Paradis
AI & Technology Investment Research
Kulicke and Soffa is mispriced because JEDEC's relaxation to 775μm extends thermocompression bonding into HBM4 and HBM5, which is KLIC's field. Its fluxless TCB tools, which with ASMPT's aim to delay hybrid bonding adoption, are a structural tailwind the market reads as a delay risk. The stock has been a chronic disappointer for about three years on wirebonder cyclicality, and the dominant TCB narrative belongs to BESI and ASMPT. x.com
Applied Materials' larger stake in BE Semiconductor points to long-term hybrid-bonding potential despite delayed adoption in HBM. Applied agreed to raise its BESI holding from 9% to about 20%; HBM adoption timelines may not reflect the prospects for the technology as a whole. x.com x.com
Logic folding could create more hybrid-bonding demand for BE Semiconductor than shorter HBM stacks take away. Customers are requesting four- or six-die stacks because HBM is expensive; fewer layers need fewer bonding steps and fit packages more easily with conventional solder, weakening the near-term HBM opportunity. Vertically stacked logic still needs exceptional alignment as it shortens signal paths and cuts energy use. x.com
Hanmi Semiconductor combines strong structural positioning with positive sensitivity to hybrid-bonding delays, backed by real TC bonder orders, and it just won a packaging equipment order for Elon Musk's Terafab. SK hynix said at Hot Chips on August 23 that it stays on the existing bonding method through HBM4E, with hybrid bonding starting at 20-high, opening a window of close to three years for delay beneficiaries. x.com
Tight memory supply keeps the AI memory supplier trade in bill-of-materials names if it persists. The investor aleabitoreddit reports an SK hynix shortage outlook through 2030 and an operating-margin result near 80%. HBM allocation and average-selling-price guidance are the next tests; Korea has already priced the earnings print, so the next move depends on who locks capacity. x.com
Higher HBM stack requirements favor Samsung Electronics shares, even as the broader semiconductor industry suffers and DRAM is in dire straits. Daishin Securities says plans for 4-high and 6-high HBM in low-end inference appear shelved, while Nvidia's 8-high Rubin option appears dropped amid base-die supply constraints and customers' preference for bandwidth. x.com
ASMPT should outperform BE Semiconductor over the next two years because hybrid bonding for HBM has slipped toward 2028, keeping older bonding in use longer. At Hot Chips 2026, SK hynix signaled it will extend MR-MUF through 16-high HBM4; per Yole, wafer-to-wafer hybrid bonding arrives around 2028. ASMPT specializes in the conventional bonding that is sticking around; BESI loses its main near-term driver. BESI is no slam-dunk short, since Intel's Diamond Rapids uses hybrid bonding in chiplets. … Show more
BE Semiconductor Industries is a good long on its supply-chain importance, as recent Lam Research and Applied Materials news validates how important hybrid bonding is to advanced packaging. Upside looks more limited, though; it is unlikely to rise 300% like names such as Applied Optoelectronics. x.com
The semiconductor cycle looks stronger than ever because repeated delays in next-generation technology extend the life of existing solutions. When engineering hits a breaking change, customers lean on older approaches, keeping bonding techniques such as MR-MUF relevant for much longer, which is better for the industry. @jukan05 notes hybrid bonding may not be used even in HBM5. Likewise, Nvidia's Rubin NVL72 Oberon racks become more relevant on the path to Kyber. x.com
SK hynix's earnings miss is misleading; the quarter confirms its pricing power in a market where suppliers and customers alike say demand cannot be met. Revenue was ₩79 trillion and operating profit ₩61 trillion at a 76.3% margin. HBM prices were locked in late 2025 and now sit below commodity server DRAM per gigabyte, while DRAM contract prices rose 90-95% in Q1 and 58-63% in Q2. Management said 2027 HBM talks are going smoothly and reflect the opportunity cost of forgone DRAM. x.com x.com
SK hynix weighing Intel Foundry for HBM4E base dies would be the validation Intel Foundry has been waiting for: a customer judging it on cost and capability for the most in-demand product in semiconductors, not a subsidy or national-security case. Per @jukan05, SK hynix wants to diversify beyond TSMC, whose base die reportedly costs 3-4x the core memory die, a cost it cannot pass through under long-term HBM deals. Its packaging VP is also evaluating Intel's EMIB against TSMC's CoWoS. x.com
