The Next AI Chip Shortage May Be the Machines That Make HBM Dr. Robert Castellano Tue, November 10, 2026 at 10:26 AM EST 16 min read AMAT +1.51% 240810.KQ +1.63% LRCX +1.49% NVDA -4.25% MU -1.62% Explore stocks on Coinbase Trading disclosure Trading disclosure The above button links to Coinbase. Yahoo Finance is not a broker-dealer or investment adviser and does not offer securities or cryptocurrencies for sale or facilitate trading. Coinbase pays us for certain activity generated through this link. Prices displayed are informational.
HBM capacity expansion is creating a second AI investment opportunity in semiconductor manufacturing equipment, extending from wafer fabrication to advanced packaging.
TOWA's quarterly equipment orders increased 125.8%, while South Korea's Wonik IPS is projected to nearly double semiconductor equipment revenue between 2025 and 2027.
Lam Research, KLA, Applied Materials, and Camtek offer different investment opportunities, with Applied's planned 20% stake in BESI adding exposure to next-generation hybrid bonding.
The semiconductor industry's artificial intelligence investment cycle has largely been evaluated through the revenue growth of Nvidia, Micron, SK hynix, and other companies supplying processors and memory for AI data centers. However, the expansion of high-bandwidth memory (HBM) production is creating a second investment opportunity involving the semiconductor equipment manufacturers responsible for producing increasingly complex memory devices.
HBM differs from conventional DRAM because its manufacturing requirements extend well beyond fabricating memory cells on silicon wafers. Multiple DRAM dies must be thinned, interconnected through vertical structures known as through-silicon vias (TSVs), stacked, bonded, and inspected before the completed memory package can be integrated with an AI accelerator. Each additional manufacturing operation introduces equipment requirements, processing costs, and potential yield losses.
The investment significance is that HBM manufacturing creates two sources of equipment demand. The first is conventional capacity expansion, as memory manufacturers increase wafer processing and advanced packaging capacity to meet growing AI demand. The second is increasing manufacturing complexity, as successive generations of HBM require more sophisticated deposition, etching, bonding, and process-control technologies.
These developments are occurring simultaneously. SK hynix, Samsung, and Micron are investing in HBM capacity, while equipment manufacturers are reporting stronger orders and revenue associated with memory manufacturing and advanced packaging. The opportunity extends across several equipment categories, although the financial benefits vary considerably among suppliers.
If you've saved over $1,000,000, this guide is for you. The last thing you want in retirement is to run out of money, you want your money to generate lasting income while you enjoy your life.
Now you can learn the strategies wealthy retirees use to fund their retirement with The Definitive Guide to Retirement Income from Fisher Investments. Download the guide today! (sponsor)
The central question for investors is whether this represents another cyclical increase in semiconductor capital expenditures or a more sustained expansion of equipment spending driven by increasingly demanding memory manufacturing processes. Recent financial results from Japanese and South Korean equipment manufacturers provide evidence that the investment cycle is extending beyond the largest U.S. wafer fabrication equipment companies.
The primary U.S.-listed investment opportunities are concentrated among Lam Research, KLA, Applied Materials, and Camtek. These companies participate in different stages of HBM manufacturing, creating different relationships between equipment spending, revenue growth, and profitability.
Lam Research (LRCX) is positioned primarily through its etching, deposition, and wafer cleaning technologies. HBM manufacturing requires TSV structures that connect stacked DRAM dies, and the fabrication of these structures involves deep silicon etching and thin-film deposition. Lam also benefits from conventional DRAM manufacturing investment, particularly as memory producers transition to more advanced process technologies.
Lam reported September-quarter 2026 revenue of $8.22 billion, increasing 54.4% year over year and 22% sequentially. Its GAAP operating margin reached 40.0%, compared with 37.4% in the preceding quarter and 34.4% one year earlier. This demonstrates the operating leverage available when semiconductor equipment sales accelerate. The company attributed its sequential revenue growth to strong demand from both foundry and memory customers, rather than HBM alone.
KLA (KLAC) provides a different investment opportunity through semiconductor inspection, metrology, and process control. As HBM manufacturers increase stack heights and reduce interconnect dimensions, manufacturing defects become more expensive because a failure in one component can reduce the yield of an entire memory stack.
This creates demand for equipment capable of identifying defects before additional manufacturing costs are incurred. KLA's inspection and metrology technologies address critical dimensions, surface defects, and process variations throughout semiconductor manufacturing. Its opportunity extends beyond the number of additional wafers processed because more complex manufacturing processes can require additional measurement and inspection steps.
The financial distinction is important. Lam benefits directly from increased spending on etching and deposition equipment, while KLA benefits from the process-control requirements associated with manufacturing increasingly complex devices. These two sources of demand can grow at different rates, even when they serve the same semiconductor manufacturers.
Applied Materials (AMAT) participates in several HBM-related manufacturing processes, including deposition, chemical mechanical planarization, and advanced interconnect technologies. Its equipment is relevant to TSV fabrication, wafer surface preparation, and other steps involved in producing advanced memory structures.
Applied's position became more interesting on October 26, 2026, when it announced an agreement to increase its ownership of BE Semiconductor Industries, commonly known as BESI, to approximately 20% of outstanding shares. The Dutch company manufactures advanced semiconductor assembly equipment, including technologies used in hybrid bonding.
The investment expands a strategic relationship that began in 2020 and provides Applied with additional exposure to next-generation semiconductor packaging. The proposed ownership increase remains subject to regulatory requirements, so the approximately 20% figure represents the announced ownership target rather than a confirmed completed transaction.
Applied's investment in BESI is strategically significant because hybrid bonding may become increasingly important as semiconductor manufacturers pursue finer interconnect pitches and more advanced chip stacking. However, the investment also introduces a financial consideration: Applied is committing capital to a specialized equipment manufacturer whose future growth depends partly on the timing and economics of hybrid bonding adoption.
Camtek (CAMT) provides more concentrated exposure to advanced packaging inspection and metrology. Its equipment addresses manufacturing defects and dimensional measurement requirements associated with increasingly complex semiconductor packages.
Camtek's September-quarter revenue reached approximately $161 million, increasing 28% year over year. The company also projected revenue exceeding $800 million in 2027, compared with approximately $595 million expected in 2026. This implies growth of roughly 35%, substantially above the growth rate of the broader semiconductor equipment market.
Unlike Lam, KLA, and Applied Materials, Camtek operates with a much smaller revenue base and greater concentration in advanced packaging inspection. This provides the potential for stronger percentage revenue growth as advanced packaging investment expands, but it also increases sensitivity to customer spending decisions and equipment qualification schedules.
Table 1. Latest Reported Quarterly Revenue of Principal HBM-Related Equipment Companies
Japanese semiconductor equipment manufacturer TOWA reported orders of ¥32.97 billion during the July–September 2026 quarter, an increase of 125.8% from the corresponding period of 2025. Quarterly sales increased 14.1% to ¥17.53 billion, substantially below the growth in incoming orders.
This divergence provides a useful indication of equipment demand. TOWA's order-to-sales ratio reached approximately 1.88, meaning that new orders during the quarter were nearly twice the company's recognized revenue. Although orders do not translate immediately into shipments, the ratio suggests that customer commitments are expanding faster than current equipment deliveries.
TOWA also increased its revenue forecast for the fiscal year ending March 2027 from ¥64 billion to ¥71 billion, representing a 10.9% upward revision. Its operating profit forecast increased from ¥10.24 billion to ¥13.15 billion, or 28.4%.
The difference between the two growth rates is particularly important. The additional ¥7 billion in projected revenue is accompanied by ¥2.91 billion in additional operating profit, producing an implied incremental operating margin of 41.6%. This suggests substantial operating leverage as equipment sales increase, although changes in product mix and expenses may also contribute.
TOWA's financial disclosures identify expanding AI and data center investment, including HBM, as important demand drivers. The company also serves other semiconductor packaging applications, so its order growth should not be attributed exclusively to HBM. Nevertheless, the results demonstrate that the broader advanced packaging equipment market is experiencing substantial investment activity.
TOWA's performance becomes more significant when compared with equipment spending at the front end of memory manufacturing. That comparison is possible using the semiconductor revenue breakdown of South Korea's Wonik IPS.
Wonik IPS (240810.KQ), headquartered in Pyeongtaek, South Korea, manufactures semiconductor processing equipment used in DRAM, NAND, and other semiconductor applications. Its products include plasma-enhanced chemical vapor deposition (PECVD), atomic layer deposition (ALD), and thermal processing systems.
Unlike TOWA, which provides exposure to downstream semiconductor packaging, Wonik IPS participates primarily in front-end wafer processing. Its financial performance therefore provides an additional indication of how memory manufacturers are allocating capital across the semiconductor manufacturing process.
Wonik IPS reported preliminary third-quarter 2026 revenue of approximately ₩337.8 billion, an increase of 25.8% year over year. Operating profit reached approximately ₩66.7 billion, increasing 142.0% from the corresponding quarter of 2025. The operating margin was approximately 19.7%.
The company's semiconductor equipment business accounted for most of its revenue. According to the supplied divisional estimates, semiconductor equipment sales increased from ₩189.6 billion in the second quarter to ₩306.6 billion in the third quarter, representing sequential growth of 61.7%.
The annual projections are equally noteworthy. Semiconductor equipment revenue is expected to increase from ₩707.3 billion in 2025 to ₩1.049 trillion in 2026 and ₩1.408 trillion in 2027. That represents cumulative growth of 99.1% over two years.
Table 1 separates Wonik IPS's semiconductor equipment revenue from its display equipment business, which is important because the two segments are experiencing substantially different investment conditions.
Table 2. Wonik IPS Semiconductor Equipment Revenue Growth, 2025–2027 (KRW Billions)
Source: Company Reports. The figures cover all semiconductor equipment applications, not HBM exclusively.
According to Table 2, semiconductor equipment revenue is projected to increase 48.3% in 2026 and another 34.2% in 2027. Meanwhile, display equipment revenue is projected to decline from ₩202.6 billion in 2025 to ₩123.3 billion in 2026 before recovering to ₩178.1 billion in 2027.
The resulting shift in revenue composition is substantial. Semiconductor equipment represented 77.7% of Wonik IPS's total revenue in 2025 but is expected to account for 89.5% in 2026 and 88.8% in 2027.
These figures suggest that semiconductor manufacturing investment, rather than display equipment demand, is driving the company's anticipated growth. However, Wonik IPS supplies equipment for conventional DRAM and NAND manufacturing as well as HBM-related applications. The financial projections therefore represent exposure to the broader memory investment cycle rather than a direct forecast of HBM equipment spending.
The comparison with TOWA is particularly relevant. Wonik IPS is experiencing stronger semiconductor equipment sales associated with front-end manufacturing, while TOWA is receiving substantially higher orders for downstream packaging equipment. Together, the results indicate that memory investment is expanding across multiple stages of semiconductor manufacturing.
The HBM investment opportunity is not confined to American equipment manufacturers. Japanese, South Korean, and European suppliers occupy important positions in wafer thinning, bonding, molding, and other advanced packaging processes.
Table 2 identifies the major equipment manufacturers, their headquarters, and their principal exposure to HBM-related manufacturing. The distinction between front-end wafer fabrication and advanced packaging is important because investment in these equipment categories is driven by different manufacturing requirements.
Table 3. Global Semiconductor Equipment Manufacturers Positioned to Benefit From HBM Investment
Multiple HBM processes; agreement to increase BESI ownership to ~20%
Front-end DRAM, NAND, and HBM-related manufacturing
According to Table 3, the HBM equipment market involves a broad range of manufacturing technologies. Lam Research and Wonik IPS participate in front-end semiconductor processing, while KLA and Camtek provide inspection and metrology capabilities. TOWA, Hanmi Semiconductor, DISCO, and BESI supply specialized technologies associated with advanced packaging.
This distribution of equipment suppliers also illustrates why increasing HBM production does not benefit every manufacturer equally. The transition to more advanced memory structures can increase demand for specific technologies without producing a proportional increase in spending across every equipment category.
For example, higher HBM stack counts increase the importance of die thinning and bonding, while smaller interconnect dimensions place additional requirements on inspection, metrology, and surface preparation. Front-end DRAM capacity expansion, meanwhile, benefits equipment manufacturers supplying etching, deposition, and thermal processing systems.
The opportunity therefore depends on the manufacturing processes being expanded or upgraded, rather than simply the number of HBM devices being produced.
One of the most important technology transitions involves the potential adoption of hybrid bonding in future HBM manufacturing.
Current HBM production primarily uses established bonding processes to connect vertically stacked DRAM dies. These processes have supported commercial production of increasingly complex HBM devices, but higher stack counts and smaller interconnect dimensions create additional challenges involving package thickness, electrical performance, and manufacturing yields.
Hybrid bonding offers a potential alternative by enabling finer-pitch interconnects through direct bonding of prepared surfaces, including copper-to-copper connections. The technology can reduce dependence on conventional microbumps and support more advanced semiconductor integration.
However, hybrid bonding introduces its own manufacturing challenges. The process requires exceptionally flat and clean surfaces, precise alignment, and tight control over bonding defects. These requirements can increase spending on surface preparation, metrology, inspection, and specialized bonding equipment.
This creates an opportunity for companies supplying the technologies required to implement hybrid bonding, but the timing of commercial adoption remains uncertain. Equipment costs, manufacturing yields, and process qualification requirements will influence how rapidly memory manufacturers transition from established bonding technologies.
Applied Materials' planned increase in its BESI ownership provides a strategic connection to this technology transition. BESI has developed advanced bonding and assembly equipment, while Applied supplies materials engineering technologies relevant to surface preparation and semiconductor interconnect formation.
Nevertheless, investors should distinguish between current equipment revenue and prospective demand associated with hybrid bonding. Existing thermocompression bonding equipment continues to support commercial HBM production, while broader adoption of hybrid bonding depends on future manufacturing requirements and economic considerations.
The two technologies may coexist for an extended period, particularly if different HBM configurations require different bonding approaches.
The financial implications of HBM manufacturing extend beyond the current expansion of memory production capacity. As semiconductor manufacturers introduce more complex memory devices, additional spending may be required to upgrade existing equipment, improve manufacturing yields, and qualify new processes.
This is an important distinction from conventional semiconductor capacity expansion. A memory manufacturer can increase output by adding wafer processing capacity, but manufacturing more advanced HBM devices may also require new equipment even when existing wafer capacity remains sufficient.
However, increasing manufacturing complexity does not guarantee that equipment spending will rise proportionately. Improvements in equipment throughput, process integration, and manufacturing yields can reduce the number of additional systems required. Manufacturers may also delay equipment purchases if memory prices weaken or AI infrastructure investment grows more slowly than anticipated.
The distinction is particularly relevant to the outlook for 2027 and 2028. Memory manufacturers are expanding capacity in response to strong demand, but new facilities and equipment installations can eventually increase available supply. If HBM production capacity grows faster than demand, the resulting pressure on memory prices could lead manufacturers to moderate capital expenditures.
Consequently, the equipment investment cycle should not be considered independent of the broader semiconductor market. HBM manufacturing complexity can support additional equipment spending, but it does not eliminate the traditional relationship between memory supply, pricing, profitability, and capital investment.
The expansion of HBM manufacturing is producing measurable financial benefits across the semiconductor equipment industry. TOWA's 125.8% increase in quarterly orders and Wonik IPS's projected 99.1% semiconductor equipment revenue growth between 2025 and 2027 provide evidence that investment is accelerating in both advanced packaging and front-end wafer fabrication. The opportunity is broader than bonding equipment alone, extending to etching, deposition, wafer thinning, inspection, and metrology.
Among the major U.S.-listed equipment manufacturers, KLA offers a compelling combination of process-control technology and exposure to increasingly demanding semiconductor manufacturing requirements. Lam Research provides more direct participation in DRAM wafer fabrication and TSV processing, supported by strong recent revenue growth and operating margin expansion. Applied Materials participates in several manufacturing processes, while its planned increase in BESI ownership provides additional exposure to hybrid bonding. Camtek offers a more concentrated opportunity in advanced packaging inspection, with management projecting approximately 35% revenue growth in 2027.
The principal investment risk is that equipment orders can accelerate ahead of semiconductor production requirements. Memory manufacturers are committing substantial capital to HBM expansion, but additional capacity, improved yields, and changing AI infrastructure demand could eventually moderate equipment spending. The most attractive investment opportunities will therefore depend on the relationship between equipment demand, earnings growth, and stock valuations, rather than exposure to HBM alone.
The broader conclusion is that HBM is changing the composition of semiconductor capital spending. Increasingly complex memory devices require additional manufacturing capabilities, creating opportunities for equipment manufacturers whose technologies improve processing precision, interconnect density, and production yields. For investors, the companies best positioned to benefit are not necessarily those with the greatest exposure to HBM capacity expansion, but those whose equipment remains essential as memory manufacturing advances through successive technology generations.
If you've saved over $1,000,000, this guide is for you. The last thing you want in retirement is to run out of money, you want your money to generate lasting income while you enjoy your life.
Now you can learn the strategies wealthy retirees use to fund their retirement with The Definitive Guide to Retirement Income from Fisher Investments. Download the guide today! (sponsor)
Contact editorial@247wallst.com for any questions or corrections.
