut 2025 assembly market growth to 1.8% from 9%, pushing the anticipated upturn into 2026. That revision would normally be a red flag for assembly equipment, and for mainstream exposure it still is. But it also makes BESI’s order improvement more valuable, because the company posted rising bookings while the market forecast deteriorated. Blickman added that TechInsights expects cumulative growth of 42% in 2026-’29, based on AI use cases, new product introductions, and a cyclical recovery in mainstream assembly. The variant perception is that the stock should not be valued only on the lowered 2025 market number. The investable question is whether BESI’s advanced-packaging order recovery is a leading indicator into that 2026-’29 window, and Q3 gives more evidence for yes than the headline revenue miss suggests.

The hybrid bonding discussion is more complicated, and it is where the bull case should avoid overclaiming. Management said internal operations include 6 Kinex lines with 30 hybrid bonding tools, and it described the likely adoption window as 2026-’27. Blickman’s most important hedge was that first orders for initial capacities should emerge as customers complete evaluations, including designs stacking the 12 and 16. That is not a statement of immediate volume revenue. It is a statement that the evaluation funnel is still active and that the company sees its technology in front. The aggressive claim came when he said, “So far, our concept is certainly leading with a market share of over 80%, even some people say 90%.” That quote matters because it is unusually direct on competitive position, but investors should pair it with the near-term reality that lower hybrid bonding revenue was part of the Q3 decline. Leadership in evaluations is not the same as current revenue conversion.

The tone of the call reinforces that distinction: management sounded materially more constructive on guidance while becoming less uncertain, but confidence scores did not rise. The tone history shows guidance_tone moving to 0.50 in Q3 FY2025 from 0.02 in Q2 FY2025, while uncertainty fell to 51.2 from 71.6. That is a meaningful delivery change, and it aligns with the bookings inflection rather than contradicting it. The caveat is tone_confidence at 0.41, so the call became more positive without the model reading it as fully settled. That is exactly how the numbers should look at an inflection point: management has more evidence in bookings and guidance, but the revenue base and hybrid bonding conversion have not yet caught up.

The comparative context argues against treating BESI as a generic assembly equipment laggard. In the peer table, DSCSY posted gross margin of 70.8% with revenue YoY of +12.3%, while ATEYY posted revenue YoY of +43.8% with gross margin of 67.4%. BESI’s latest reported gross margin in the income-statement series at 58.0% is therefore not sector-leading on current profitability, and the Q3 revenue decline makes that obvious. But that is not the right comparative takeaway. The better point is that peers already show parts of the test and assembly complex can grow while BESI is still digesting mainstream weakness. If BESI’s Q4 revenue guide and order composition convert, the relative gap becomes a recovery opportunity rather than proof of structural impairment. If they do not, the peer data will make the miss look more idiosyncratic.

Capital allocation adds a floor to the thesis, but it should not be the reason to own the event. Liquidity was €518.6 million at September 30, and management completed a €100 million share buyback program in October ’25 before authorizing a new €60 million program with an anticipated completion date of October 2026. That tells investors the company is not conserving cash as if the downturn is worsening. It also gives management a way to return capital while orders rebuild, but the buyback is secondary to the operating question. A €60 million authorization does not fix a revenue miss, and it does not prove hybrid bonding adoption. It does, however, support the interpretation that the company sees enough balance-sheet flexibility to invest through the R&D step-up it guided for Q4.

The risk to the thesis is that the bookings inflection fails to become revenue at acceptable margin. Management guided operating expenses to increase by 5% to 10% versus Q3, primarily from higher R&D, so the margin recovery has to do real work. The company’s first 9 months of 2025 also show why investors are right to demand proof: revenue was €425 million and orders were €434.6 million, down 6.4% and 6.5% versus the comparable prior-year period. Net income was €88.8 million, down 27.6%, due to lower revenue, lower gross margins, adverse ForEx effects, and higher interest expense. Those figures conflict with the bullish bookings read because the year-to-date account still looks like a downcycle. The reconciliation is timing: Q3 orders and Q4 guidance point forward, while the year-to-date financials measure the period before the recovery broadened.

What to watch next is concrete. The thesis is confirmed if Q4 revenue lands within management’s approximately 15% to 25% sequential growth guide and gross margin returns to the 61% to 63% range, because that would show Q3 bookings are converting without the 58.0% margin trough becoming the new base. Orders need to remain anchored near or above the Q3 bookings level of €174.7 million, with management again tying demand to Asian subcontractors for 2.5D data center applications and leading photonics customers rather than relying on generic mainstream recovery language. The thesis breaks if Q4 revenue misses that guide, if gross margin stays closer to 58.0% than to 61% to 63%, or if the 2026-’27 hybrid bonding adoption language slips without initial capacity orders from the customer evaluations management described. The next quarter is not about whether Q3 revenue was disappointing; it was. It is about whether the market is still underwriting the wrong cycle at the moment orders have already started to say where the next one begins.