{"id":24276292,"date":"2026-05-12T00:01:51","date_gmt":"2026-05-12T07:01:51","guid":{"rendered":"https:\/\/semiengineering.com\/?p=24276292"},"modified":"2026-05-12T08:23:53","modified_gmt":"2026-05-12T15:23:53","slug":"hbm-shifts-testing-left-to-preserve-ai-chip-yield","status":"publish","type":"post","link":"https:\/\/semiengineering.com\/hbm-shifts-testing-left-to-preserve-ai-chip-yield\/","title":{"rendered":"HBM Shifts Testing Left To Preserve AI Chip Yield"},"content":{"rendered":"<p><strong>Key Takeaways:<\/strong><\/p>\n<ul>\n<li>A high-yield, known-good stack requires multiple test insertions.<\/li>\n<li>Known good stack testing poses challenges for power delivery and thermal management.<\/li>\n<li>The shift to HBM4 and HBM5 will increase the pressure for shift-left test flows.<\/li>\n<\/ul>\n<hr \/>\n<p>Taller high-bandwidth memory (HBM) stacks and tighter TSV pitch are impacting AI module yields. The solution is to push test further left in the manufacturing flow, but that shift comes at a cost.<\/p>\n<p>HBM is an essential ingredient in AI systems, which have become insatiable consumers of memory as the volume of data to be processed and stored continues to grow. Over the past decade, HBM memories have grown from 2 to 12 dies in a stack, and soon will reach 16. Meanwhile, the number of HBM stacks in a multi-die assembly inside an AI data center has increased from 4 to 8.<\/p>\n<p>Today, HBM accounts for nearly half the cost of AI chips. So a defective memory stack found during final test is an expensive failure, which is why there is an increasing focus on known good stacks (KGS). But stacking dies is an intricate and difficult manufacturing process. Alignment of through-silicon vias (TSVs) to microbumps is measured in microns. The thinning and sawing of wafers introduce mechanical stresses that can exacerbate existing cracks, slips, and scratches. On top of that, thermal compression bonding can cause opens, shorts, and head-in-pillow and high-resistive connections.<\/p>\n<p>Worse, finding these real and potential defects is a huge challenge. Stacked die testing includes balancing test coverage with test time, mechanical handling, thermal management, and power delivery. While engineering teams can reduce test costs with DFT and high site count parallel test, stacked die have difficult thermal management problems due to stack height and high power. And all of this will become more difficult with HBM4 and HBM5.<\/p>\n<p>\u201cData from hyperscalers indicates that HBM failures are the number one cause of GPU failures in data centers,&#8221; said Faisal Goriawalla, director of product management for SLM at <a href=\"https:\/\/semiengineering.com\/entities\/synopsys-inc\/\">Synopsys<\/a>. &#8220;Studies also indicate that HBM is more prone to faults than traditional DRAM due to its complex vertical stacking, with column failure (e.g., TSV defects) being particularly common. The move from HBM3 to HBM4 will require further evolution in multi-die support. The 2,048-bit memory interface requires a significant increase in the number of TSVs routed through a memory stack. This will mean shrinking the external bump pitch as the total number of microbumps increases significantly. In addition, support for 16-high TSV stacks brings new complexity in wiring up an even larger number of DRAM dies without defects.\u201d<\/p>\n<p>This calls for more testing earlier in the manufacturing flow so that faulty stacks can be scrapped prior to assembly. Today, to produce a shippable HBM stacked die, the test process includes multiple insertions at both wafer-level and stacked-die level. Both the HBM logic and HBM DRAM dies go through wafer test. Each DRAM will go through multiple insertions \u2014 wafer-level burn-in, hot and cold testing, and repair \u2014 after which the DRAM wafer is thinned, bumped, and singulated. DRAM dies are then stacked upon the wafer of logic base die and go through a series of test insertions. Depending on the assembly manufacturer\u2019s process, the testing could be done after each DRAM die is stacked, or it could be done after 2 or 4 dies are stacked. Eventually the stacked wafer is singulated.<\/p>\n<p><img data-recalc-dims=\"1\" loading=\"lazy\" decoding=\"async\" class=\"alignnone wp-image-24276295\" src=\"https:\/\/i0.wp.com\/semiengineering.com\/wp-content\/uploads\/2026\/05\/Ann-5_11-Fig-1.jpg?resize=750%2C578&#038;ssl=1\" alt=\"\" width=\"750\" height=\"578\" srcset=\"https:\/\/i0.wp.com\/semiengineering.com\/wp-content\/uploads\/2026\/05\/Ann-5_11-Fig-1.jpg?resize=300%2C231&amp;ssl=1 300w, https:\/\/i0.wp.com\/semiengineering.com\/wp-content\/uploads\/2026\/05\/Ann-5_11-Fig-1.jpg?resize=600%2C461&amp;ssl=1 600w, https:\/\/i0.wp.com\/semiengineering.com\/wp-content\/uploads\/2026\/05\/Ann-5_11-Fig-1.jpg?w=1431&amp;ssl=1 1431w\" sizes=\"auto, (max-width: 750px) 100vw, 750px\" \/><\/p>\n<p><strong>Fig. 1: Generic manufacturing test process for HBM known good stacks comparing pre-singulated and post-singulated flows. Source Teradyne<\/strong><\/p>\n<p>Hypothetically, one could test the singulated stacked HBM DRAM. However, that is not currently done.<\/p>\n<p>\u201cThe test flow on the left is the standard today,&#8221; explained Hanh Lai, product marketing manager in the memory division at <a href=\"https:\/\/semiengineering.com\/entities\/teradyne-corporation\/\">Teradyne<\/a>. &#8220;It seems to be the most cost-effective and more risk-averse, because it doesn&#8217;t require probing an exotic structure, such as a singulated stacked die, though having dies stacked on a wafer does present a planarity challenge. But testing of singulated stacked dies is of interest, and prober companies are developing solutions. The reason is that integrators (i.e., Nvidia and AMD) are concerned about final package yield. In the typical GPU package, a GPU is surrounded by eight HBM stacks. One defective HBM stack becomes very costly at that particular point.\u201d<\/p>\n<p>The call for shift left becomes louder as the cost of one bad die or bad stack rises.<\/p>\n<p>\u201cIt all comes down to cost,&#8221; said Vernon Rodgers, executive vice president of sales and marketing at Aehr Test Systems. &#8220;The ability to reduce the scrappage, improve yield, and reduce waste drives the test choices. Maybe in the past it was too expensive to move left, but today the yield cost curve definitely is driving it more and more left. Consider wafer level burn-in. Wafer level burn-in reduces infant mortality related defects. As we move to [more] dies per stack and bigger packages, this only becomes more important.\u201d<\/p>\n<p>Others agree. \u201cAs HBM devices grow in complexity and cost, test content continues to shift earlier in the flow,&#8221; said Kevin Tran, senior director of product marketing at FormFactor. &#8220;This shift left helps prevent defective dies from entering expensive stacking processes and drives increased use of high\u2011speed test, wider parallelism, and tighter thermal control at wafer test.\u201d<\/p>\n<p><strong>Wafer test and burn-in<\/strong><br \/>\nThe path to known-good stacks begins with known-good dies (KGDs). A thorough wafer test of each DRAM and the logic base die requires testing internal circuitry, core memories, and the TSVs.<\/p>\n<p>DRAMs require thousands of test patterns specific to the memory architecture. Due to the bit-cell density, the application of redundancy plays an essential role during the test process. Without it the wafer-level yield would be significantly lower. Patterns are supplied by ATEs, and to reduce test costs, DRAM dies are tested in parallel \u2014 typically 64 to 128 sites.<\/p>\n<p>Industry experts emphasize the importance of testing the logic base die because it provides the only access to the memory dies in the stack. In addition, the impact to stacked die yield is significant. \u201cConsider the stacked die \u2014 one base logic, and then 8 to 16 HBM,&#8221; said Aehr Test\u2019s Rodgers. &#8220;Making sure that base logic device is of the highest quality is critical, because if it&#8217;s bad, 16 dies get thrown away. So it&#8217;s a huge multiplier in the yield curve.&#8221;<\/p>\n<p>The logic base die test requirements focus on the DFT circuitry, which enables HBM DRAM test during the stacking process and throughout its lifecycle. This testing relies upon JEDEC-specified direct access [1-3] or IEEE 1500 [4] using a limited number of pads or microbumps. Applying test content at logic wafer test ensures there are no defects in the internal logic, circuitry for IEEE 1500, direct access bus, memory built-in self-test (MBiST), internal logic, TSV connectivity, and PHY circuitry.<\/p>\n<p>Nevertheless, the challenges with wafer probing increase with each new HBM generation.<\/p>\n<p>\u201cAt advanced DRAM process nodes, especially those used in HBM, wafer-level test is not just limited to contact and functional screening. It becomes a multi-dimensional challenge across mechanical, power delivery, signal integrity, and throughput,\u201d said FormFactor\u2019s Tran. &#8220;Shrinking pad geometries can be addressed with advanced MEMS probe technologies, which provide tighter pitch and improved accuracy control. New speed and power requirements in HBM4 and HBM5 demand data transfer rates exceeding 10\u202fGbps in future generations, and power levels of up to 100\u202fwatts per HBM stack. MEMS probes also offer higher current capability and, when combined with optimized probe card-level power delivery design, can address the high-power, high-speed requirements of KGD testing.\u201d<\/p>\n<p>Typical DRAM test flows include wafer-level burn-in, during which latent defects are accelerated so that standard testing can be done. \u201cWith burn in, two things are addressed. One, you\u2019re looking for devices that are weak, such as gate oxide issues. Two, because bit-cells are capacitors, you need to stabilize their values,\u201d said Aehr Test\u2019s Rodgers. \u201cThe debate has always been, do you burn-in at wafer, singulated die, or package? But now, when you start stacking, you want to move left, and it&#8217;s really what&#8217;s driving wafer-level burn-in.\u201d<\/p>\n<p>The contact solution for wafer-level burn-in needs to address the mechanical challenges that come with electrically contacting the test access pads or bumps. This can be done for 300mm wafers using MEMS technology or with micro-pogo pins.<\/p>\n<p><img data-recalc-dims=\"1\" loading=\"lazy\" decoding=\"async\" class=\"alignnone wp-image-24276294\" src=\"https:\/\/i0.wp.com\/semiengineering.com\/wp-content\/uploads\/2026\/05\/Ann-5_11-Fig-2.jpg?resize=750%2C385&#038;ssl=1\" alt=\"\" width=\"750\" height=\"385\" srcset=\"https:\/\/i0.wp.com\/semiengineering.com\/wp-content\/uploads\/2026\/05\/Ann-5_11-Fig-2.jpg?resize=300%2C154&amp;ssl=1 300w, https:\/\/i0.wp.com\/semiengineering.com\/wp-content\/uploads\/2026\/05\/Ann-5_11-Fig-2.jpg?resize=1024%2C525&amp;ssl=1 1024w, https:\/\/i0.wp.com\/semiengineering.com\/wp-content\/uploads\/2026\/05\/Ann-5_11-Fig-2.jpg?resize=768%2C393&amp;ssl=1 768w, https:\/\/i0.wp.com\/semiengineering.com\/wp-content\/uploads\/2026\/05\/Ann-5_11-Fig-2.jpg?resize=600%2C307&amp;ssl=1 600w, https:\/\/i0.wp.com\/semiengineering.com\/wp-content\/uploads\/2026\/05\/Ann-5_11-Fig-2.jpg?w=1427&amp;ssl=1 1427w\" sizes=\"auto, (max-width: 750px) 100vw, 750px\" \/><\/p>\n<p><strong>Fig. 2: MEMS wafer-level contactor for burn-in. Source Aehr Test<\/strong><\/p>\n<p>Combining DFT with probing on aluminum test pads assists in lowering test costs. In specifying the microbump layout for HBM I\/O, the JEDEC standard includes space for adding sacrificial test pads.<\/p>\n<p>\u201cWhen you use a sacrificial pad and space them out, your probe card becomes much cheaper. You don&#8217;t need to go spend $500,000 on a probe card. They can save up to 80% on the cost of a probe card. DFT enables quality, but what it really enables is a lower cost approach to wafer-level burn-in,\u201d noted Rodgers. \u201cYou can go to micro pogo versus MEMS. I can offer the technology at two very different cost points, but the DFT will determine which cost point you actually enter.\u201d<\/p>\n<p><strong>Stack test<\/strong><br \/>\nTesting stacked dies reduces the yield risk of the AI product\u2019s final test. As noted earlier, the standard manufacturing and test flow stacks HBM on the base die in wafer form. Then, the test interface probes the backside of the wafer. Multi-site testing is standard. But stacking dies poses challenges in terms of thermal management, power delivery, and mechanical handling. And as test insertions increase, containing test costs becomes more difficult. The number of test insertions for a 12-stack die can range from 3 to 12, depending on the assembly house\u2019s quality levels.<\/p>\n<p>\u201cStacking DRAM dies can introduce new errors into the process, including issues related to higher\u2011speed transfers within the stack, increased power and current requirements for taller stacks, and thermal challenges associated with these factors,\u201d noted FormFactor\u2019s Tran. \u201cTesting and sorting stacked die can eliminate bad die early and help reduce the overall cost of test. Stacked die testing requires alignment accuracy at the single-digit micron level. HBM5 requires up to 16\u2011high stacks, making alignment accuracy even more critical to accommodate TSV and bonding tolerances.\u201d<\/p>\n<p>Others point to the importance of intermediate testing of stacks during the assembly process.<\/p>\n<p>\u201cIntermediate testing during the assembly process is becoming increasingly critical, especially as the cost of HBM continues to rise. To support this, new contact mechanisms are being developed to enable reliable testing at intermediate stages,\u201d said Omer Dossani, vice president of global test services at <a href=\"https:\/\/semiengineering.com\/entities\/amkor-technology\/\">Amkor Technology<\/a>. \u201cMany of these challenges are resolved by the time we reach high-volume manufacturing (HVM) in our factory. However, they remain significant manufacturing considerations, requiring increasingly stringent control of temperature stability during testing, as well as the use of specialized sockets, dedicated cleaning materials, and enhanced data monitoring throughout the manufacturing process.\u201d<\/p>\n<p>When it comes to a test solution, power delivery and thermal management are always considerations, but they become complicated as the stack height grows. \u201cIf you look at a 16-story building, the sun hits the outside and the center doesn&#8217;t really get the heat,&#8221; said Aehr Test\u2019s Rodgers. &#8220;It&#8217;s the reverse in a stacked die. The outside is cooled, but how do you get the heat out of the center? With stack die burn-in or test, how you manage the thermals in the center die is important.&#8221;<\/p>\n<p>Stacking dies necessitates good thermal and power management. \u201cThe problem is how to manage thermal output from these devices,&#8221; said Teradyne\u2019s Lai. &#8220;The prober company needs to dissipate the heat from the HBM stacks. Right now, our tester has the capability of testing up to 128 devices in parallel, depending upon the device requirement in terms of pin count and power. From generation HBM3 to HBM4, the power increase is likely to be more than 2X. The challenge for both the prober and probe card companies is to dissipate the heat from these high-powered devices.\u201d<\/p>\n<p>Testing at post-singulated die is an attractive shift-left test prior to 2.5D integration with the final AI chip. It also enables active thermal control versus the passive thermal control used for full-wafer testing. This capability provides more precise temperature during test. The solution for testing singulated stacks involves several technologies \u2014 die carrier for stacks, loader\/unloader equipment, handler for stacked die, and active thermal control. None of these approaches is inexpensive, and each requires development into a high-volume manufacturing (HVM) solution.<\/p>\n<p>The current approach for testing stacked die is pre-singulation, probing the backside of the logic base die on aluminum pads for which there is a specified space amid the layout of microbumps. As such, the ATE needs the instrumentation to test both logic and memory, and with up to 128 test sites for parallel test, the power delivery demand is significant.<\/p>\n<p>After the DRAM is stacked on base die, testing the core memory can be done with the logic base die\u2019s MBiST (often programmable) or a direct access bus. Repair for defective TSVs occurs during each test insertion.<\/p>\n<p>\u201cThe SoC designer must therefore be able to deploy a flexible BiST engine that allows different algorithms to be used to trade off high coverage versus test time, depending on the use case scenario (manufacturing test vs. power-on self-test (POST) vs. in-system debug and diagnosis),\u201d said Synopsys\u2019s Goriawalla. \u201cThis engine must be programmable to handle different latencies, and address ranges and timing of test operations that vary across DRAM manufacturers. It may also need to support post-package repair (PPR) for HBM DRAM to delay any &#8216;truck roll-out&#8217; for in-field service. The diagnostics performed by the BiST engine must be precise, showing the failing bank, row address, column address, etc., if there is a defect detected in the DRAM stack.&#8221;<\/p>\n<p><strong><img data-recalc-dims=\"1\" loading=\"lazy\" decoding=\"async\" class=\"alignnone wp-image-24276293\" src=\"https:\/\/i0.wp.com\/semiengineering.com\/wp-content\/uploads\/2026\/05\/Ann-5_11-Fig-3.jpg?resize=750%2C533&#038;ssl=1\" alt=\"\" width=\"750\" height=\"533\" srcset=\"https:\/\/i0.wp.com\/semiengineering.com\/wp-content\/uploads\/2026\/05\/Ann-5_11-Fig-3.jpg?resize=300%2C213&amp;ssl=1 300w, https:\/\/i0.wp.com\/semiengineering.com\/wp-content\/uploads\/2026\/05\/Ann-5_11-Fig-3.jpg?resize=1024%2C726&amp;ssl=1 1024w, https:\/\/i0.wp.com\/semiengineering.com\/wp-content\/uploads\/2026\/05\/Ann-5_11-Fig-3.jpg?resize=768%2C545&amp;ssl=1 768w, https:\/\/i0.wp.com\/semiengineering.com\/wp-content\/uploads\/2026\/05\/Ann-5_11-Fig-3.jpg?resize=600%2C426&amp;ssl=1 600w, https:\/\/i0.wp.com\/semiengineering.com\/wp-content\/uploads\/2026\/05\/Ann-5_11-Fig-3.jpg?w=1066&amp;ssl=1 1066w\" sizes=\"auto, (max-width: 750px) 100vw, 750px\" \/><\/strong><\/p>\n<p><strong>Fig. 3: Example of fault detection in an HBM stack, which is required for repairing TSV connectivity. Source Synopsys<\/strong><\/p>\n<p><strong>Conclusion<\/strong><br \/>\nWhile HBM DRAM vendors can now charge a premium, their focus remains low-cost. \u201cThe mentality of memory manufacturers is that the test solution needs to be optimized for low-cost \u2014 more so than the SoC vendors who operate with very short lifecycle products,&#8221; said Teradyne&#8217;s Lai. &#8220;HBM manufacturers are in the business for a very long time in a very competitive market, which historically has low margins.&#8221;<\/p>\n<p>Still, the economic pressures of scrappage are driving HBM stacked-die vendors to more testing earlier in the flow, which invariably increases test costs. But that may be offset with a flexible MBiST on the base die, which permits tradeoffs on test content. Testing stacked dies with a high level of parallelism raises the bar for both ATE power delivery and thermal management solutions. And finally, there is still the prospect of testing post-singulated stacked dies, which remains to be proven with unknown economic impact.<\/p>\n<p><strong>References<\/strong><\/p>\n<ol>\n<li>HBM JEDEC <a href=\"https:\/\/www.jedec.org\/standards-documents\/docs\/jesd235a\">https:\/\/www.jedec.org\/standards-documents\/docs\/jesd235a<\/a><\/li>\n<li>HBM3 JEDEC <a href=\"https:\/\/www.jedec.org\/system\/files\/docs\/JESD238B.01.pdf\">https:\/\/www.jedec.org\/system\/files\/docs\/JESD238B.01.pdf<\/a><\/li>\n<li>HBM4 JEDEC <a href=\"https:\/\/www.jedec.org\/system\/files\/docs\/JESD270-4A.pdf\">https:\/\/www.jedec.org\/system\/files\/docs\/JESD270-4A.pdf<\/a><\/li>\n<li>IEEE 1500 <a href=\"https:\/\/standards.ieee.org\/ieee\/1500\/7704\/\">https:\/\/standards.ieee.org\/ieee\/1500\/7704\/<\/a><\/li>\n<\/ol>\n<hr \/>\n<p><strong>Related Articles<\/strong><\/p>\n<p><a href=\"https:\/\/semiengineering.com\/ai-accelerators-usher-in-new-era-of-semiconductor-test\/\">AI Accelerators Usher In New Era For IC Test<\/a><br \/>\nThe number and variety of test interfaces, coupled with increased packaging complexity, are adding a slew of new challenges.<\/p>\n<p><a href=\"https:\/\/semiengineering.com\/hbm-leads-the-way-to-defect-free-bumps\/\">HBM Leads The Way To Defect-Free Bumps<\/a><br \/>\nBump scaling is pushing defect inspection to the limit. What comes next and why it matters.<\/p>\n<p><a href=\"https:\/\/semiengineering.com\/hbm4-sticks-with-microbumps-postponing-hybrid-bonding\/\">HBM4 Sticks With Microbumps, Postponing Hybrid Bonding<\/a><br \/>\nProcess cost and yield issues delay the adoption of hybrid bonding.<\/p>\n<p><a href=\"https:\/\/semiengineering.com\/chiplet-interfaces-embrace-failures\/\">Chiplet Interfaces Embrace Failures<\/a><br \/>\nWhy lane swapping is essential to meet assembly yield.<\/p>\n","protected":false},"excerpt":{"rendered":"<p>Testing sooner and more often can improve quality and reduce scrap, but it&#8217;s also more costly.<\/p>\n","protected":false},"author":789,"featured_media":24276293,"comment_status":"open","ping_status":"open","sticky":false,"template":"","format":"standard","meta":{"_acf_changed":false,"_jetpack_memberships_contains_paid_content":false,"footnotes":"","jetpack_publicize_message":"","jetpack_publicize_feature_enabled":true,"jetpack_social_post_already_shared":true,"jetpack_social_options":{"image_generator_settings":{"template":"highway","default_image_id":0,"font":"","enabled":false},"version":2}},"categories":[9786,2],"tags":[11805,793,68680,336,27,688,961,1379,701,43219,68678,68677,2362,48486,6712,7,1728,26],"class_list":["post-24276292","post","type-post","status-publish","format-standard","has-post-thumbnail","hentry","category-packaging-and-test","category-top-stories","tag-aehr-test-systems","tag-amkor","tag-back-side-test","tag-dft","tag-dram","tag-formfactor","tag-hbm","tag-high-bandwidth-memory","tag-jedec","tag-kgd","tag-kgs","tag-known-good-stack","tag-mbist","tag-parallel-test","tag-repair","tag-synopsys","tag-teradyne","tag-tsv"],"acf":[],"yoast_head":"<!-- This site is optimized with the Yoast SEO Premium plugin v27.7 (Yoast SEO v27.7) - https:\/\/yoast.com\/product\/yoast-seo-premium-wordpress\/ -->\n<title>HBM Shifts Testing Left To Preserve AI Chip Yield<\/title>\n<meta name=\"robots\" content=\"index, follow, max-snippet:-1, max-image-preview:large, max-video-preview:-1\" \/>\n<link rel=\"canonical\" href=\"https:\/\/semiengineering.com\/hbm-shifts-testing-left-to-preserve-ai-chip-yield\/\" \/>\n<meta property=\"og:locale\" content=\"en_US\" \/>\n<meta property=\"og:type\" content=\"article\" \/>\n<meta property=\"og:title\" content=\"HBM Shifts Testing Left To Preserve AI Chip Yield\" \/>\n<meta property=\"og:description\" content=\"Testing sooner and more often can improve quality and reduce scrap, but it&#039;s also more costly.\" \/>\n<meta property=\"og:url\" content=\"https:\/\/semiengineering.com\/hbm-shifts-testing-left-to-preserve-ai-chip-yield\/\" \/>\n<meta property=\"og:site_name\" content=\"Semiconductor Engineering\" \/>\n<meta property=\"article:publisher\" content=\"https:\/\/www.facebook.com\/SemiEngineering\" \/>\n<meta property=\"article:published_time\" content=\"2026-05-12T07:01:51+00:00\" \/>\n<meta property=\"article:modified_time\" content=\"2026-05-12T15:23:53+00:00\" \/>\n<meta property=\"og:image\" content=\"https:\/\/semiengineering.com\/wp-content\/uploads\/2026\/05\/Ann-5_11-Fig-3.jpg\" \/>\n\t<meta property=\"og:image:width\" content=\"1066\" \/>\n\t<meta property=\"og:image:height\" content=\"756\" \/>\n\t<meta property=\"og:image:type\" content=\"image\/jpeg\" \/>\n<meta name=\"author\" content=\"Anne Meixner\" \/>\n<meta name=\"twitter:card\" content=\"summary_large_image\" \/>\n<meta name=\"twitter:creator\" content=\"@SemiEngineering\" \/>\n<meta name=\"twitter:site\" content=\"@SemiEngineering\" \/>\n<meta name=\"twitter:label1\" content=\"Written by\" \/>\n\t<meta name=\"twitter:data1\" content=\"Anne Meixner\" \/>\n\t<meta name=\"twitter:label2\" content=\"Est. reading time\" \/>\n\t<meta name=\"twitter:data2\" content=\"12 minutes\" \/>\n<script type=\"application\/ld+json\" class=\"yoast-schema-graph\">{\"@context\":\"https:\\\/\\\/schema.org\",\"@graph\":[{\"@type\":\"Article\",\"@id\":\"https:\\\/\\\/semiengineering.com\\\/hbm-shifts-testing-left-to-preserve-ai-chip-yield\\\/#article\",\"isPartOf\":{\"@id\":\"https:\\\/\\\/semiengineering.com\\\/hbm-shifts-testing-left-to-preserve-ai-chip-yield\\\/\"},\"author\":{\"name\":\"Anne Meixner\",\"@id\":\"https:\\\/\\\/semiengineering.com\\\/#\\\/schema\\\/person\\\/09642fe6b88c48eed5f0273e74c05f98\"},\"headline\":\"HBM Shifts Testing Left To Preserve AI Chip Yield\",\"datePublished\":\"2026-05-12T07:01:51+00:00\",\"dateModified\":\"2026-05-12T15:23:53+00:00\",\"mainEntityOfPage\":{\"@id\":\"https:\\\/\\\/semiengineering.com\\\/hbm-shifts-testing-left-to-preserve-ai-chip-yield\\\/\"},\"wordCount\":2571,\"commentCount\":1,\"image\":{\"@id\":\"https:\\\/\\\/semiengineering.com\\\/hbm-shifts-testing-left-to-preserve-ai-chip-yield\\\/#primaryimage\"},\"thumbnailUrl\":\"https:\\\/\\\/i0.wp.com\\\/semiengineering.com\\\/wp-content\\\/uploads\\\/2026\\\/05\\\/Ann-5_11-Fig-3.jpg?fit=1066%2C756&ssl=1\",\"keywords\":[\"Aehr Test Systems\",\"Amkor\",\"back-side test\",\"DFT\",\"DRAM\",\"FormFactor\",\"HBM\",\"high-bandwidth memory\",\"JEDEC\",\"KGD\",\"KGS\",\"known good stack\",\"MBiST\",\"parallel test\",\"Repair\",\"Synopsys\",\"Teradyne\",\"TSV\"],\"articleSection\":[\"Test, Measurement and Analytics\",\"Top Stories\"],\"inLanguage\":\"en-US\",\"potentialAction\":[{\"@type\":\"CommentAction\",\"name\":\"Comment\",\"target\":[\"https:\\\/\\\/semiengineering.com\\\/hbm-shifts-testing-left-to-preserve-ai-chip-yield\\\/#respond\"]}],\"copyrightYear\":\"2026\",\"copyrightHolder\":{\"@id\":\"https:\\\/\\\/semiengineering.com\\\/#organization\"}},{\"@type\":\"WebPage\",\"@id\":\"https:\\\/\\\/semiengineering.com\\\/hbm-shifts-testing-left-to-preserve-ai-chip-yield\\\/\",\"url\":\"https:\\\/\\\/semiengineering.com\\\/hbm-shifts-testing-left-to-preserve-ai-chip-yield\\\/\",\"name\":\"HBM Shifts Testing Left To Preserve AI Chip Yield\",\"isPartOf\":{\"@id\":\"https:\\\/\\\/semiengineering.com\\\/#website\"},\"primaryImageOfPage\":{\"@id\":\"https:\\\/\\\/semiengineering.com\\\/hbm-shifts-testing-left-to-preserve-ai-chip-yield\\\/#primaryimage\"},\"image\":{\"@id\":\"https:\\\/\\\/semiengineering.com\\\/hbm-shifts-testing-left-to-preserve-ai-chip-yield\\\/#primaryimage\"},\"thumbnailUrl\":\"https:\\\/\\\/i0.wp.com\\\/semiengineering.com\\\/wp-content\\\/uploads\\\/2026\\\/05\\\/Ann-5_11-Fig-3.jpg?fit=1066%2C756&ssl=1\",\"datePublished\":\"2026-05-12T07:01:51+00:00\",\"dateModified\":\"2026-05-12T15:23:53+00:00\",\"author\":{\"@id\":\"https:\\\/\\\/semiengineering.com\\\/#\\\/schema\\\/person\\\/09642fe6b88c48eed5f0273e74c05f98\"},\"breadcrumb\":{\"@id\":\"https:\\\/\\\/semiengineering.com\\\/hbm-shifts-testing-left-to-preserve-ai-chip-yield\\\/#breadcrumb\"},\"inLanguage\":\"en-US\",\"potentialAction\":[{\"@type\":\"ReadAction\",\"target\":[\"https:\\\/\\\/semiengineering.com\\\/hbm-shifts-testing-left-to-preserve-ai-chip-yield\\\/\"]}]},{\"@type\":\"ImageObject\",\"inLanguage\":\"en-US\",\"@id\":\"https:\\\/\\\/semiengineering.com\\\/hbm-shifts-testing-left-to-preserve-ai-chip-yield\\\/#primaryimage\",\"url\":\"https:\\\/\\\/i0.wp.com\\\/semiengineering.com\\\/wp-content\\\/uploads\\\/2026\\\/05\\\/Ann-5_11-Fig-3.jpg?fit=1066%2C756&ssl=1\",\"contentUrl\":\"https:\\\/\\\/i0.wp.com\\\/semiengineering.com\\\/wp-content\\\/uploads\\\/2026\\\/05\\\/Ann-5_11-Fig-3.jpg?fit=1066%2C756&ssl=1\",\"width\":1066,\"height\":756},{\"@type\":\"BreadcrumbList\",\"@id\":\"https:\\\/\\\/semiengineering.com\\\/hbm-shifts-testing-left-to-preserve-ai-chip-yield\\\/#breadcrumb\",\"itemListElement\":[{\"@type\":\"ListItem\",\"position\":1,\"name\":\"Home\",\"item\":\"https:\\\/\\\/semiengineering.com\\\/\"},{\"@type\":\"ListItem\",\"position\":2,\"name\":\"HBM Shifts Testing Left To Preserve AI Chip Yield\"}]},{\"@type\":\"WebSite\",\"@id\":\"https:\\\/\\\/semiengineering.com\\\/#website\",\"url\":\"https:\\\/\\\/semiengineering.com\\\/\",\"name\":\"Semiconductor Engineering\",\"description\":\"Deep Insights For Chip Engineers\",\"potentialAction\":[{\"@type\":\"SearchAction\",\"target\":{\"@type\":\"EntryPoint\",\"urlTemplate\":\"https:\\\/\\\/semiengineering.com\\\/?s={search_term_string}\"},\"query-input\":{\"@type\":\"PropertyValueSpecification\",\"valueRequired\":true,\"valueName\":\"search_term_string\"}}],\"inLanguage\":\"en-US\"},{\"@type\":\"Person\",\"@id\":\"https:\\\/\\\/semiengineering.com\\\/#\\\/schema\\\/person\\\/09642fe6b88c48eed5f0273e74c05f98\",\"name\":\"Anne Meixner\",\"image\":{\"@type\":\"ImageObject\",\"inLanguage\":\"en-US\",\"@id\":\"https:\\\/\\\/secure.gravatar.com\\\/avatar\\\/5ef77e855d4b5df19e9d5a817f51e2e3c2bcc80cfba5c2285d2f4886ed94cd1a?s=96&d=mm&r=g\",\"url\":\"https:\\\/\\\/secure.gravatar.com\\\/avatar\\\/5ef77e855d4b5df19e9d5a817f51e2e3c2bcc80cfba5c2285d2f4886ed94cd1a?s=96&d=mm&r=g\",\"contentUrl\":\"https:\\\/\\\/secure.gravatar.com\\\/avatar\\\/5ef77e855d4b5df19e9d5a817f51e2e3c2bcc80cfba5c2285d2f4886ed94cd1a?s=96&d=mm&r=g\",\"caption\":\"Anne Meixner\"},\"description\":\"Anne Meixner is a contributing editor at Semiconductor Engineering. She has 30+ years in the semiconductor industry. She became fascinated by defects in the semiconductor manufacturing process as a young engineer at IBM. Over that period, she has focused on test methodology with emphasis on mixed-signal and analog DFT and test. As a technical communicator she takes complex ideas and explains them in consumable understandable pieces. Meixner has worked at IBM, Carnegie Mellon University and Intel. She holds three U.S. patents and her peers have recognized her work with two best papers at IEEE International Test Conference. She founded The Engineers\u2019 Daughter in 2015 to consult on semiconductor testing and to coach engineers.\",\"url\":\"https:\\\/\\\/semiengineering.com\\\/author\\\/anne-meixner\\\/\"}]}<\/script>\n<!-- \/ Yoast SEO Premium plugin. -->","yoast_head_json":{"title":"HBM Shifts Testing Left To Preserve AI Chip Yield","robots":{"index":"index","follow":"follow","max-snippet":"max-snippet:-1","max-image-preview":"max-image-preview:large","max-video-preview":"max-video-preview:-1"},"canonical":"https:\/\/semiengineering.com\/hbm-shifts-testing-left-to-preserve-ai-chip-yield\/","og_locale":"en_US","og_type":"article","og_title":"HBM Shifts Testing Left To Preserve AI Chip Yield","og_description":"Testing sooner and more often can improve quality and reduce scrap, but it's also more costly.","og_url":"https:\/\/semiengineering.com\/hbm-shifts-testing-left-to-preserve-ai-chip-yield\/","og_site_name":"Semiconductor Engineering","article_publisher":"https:\/\/www.facebook.com\/SemiEngineering","article_published_time":"2026-05-12T07:01:51+00:00","article_modified_time":"2026-05-12T15:23:53+00:00","og_image":[{"width":1066,"height":756,"url":"https:\/\/semiengineering.com\/wp-content\/uploads\/2026\/05\/Ann-5_11-Fig-3.jpg","type":"image\/jpeg"}],"author":"Anne Meixner","twitter_card":"summary_large_image","twitter_creator":"@SemiEngineering","twitter_site":"@SemiEngineering","twitter_misc":{"Written by":"Anne Meixner","Est. reading time":"12 minutes"},"schema":{"@context":"https:\/\/schema.org","@graph":[{"@type":"Article","@id":"https:\/\/semiengineering.com\/hbm-shifts-testing-left-to-preserve-ai-chip-yield\/#article","isPartOf":{"@id":"https:\/\/semiengineering.com\/hbm-shifts-testing-left-to-preserve-ai-chip-yield\/"},"author":{"name":"Anne Meixner","@id":"https:\/\/semiengineering.com\/#\/schema\/person\/09642fe6b88c48eed5f0273e74c05f98"},"headline":"HBM Shifts Testing Left To Preserve AI Chip Yield","datePublished":"2026-05-12T07:01:51+00:00","dateModified":"2026-05-12T15:23:53+00:00","mainEntityOfPage":{"@id":"https:\/\/semiengineering.com\/hbm-shifts-testing-left-to-preserve-ai-chip-yield\/"},"wordCount":2571,"commentCount":1,"image":{"@id":"https:\/\/semiengineering.com\/hbm-shifts-testing-left-to-preserve-ai-chip-yield\/#primaryimage"},"thumbnailUrl":"https:\/\/i0.wp.com\/semiengineering.com\/wp-content\/uploads\/2026\/05\/Ann-5_11-Fig-3.jpg?fit=1066%2C756&ssl=1","keywords":["Aehr Test Systems","Amkor","back-side test","DFT","DRAM","FormFactor","HBM","high-bandwidth memory","JEDEC","KGD","KGS","known good stack","MBiST","parallel test","Repair","Synopsys","Teradyne","TSV"],"articleSection":["Test, Measurement and Analytics","Top Stories"],"inLanguage":"en-US","potentialAction":[{"@type":"CommentAction","name":"Comment","target":["https:\/\/semiengineering.com\/hbm-shifts-testing-left-to-preserve-ai-chip-yield\/#respond"]}],"copyrightYear":"2026","copyrightHolder":{"@id":"https:\/\/semiengineering.com\/#organization"}},{"@type":"WebPage","@id":"https:\/\/semiengineering.com\/hbm-shifts-testing-left-to-preserve-ai-chip-yield\/","url":"https:\/\/semiengineering.com\/hbm-shifts-testing-left-to-preserve-ai-chip-yield\/","name":"HBM Shifts Testing Left To Preserve AI Chip Yield","isPartOf":{"@id":"https:\/\/semiengineering.com\/#website"},"primaryImageOfPage":{"@id":"https:\/\/semiengineering.com\/hbm-shifts-testing-left-to-preserve-ai-chip-yield\/#primaryimage"},"image":{"@id":"https:\/\/semiengineering.com\/hbm-shifts-testing-left-to-preserve-ai-chip-yield\/#primaryimage"},"thumbnailUrl":"https:\/\/i0.wp.com\/semiengineering.com\/wp-content\/uploads\/2026\/05\/Ann-5_11-Fig-3.jpg?fit=1066%2C756&ssl=1","datePublished":"2026-05-12T07:01:51+00:00","dateModified":"2026-05-12T15:23:53+00:00","author":{"@id":"https:\/\/semiengineering.com\/#\/schema\/person\/09642fe6b88c48eed5f0273e74c05f98"},"breadcrumb":{"@id":"https:\/\/semiengineering.com\/hbm-shifts-testing-left-to-preserve-ai-chip-yield\/#breadcrumb"},"inLanguage":"en-US","potentialAction":[{"@type":"ReadAction","target":["https:\/\/semiengineering.com\/hbm-shifts-testing-left-to-preserve-ai-chip-yield\/"]}]},{"@type":"ImageObject","inLanguage":"en-US","@id":"https:\/\/semiengineering.com\/hbm-shifts-testing-left-to-preserve-ai-chip-yield\/#primaryimage","url":"https:\/\/i0.wp.com\/semiengineering.com\/wp-content\/uploads\/2026\/05\/Ann-5_11-Fig-3.jpg?fit=1066%2C756&ssl=1","contentUrl":"https:\/\/i0.wp.com\/semiengineering.com\/wp-content\/uploads\/2026\/05\/Ann-5_11-Fig-3.jpg?fit=1066%2C756&ssl=1","width":1066,"height":756},{"@type":"BreadcrumbList","@id":"https:\/\/semiengineering.com\/hbm-shifts-testing-left-to-preserve-ai-chip-yield\/#breadcrumb","itemListElement":[{"@type":"ListItem","position":1,"name":"Home","item":"https:\/\/semiengineering.com\/"},{"@type":"ListItem","position":2,"name":"HBM Shifts Testing Left To Preserve AI Chip Yield"}]},{"@type":"WebSite","@id":"https:\/\/semiengineering.com\/#website","url":"https:\/\/semiengineering.com\/","name":"Semiconductor Engineering","description":"Deep Insights For Chip Engineers","potentialAction":[{"@type":"SearchAction","target":{"@type":"EntryPoint","urlTemplate":"https:\/\/semiengineering.com\/?s={search_term_string}"},"query-input":{"@type":"PropertyValueSpecification","valueRequired":true,"valueName":"search_term_string"}}],"inLanguage":"en-US"},{"@type":"Person","@id":"https:\/\/semiengineering.com\/#\/schema\/person\/09642fe6b88c48eed5f0273e74c05f98","name":"Anne Meixner","image":{"@type":"ImageObject","inLanguage":"en-US","@id":"https:\/\/secure.gravatar.com\/avatar\/5ef77e855d4b5df19e9d5a817f51e2e3c2bcc80cfba5c2285d2f4886ed94cd1a?s=96&d=mm&r=g","url":"https:\/\/secure.gravatar.com\/avatar\/5ef77e855d4b5df19e9d5a817f51e2e3c2bcc80cfba5c2285d2f4886ed94cd1a?s=96&d=mm&r=g","contentUrl":"https:\/\/secure.gravatar.com\/avatar\/5ef77e855d4b5df19e9d5a817f51e2e3c2bcc80cfba5c2285d2f4886ed94cd1a?s=96&d=mm&r=g","caption":"Anne Meixner"},"description":"Anne Meixner is a contributing editor at Semiconductor Engineering. She has 30+ years in the semiconductor industry. She became fascinated by defects in the semiconductor manufacturing process as a young engineer at IBM. Over that period, she has focused on test methodology with emphasis on mixed-signal and analog DFT and test. As a technical communicator she takes complex ideas and explains them in consumable understandable pieces. Meixner has worked at IBM, Carnegie Mellon University and Intel. She holds three U.S. patents and her peers have recognized her work with two best papers at IEEE International Test Conference. She founded The Engineers\u2019 Daughter in 2015 to consult on semiconductor testing and to coach engineers.","url":"https:\/\/semiengineering.com\/author\/anne-meixner\/"}]}},"jetpack_publicize_connections":[],"jetpack_featured_media_url":"https:\/\/i0.wp.com\/semiengineering.com\/wp-content\/uploads\/2026\/05\/Ann-5_11-Fig-3.jpg?fit=1066%2C756&ssl=1","jetpack_sharing_enabled":true,"jetpack_shortlink":"https:\/\/wp.me\/p9HsLC-1DRn6","_links":{"self":[{"href":"https:\/\/semiengineering.com\/wp-json\/wp\/v2\/posts\/24276292","targetHints":{"allow":["GET"]}}],"collection":[{"href":"https:\/\/semiengineering.com\/wp-json\/wp\/v2\/posts"}],"about":[{"href":"https:\/\/semiengineering.com\/wp-json\/wp\/v2\/types\/post"}],"author":[{"embeddable":true,"href":"https:\/\/semiengineering.com\/wp-json\/wp\/v2\/users\/789"}],"replies":[{"embeddable":true,"href":"https:\/\/semiengineering.com\/wp-json\/wp\/v2\/comments?post=24276292"}],"version-history":[{"count":14,"href":"https:\/\/semiengineering.com\/wp-json\/wp\/v2\/posts\/24276292\/revisions"}],"predecessor-version":[{"id":24276436,"href":"https:\/\/semiengineering.com\/wp-json\/wp\/v2\/posts\/24276292\/revisions\/24276436"}],"wp:featuredmedia":[{"embeddable":true,"href":"https:\/\/semiengineering.com\/wp-json\/wp\/v2\/media\/24276293"}],"wp:attachment":[{"href":"https:\/\/semiengineering.com\/wp-json\/wp\/v2\/media?parent=24276292"}],"wp:term":[{"taxonomy":"category","embeddable":true,"href":"https:\/\/semiengineering.com\/wp-json\/wp\/v2\/categories?post=24276292"},{"taxonomy":"post_tag","embeddable":true,"href":"https:\/\/semiengineering.com\/wp-json\/wp\/v2\/tags?post=24276292"}],"curies":[{"name":"wp","href":"https:\/\/api.w.org\/{rel}","templated":true}]}}