Home / Themes / Memory '26: HBM Testing
Memory '26: HBM Testing
Last updated
Theme thesis · 4/5 sections · Tickers 4 with notes · 2 pending
Bull / Bear Details has the investment thesis and bull/bear points. Overview is monitoring guidance (hiring, forums, second-order trends, search keywords, Google Trends, datasets).
Bull / Bear DetailsThe escalating demand for High Bandwidth Memory (HBM) in AI accelerators, coupled with its increasing architectural complexity and the transition to advanced pa
Thesis
The escalating demand for High Bandwidth Memory (HBM) in AI accelerators, coupled with its increasing architectural complexity and the transition to advanced packaging like hybrid bonding, makes rigorous and multi-stage testing a critical and growing bottleneck, driving significant investment in HBM testing solutions.
Bull case
The rapid scaling of AI GPUs and the demand for long-context, agentic AI systems are driving an exponential increase in HBM content per accelerator and overall HBM demand. This necessitates more extensive and complex testing throughout the HBM manufacturing process, from wafer sort to final package test, to ensure functionality and reliability.
The industry's transition to HBM4E and the adoption of hybrid bonding technology introduce new and more challenging testing requirements. Hybrid bonding, with its metal-to-metal connections and tighter wiring, makes 'known good die' screening and subsequent stack testing significantly more difficult, requiring specialized and advanced probe technologies and equipment.
The high cost of advanced AI accelerators means that a failure in HBM can lead to substantial financial losses. This drives a heightened focus on yield optimization and rigorous testing at every layer of HBM assembly to catch defects early, making advanced testing solutions indispensable for minimizing scrap and maximizing the value of expensive HBM stacks.
Bear case
Despite the strong demand for HBM, the broader memory market remains cyclical. A potential future oversupply in the general DRAM or NAND market, driven by increased capacity from major memory OEMs, could eventually temper overall capital expenditure in memory production, potentially impacting the HBM testing segment, even if HBM demand remains robust.
The fast pace of technological evolution in HBM, particularly with the shift to hybrid bonding, poses a risk of rapid obsolescence for existing testing solutions or a significant shift in market share among testing equipment providers. Companies unable to adapt quickly to new bonding techniques or testing methodologies could lose out to more agile competitors.
There is a risk that large semiconductor manufacturers (memory OEMs or GPU designers) could increasingly internalize advanced HBM testing processes or develop proprietary in-house solutions. This could reduce their reliance on third-party testing equipment and services, potentially limiting the market opportunity for independent HBM testing theme participants.
Key Metrics
| Metric | Cadence | What It Signals | Update Source |
|---|---|---|---|
| Global High Bandwidth Memory (HBM) Market Size (Revenue) | Annually (forecasts, with quarterly updates from major HBM manufacturers) | Accelerating growth in HBM market size indicates strong adoption in AI, HPC, and data centers, directly increasing demand for HBM testing solutions, supporting a bullish view. | LLM_Approved |
| Global Semiconductor Test Equipment Market Revenue (Memory Segment) | Annually (forecasts, with quarterly updates from major test equipment suppliers) | Increasing revenue in the memory test equipment market signifies rising investment in HBM testing capacity and advanced capabilities, indicating a bullish trend for the theme. | LLM_Approved |
| Global Advanced Packaging Equipment Spending | Annually (forecasts, with quarterly updates from major equipment manufacturers) | Growth in advanced packaging equipment spending, driven by HBM and chiplet architectures, indicates increasing complexity and investment in the HBM supply chain, boosting testing demand. | LLM_Approved |
Upcoming Catalysts
| Catalyst ID | Estimated Timing | Estimated Date Start | Estimated Date End | Catalyst | Why It Matters | Ticker Or Theme Specific | Transcript Date | Source Type | Catalyst Source |
|---|---|---|---|---|---|---|---|---|---|
| COHU_e73fa5e1 | this summer and follow-on units later in the year | 2026-06-01 | 2026-12-31 | Initial qualification shipment and subsequent units of a new handler, currently in development, targeting automotive ADAS and physical AI device test. | Successful qualification and ramp of this new handler could drive new revenue streams and expand Cohu's market presence in high-growth automotive ADAS and physical AI segments, which are strategic verticals for the company. | Ticker | 2026-02-12 | earnings_transcript | COHU (ticker) |
| COHU_106d1a34 | this year in HBM between $15 million and $20 million | 2026-01-01 | 2026-12-31 | Achievement of Cohu's High Bandwidth Memory (HBM) revenue target of $15 million to $20 million for the full year 2026. | HBM is a critical growth driver in the AI market, and meeting or exceeding this target would signal strong execution and capitalize on the HBM supercycle, positively impacting revenue and investor sentiment. | Ticker | 2026-02-12 | earnings_transcript | COHU (ticker) |
| COHU_cb758a6c | in the second quarter going into third quarter | 2026-04-01 | 2026-09-30 | Increased shipment rate of the Eclipse handler, particularly for high-performance computing opportunities. | The Eclipse handler is crucial for AI-related high-performance computing and HBM memory. A successful ramp indicates strong demand and execution in a key growth area, impacting revenue and market position. | Ticker | 2026-02-12 | earnings_transcript | COHU (ticker) |
| COHU_7c975498 | coming up towards the middle of the year | 2026-05-01 | 2026-06-30 | Acceleration of a mixed-signal tester design win, specifically for digital controllers used in data centers, from a large semiconductor manufacturer. | This acceleration would signify successful penetration and ramp-up in the critical data center market, driving system revenue and strengthening Cohu's position in high-performance computing. | Ticker | 2026-02-12 | earnings_transcript | COHU (ticker) |
| COHU_93668a56 | in 2026 | 2026-01-01 | 2026-12-31 | Cohu modeling another year of revenue growth in 2026, following 13% growth in 2025. | This is a key financial outlook. Achieving continued revenue growth would validate management's market recovery thesis and execution, positively impacting investor sentiment and valuation. | Ticker | 2026-02-12 | earnings_transcript | COHU (ticker) |
NotesEarnings Summary
| Date | Type | Comment | Detail | Sentiment | Tickers | IS CHANGE |
|---|---|---|---|---|---|---|
| 2026-03-04 | group_thesis | The transcript reinforces HBM testing as a critical AI bottleneck, intensifying with HBM4E's hybrid bonding transition, though some hybrid bonding adoption for HBM4 may be delayed. This drives demand for specialized testing and bonding equipment (FormFactor, BESI, ASMPT, Teradyne, Onto Innovation). Companies like Cohu and PDF Solutions also benefit from increasing complexity. Hanmi Semiconductor is adapting by launching a 'Wide TC Bonder' for HBM5/6, challenging the direct hybrid bonder impact. | Earnings Summary | Bullish | FORM, ONTO, TER, BESI NA, ASMPT, COHU, 6239 TT, PDFS US, 042700 KS | False |