Semiconductor equipment change is imminent! Applied Materials and Besi are hunting down the "mixed bonding" of AI and storage technologies.

With the advanced chip packaging technology exclusively possessed by BE Semiconductor that is, Hybrid Bonding advanced packaging technology, it has become increasingly crucial for AI chips and high-performance storage systems in data centers under the global AI super wave. The two semiconductor equipment manufacturing giants in the world the headquartered in the United States Applied Materials (AMAT.US) and Lam Research Corporation (LRCX.US) have emphasized their intentions to acquire this European semiconductor equipment company. According to the latest news reported by the media citing sources familiar with the matter, BE Semiconductor, headquartered in the Netherlands, is working with the Wall Street investment bank giant Morgan Stanley to discuss the acquisition inquiries it received. One source familiar with the matter added that Lam Research Corporation has already issued one of the acquisition intentions; another source added that Applied Materials, Inc. also has strong acquisition interests. Applied Materials previously acquired about 9% of BE Semiconductor in April 2025. At that time, Applied Materials stated that it considered this stake as a "strategically significant long-term investment, indicating Applied Materials' commitment to developing together the industry's strongest hybrid bonding packaging solution a technology that is becoming increasingly important for the upgrade iterations of advanced logic and memory chip capacity and architecture that make up AI training/inference systems." On Friday, Applied Materials announced a 15% increase in its quarterly dividend to $0.53 per share, higher than the previous $0.46 per share. Earlier on Tuesday, Applied Materials, along with storage chip leader Micron Technology, Inc. (MU.US) and South Korea-based SK Hynix, announced a major partnership, to develop and build the most cutting-edge solutions and upgrade paths for DRAM, high-bandwidth memory (i.e., HBM storage systems), and data center NAND storage systems, to significantly enhance overall chip capacity and the comprehensive performance of artificial intelligence training/inference systems. Previously in February, Applied Materials' latest performance report revealed that almost all high-end semiconductor equipment provided by Applied Materials, Inc. gave a prior quarter performance that exceeded expectations, and an incredibly strong future performance guidance range, highlighting that under the rapidly growing global AI computing infrastructure wave and the "memory chip super cycle" macro background, semiconductor equipment manufacturers are also entering a super growth cycle, with them being the largest beneficiaries of the sharp expansion trend of AI chips (covering AI GPU/AI ASIC) and DRAM/NAND memory chip capacity. The largest semiconductor manufacturing equipment and advanced packaging equipment supplier in the United States anticipates its revenue for the second quarter of fiscal year 2026 to be approximately $7.65 billion, with a +/- range of about $500 million. In comparison, Wall Street analysts' average revenue expectations for Applied Materials for that quarter (ending in April of this year) are $70.3 billion knowing that with the rapid expansion of production of AI chips with advanced processes of 3nm and below, and the acceleration of CoWoS/3D advanced packaging capacity, as well as the expansion of DRAM/NAND memory chip capacity, Applied Materials' revenue expectations have been continuously raised by analysts since the beginning of this year. The Non-GAAP guideline for the second fiscal quarter earnings outlook range is expected to be between $2.44 and $2.84 per share (excluding certain items), which is significantly higher than the analysts' average expectation of $2.29 per share. Applied Materials' Non-GAAP earnings per share in the first quarter were $2.38, higher than the Wall Street average expectation of $2.21 per share, and almost the same as the same period last year; the company's gross margin in the first quarter reached 49%, up from approximately 48% the same period last year. The company's first quarter Non-GAAP free cash flow was as high as $1.04 billion, indicating a significant growth of 91%. The "new wealth code" of the chip industry Hybrid Bonding! This technology is the bottleneck for AI chips and memory chips. Advanced packaging has become a new engine driving continuous improvement in computing power in the "post-Moore era," and the performance of bonding technology directly determines the ceiling of integrated systems. The evolution of bonding technology has gone from wire bonding, flip-chip, thermocompression bonding to fan-out packaging, and ultimately to the era of hybrid bonding. Hybrid Bonding advanced packaging has become one of the core technologies for breaking through the bottlenecks of computing power and bandwidth, reshaping the value chain of AI computing. This technology is widely used in the field of AI chips and HBM such as NVIDIA Corporation's Blackwell series AI GPUs and Broadcom Inc.'s AI ASIC, SK Hynix's HBM3E/HBM4, and soon in the high-performance Ethernet switch field integrating CPO technology, it will also play a core role in advanced packaging. Hybrid bonding replaces traditional bumping with copper-copper direct bonding, achieving ultra-fine interconnects below 10m, bringing orders of magnitude improvement in interconnect density, bandwidth, energy efficiency, and unit interconnect cost. It is a key breakthrough to support 3D stacking and heterogeneous integration. Its processes are divided into wafer-to-wafer (suitable for homogeneous small chip applications such as storage) and chip-to-wafer (suitable for large chips and heterogeneous integration). BE Semiconductor's hybrid bonding advanced packaging equipment is fully adopted by leading chip manufacturers in the industry, such as Taiwan Semiconductor Manufacturing Co., Ltd. Sponsored ADR, SK Hynix, and Samsung, mainly used in the advanced packaging stage of chips, both for 2.5D CoWoS advanced packaging and even more advanced 3D/3.5D packaging require the use of hybrid bonding equipment. The bottleneck in the AI era is increasingly not in single transistors, but rather in data transfer efficiency between chips, between logic and memory, and within the packaging. In a research report, Applied Materials explicitly mentioned that hybrid bonding, through direct copper-to-copper connections, increases interconnect density, reduces line length, and simultaneously improves performance, energy consumption, and cost; hence, this direct connection can bring faster data transmission and lower power consumption. For high-performance computing chips such as GPUs, AI ASICs, and CPU+accelerators, this is crucial in determining whether packaging can continue to improve bandwidth density and energy efficiency. For HBM and high-end enterprise storage, hybrid bonding is also becoming increasingly critical, particularly for future higher-layer, higher bandwidth, and lower power memory/storage paths. Applied Materials' Kinex materials explicitly mention that their systems are oriented towards future HBM and logic-memory integration, and have already demonstrated the ability to stack 16 layers and above of DRAM; Besi also stated that their customer roadmap in the next two years will further integrate hybrid bonding/TC Next into HBM4/4e. In super large-scale AI data centers, high-performance networks themselves have become part of the computation. Besi has classified co-packaged optics (CPO) and ASIC as new use cases; Broadcom Inc. clearly stated on their 102.4Tbps Tomahawk 6 CPO switch chip that integrating optical engines with switch chips in a co-packaged manner can significantly improve energy efficiency, link stability, and support the scaling of large-scale AI clusters. The landscape of the global semiconductor equipment industry is about to change. The essence of Applied Materials and Lam Research Corporation's bids for BE Semiconductor is not an ordinary acquisition, but rather about who can acquire the control of the most scarce packaging "bottleneck equipment" in the AI era. Besi has the "world's highest precision" hybrid bonding equipment, which means it is not an ordinary packaging manufacturer, but one of the core assets of advanced packaging in the AI/HPC era. As the global construction of super large-scale AI data centers led by tech giants such as Microsoft Corporation, Alphabet Inc. Class C, and Meta becomes increasingly intense, driving all the chip manufacturing giants to accelerate the mass production of AI chips with advanced processes of 3nm and below, as well as the expansion of CoWoS/3D advanced packaging capacity and the expansion of DRAM/NAND storage chip capacity, the long-term bullish logic of the semiconductor equipment sector is becoming increasingly robust. According to the latest analyst expectations gathered by institutions, Amazon.com, Inc. together with Alphabet Inc. Class C parent company Alphabet, Facebook parent company Meta Platforms Inc., as well as Oracle Corporation and Microsoft Corporation, are expected to reach approximately $650 billion in cumulative AI-related capital expenditures in 2026, while some analysts believe that the overall spending might exceed $700 billion meaning that the year-on-year increase in AI capital expenditure may exceed 70%. If Applied Materials or Lam Research Corporation ultimately win the bid for Besi, it will signify a significant change in the entire landscape of the semiconductor equipment industry. This will also signal that advanced packaging is officially upgrading from being a "back-end process supporting role" to being a core battlefield alongside the front-end processes. For Applied Materials, this means that it can combine cleaning, activation, measurement, integration platforms with Besi's core bonding technology to create a more complete "full-stack hybrid bonding solution"; for Lam Research Corporation, it means that it can further advance towards the most critical value capture point in advanced packaging with existing TSV, RDL, high aspect ratio (HAR) etching/deposition, and hybrid bonding support tools. Regardless of the winner, the semiconductor equipment industry landscape will clearly move towards one direction: the future competition in AI chips/storage chips will not only depend on advanced semiconductor equipment updates and iterations such as EUV, etching, and thin film deposition but also on who can turn the most difficult last mile of advanced packaging into their platform's "super moat."