ASMPT's Record Order Book Comes With a Handoff Problem

The AI packaging boom is real, but the memory leg has been pushed out—and the new CEO is asking investors to underwrite logic and photonics first.

0522.HK · Earnings Call · 2026-10-27

ASMPT's first earnings call under new Group CEO Bassel Haddad was, on the surface, a triumph: record group bookings, record TCB and SMT orders, record photonics revenue, and a book-to-bill of 1.69. But the detail underneath the headline is a company in the middle of a handoff. The AI infrastructure order wave is real; the memory leg that was supposed to anchor the next phase has been pushed to the right, and management is now asking investors to underwrite logic, panel-level packaging and photonics before HBM reaccelerates.

A record quarter with a visible air pocket

Q3 revenue of $669.2 million rose 6.3% sequentially and 48.7% year-on-year, above consensus. “Group quarterly bookings reached a record high at USD 1.13 billion” — Yifan Xu, Chief Financial Officer · 2026-10-27, up 24.7% QoQ and 137.9% YoY, while “Book-to-bill was 1.69, the highest since Q1 2021.” — Yifan Xu, Chief Financial Officer · 2026-10-27 Adjusted gross margin of 42.9% expanded 35 basis points QoQ and 517 basis points YoY, and adjusted EPS of HKD 1.65 beat consensus. The record bookings were broad: TCB set a record, C2S orders hit an all-time high on AI data-center demand, SMT booked a record for the third straight quarter, and photonics set records for both bookings and revenue. Then came the guide. Q4 revenue is expected at $670–730 million, a midpoint of $700 million, up 4.6% QoQ and 37.5% YoY but only modestly above the just-reported quarter. More importantly, “SEMI revenue is expected to be flat due to longer lead times of AP products.” — Bassel Haddad, Group CEO · 2026-10-27 SMT is expected to carry the quarter with double-digit growth. And management explicitly warned that “Q4 booking is expected to be below Q3 record level due to the one-offs and pull-in in SMT.” — Bassel Haddad, Group CEO · 2026-10-27 In other words, the record order quarter included a meaningful amount of bulk order activity and pulled-forward demand; the backlog is real, but the order rate is not necessarily the new run-rate.

The HBM pause—and the moving TAM goalpost

The most consequential commentary was on HBM. Management described a stagnation in HBM layer counts and a recalibration of customer investment schedules.

... we believe what we see in HBM layer count stagnation reflects the industry reassessment of the optimal balance between memory density, bandwidth and thermal. Even though we see larger investments have been pushed out, we believe there will be the need to add capacity and selective upgrades. So all in all, it is an adjustment to the timing rather than a change in the underlying demand.
Bassel Haddad, Group CEO · 2026-10-27

That is a reasonable framing, but it is also a change from the prior script. In March, management said “2026 will be a year whereby, there will be new true demand for TCB for 16-high” — Cher Ng, Group Chief Executive Officer (CEO) · 2026-03-04, though it could not give timing. In April, the message was that the “TAM forecast is actually of the $1.6 billion is intact.” — Yifan Xu, Chief Financial Officer · 2026-04-22 Now the mix has moved. Haddad said “the majority of the TCB TAM in the next couple of years would come from logic versus memory -- versus HBM.” — Bassel Haddad, Group CEO · 2026-10-27 That is a meaningful shift from the prior framing, where HBM was described as the larger portion of the TAM until the outer years. In March, CFO Katie Xu said “the memory -- the HBM portion in the TAM definitely is much larger than logic. But as we go out -- a few years out, this dynamic will actually shift” — Yifan Xu, Finance or Financial Officer (likely CFO or similar) · 2026-03-04. The shift appears to be arriving sooner—or at least the company is asking investors to value it sooner. TCB TAM is still expected to expand beyond $1.6 billion by 2028, with the official update due at the Q1 2027 call. The new pillars are logic C2W, silicon bridge integration, and panel level packaging. Panel is still early: sample bonding with a leading advanced logic customer, with “orders to come potentially late '27 with some revenue realization starting with 2028.” — Bassel Haddad, Group CEO · 2026-10-27 Bridge die work with a leading global IDM is in development. The HBM use TCB opportunity has not disappeared—it has been deferred, and the burden of proof has shifted to logic.

Photonics is the quiet second leg

The call spent less time on photonics than on TCB, but the numbers were arguably more surprising. The Photonics business delivered record bookings and revenue, driven by production ramps of 800G, 1.6T and next-generation optical modules. AMICRA scored a breakthrough NPO bulk order from a leading communication service provider, and management expects that engagement to continue through 2027 rather than being a one-off. On the technology roadmap, “we expect the NPO to ramp earlier than CPO.” — Bassel Haddad, Group CEO · 2026-10-27 CPO volume production is framed as more of an end-of-decade event. The EIC-to-PIC bond is a TCB opportunity, which ties photonics back to the core advanced-packaging franchise. Hybrid bonding is the other optionality. Management said its second-generation tool is engaging with a Tier 1 foundry and a leading IDM as a second source, with qualification starting and volume demand potentially 12–18 months out. “We believe every big customer in such a new technology would look for more than one supplier” — Bassel Haddad, Group CEO · 2026-10-27, Haddad said. That is a credible way to enter a market dominated by an incumbent, but it is still a qualification story, not a revenue line.

The SMT question and the new CEO's discipline

SMT is the awkward piece. It just posted record bookings for the third consecutive quarter and is expected to drive Q4 growth, yet the strategic review remains unresolved. Asked directly, Haddad would only say he cannot disclose details, adding that the record bookings reinforce the quality and strategic value of the SMT business and that the assessment is focused on maximizing long-term shareholder value. Prior management had said it had received interest in SMT. The lack of an update is itself a signal: the new CEO is not rushing a sale into strength, but he is also not closing the door. Operationally, the company is spending to support the upcycle. Adjusted operating expenses rose 6.0% QoQ and 16.6% YoY on higher incentive provisions and volume. SEMI gross margin is in the mid-40s and management believes it can expand beyond that with a richer advanced-packaging mix. SMT gross margin should hold at current levels if volume stays elevated. The balance sheet and capacity position appear comfortable—management said it has enough capacity to meet demand and is using safety stocks and risk buys to manage component lead times.

What to watch

The bull case is that Q3 was a bridge quarter: logic TCB, panel, bridge die and NPO fill the gap while HBM customers digest their roadmaps, and the TCB TAM expands beyond $1.6 billion into 2028. The bear case is that Q3 marked the peak of the order cycle, with pulled-forward SMT demand, a flat SEMI revenue guide, and a memory customer base that has repeatedly pushed out the next capability node. The customer investment recalibration is the swing factor. Until HBM orders resume, ASMPT's record backlog is a show-me story—and the market will want to see Q4 bookings normalize at a level that still supports the 2027 growth narrative.