AnswersPublished21 sources
How Micron Plans to Double HBM Capacity for Nvidia’s Vera Rubin Era
Micron reportedly plans to reach about 100,000 HBM wafer inputs per month by the end of 2026 by investing aggressively in equipment and ramping 12 layer HBM4. Micron has already begun volume shipments of a 36GB, 12 high HBM4 designed for Nvidia’s Vera Rubin platform; Taiwan, Singapore and Japan projects are intended...
Published byEdited with GPT-5.6 TerraImages generated with GPT Image 2
Research answer
Key takeaways
- Micron reportedly plans to reach about 100,000 HBM wafer inputs per month by the end of 2026 by investing aggressively in equipment and ramping 12 layer HBM4.
- Micron has already begun volume shipments of a 36GB, 12 high HBM4 designed for Nvidia’s Vera Rubin platform; Taiwan, Singapore and Japan projects are intended to support supply growth mainly from 2027 and beyond.
- The expansion strengthens Micron’s position as a meaningful alternative supplier in a tight HBM market, though SK hynix remains the market leader and Samsung’s HBM4 recovery is intensifying the competitive race.
AI Prompt
openai.comCreate a landscape editorial hero image for this Studio Global article: How is Micron Technology planning to more than double its HBM production capacity to about 100,000 wafers per month by the end of 2026—parti. Article summary: Micron is pursuing a two-track strategy: rapidly convert existing DRAM output and packaging capacity to HBM4 now, while funding new Asian manufacturing and packaging capacity that supports growth from 2027 onward. This s. Topic tags: general, general web, news, user generated. Style: premium digital editorial illustration, source-backed research mood, clean composition, high detail, modern web publication hero. Use reference image context only for broad subject, composition, and topical grounding; do not copy the exact image. Avoid: logos, brand marks, copyrighted characters, real person likenesses, fake screenshots, UI text, readable text, watermarks, charts w
Micron’s route to substantially more high-bandwidth memory (HBM) is not dependent on waiting for a brand-new fab to open. The near-term plan is to commit more equipment and manufacturing capacity to HBM—especially 12-layer HBM4—while its investments in Taiwan, Singapore and Japan build the wafer and advanced-packaging base for later years.
Industry reporting says Micron is targeting roughly 100,000 HBM wafer inputs per month by the end of 2026, after adding up to 60,000 wafer inputs per month. That would be more than double its estimated 2025 level of roughly 40,000–50,000 wafer inputs per month. These figures describe wafer-input capacity, not finished HBM stacks, so realized shipments will still depend on yields, die stacking and packaging throughput. 7
The immediate lever: ramp 12-layer HBM4
The strategic product is Micron’s 36GB, 12-high HBM4. Micron said it began volume shipments in the first quarter of calendar 2026 and that the part was designed for Nvidia’s Vera Rubin platform. The company lists bandwidth above 2.8 TB/s and more than 20% better power efficiency than its HBM3E product. 9
That matters because HBM is not simply conventional DRAM sold in greater volume. It requires leading-edge DRAM dies, through-silicon-via interconnects, stacking and advanced packaging, as well as qualification in an accelerator platform. Increasing nominal wafer starts helps only if those downstream steps can scale with adequate yield.
Micron has also shipped samples of a 48GB, 16-high HBM4 product, demonstrating a path to greater capacity per HBM placement. But samples are distinct from broad volume availability; the established volume-shipping product in the company’s announcement is the 36GB, 12-high device. 9
What the capacity target means—and what it does not
The reported plan is an aggressive reallocation and equipment-investment effort, not evidence that every new wafer will become immediately available to customers. The central execution tests are:
- HBM4 manufacturing yield: More wafer starts do not necessarily translate one-for-one into qualified stacks.
- Advanced-packaging throughput: HBM requires assembling and testing vertically stacked dies at scale.
- Customer qualification and allocation: A supplier must be validated for a specific accelerator platform before capacity becomes commercially useful.
Micron’s HBM supply for calendar 2026 has been reported as sold out, with some 2027 capacity under contract. Publicly available reporting is less clear on claims that all of 2027 is fully booked or that 2028 bookings are firm. Those longer-range assertions should therefore be treated as less certain than the reported 2026 sellout. 6
11
The Asia buildout supports the next phase
The 2026 HBM ramp is the short-term story. Micron’s physical expansion projects address a different problem: creating a larger and more resilient supply base after the current bottleneck.
Taiwan: a faster route to additional DRAM capacity
Micron completed its acquisition of PSMC’s Tongluo P5 site in Miaoli County in March 2026. The acquired site gives Micron an additional manufacturing location in Taiwan; the company has indicated that wafer starts are expected in 2027. 54
19
For HBM, added DRAM manufacturing capacity is strategically valuable because stacked HBM products consume substantial volumes of advanced DRAM die. Tongluo is therefore better understood as a future supply enabler than the primary explanation for the end-of-2026 capacity target.
Singapore: packaging is as important as wafers
Micron’s Singapore HBM advanced-packaging facility carries an approximate $7 billion investment. Operations were scheduled to begin in 2026, with meaningful expansion of Micron’s total advanced-packaging capacity expected in calendar 2027. 57
That timeline is crucial: packaging capacity is a gating step for HBM, so the Singapore project can help convert DRAM output into stacked, qualified memory products.
Micron is also building a separate advanced wafer fab in Singapore with a planned $24 billion investment over 10 years and up to 700,000 square feet of cleanroom space. That fab is located at Micron’s existing NAND complex, and wafer output is scheduled for the second half of 2028. It supports Micron’s longer-term manufacturing footprint, but it should not be counted as near-term HBM capacity. 51
Japan: advanced-memory capacity for the later horizon
Micron broke ground on a ¥1.5 trillion, or roughly $9.3 billion, expansion in Hiroshima. The facility is intended to make advanced memory including HBM-relevant products, with equipment delivery and installation scheduled to begin in the second half of 2028. 25
Like the Singapore wafer fab, the Hiroshima project is a longer-dated capacity investment rather than a contributor to the 2026 HBM4 ramp.
Where Micron stands against SK hynix and Samsung
Micron is moving from a smaller HBM participant toward a more consequential third supplier, particularly for next-generation AI accelerators. But capacity ambitions alone do not make it the market leader.
TrendForce described SK hynix as the leader in early 2026, with Samsung recovering through HBM3E and HBM4 and Micron expanding aggressively. 40 More recent reported revenue-share data for the second quarter put SK hynix at 50% and Samsung at 33%, underscoring Samsung’s sharp recovery.
35
The exact shares vary by source, quarter and measurement method—revenue, bits and wafer inputs are not interchangeable. The durable conclusion is clearer than any single percentage: SK hynix retains the scale and incumbent advantage, Samsung is rapidly rebuilding momentum in HBM4, and Micron is trying to secure a larger role through credible product qualification and a sharp capacity ramp.
Bottom line
Micron’s reported 100,000-wafer-per-month HBM target rests on accelerating existing manufacturing and packaging investment around 12-layer HBM4, not on new greenfield projects coming online in 2026. Its volume HBM4 shipments for Nvidia Vera Rubin provide the product and platform foundation; Taiwan, Singapore and Japan are intended to make that growth more durable from 2027 onward. 7
9
51
If Micron can sustain yields and advanced-packaging throughput, the plan would narrow its scale gap and strengthen its importance as an alternative AI-memory supplier. It does not, by itself, resolve the HBM shortage or overturn SK hynix’s leadership—especially as Samsung expands its own HBM4 presence.
Studio Global AI
Continue your research
This page includes a source-backed answer you can continue inside Studio Global.
People also ask
What is the short answer to "How Micron Plans to Double HBM Capacity for Nvidia’s Vera Rubin Era"?
Micron reportedly plans to reach about 100,000 HBM wafer inputs per month by the end of 2026 by investing aggressively in equipment and ramping 12 layer HBM4.
What are the key points to validate first?
Micron reportedly plans to reach about 100,000 HBM wafer inputs per month by the end of 2026 by investing aggressively in equipment and ramping 12 layer HBM4. Micron has already begun volume shipments of a 36GB, 12 high HBM4 designed for Nvidia’s Vera Rubin platform; Taiwan, Singapore and Japan projects are intended to support supply growth mainly from 2027 and beyond.
What should I do next in practice?
The expansion strengthens Micron’s position as a meaningful alternative supplier in a tight HBM market, though SK hynix remains the market leader and Samsung’s HBM4 recovery is intensifying the competitive race.
Sources
en.etnews.comUS Micron to Double HBM Capacity, Closing Gap with Samsung ...
investors.micron.comMicron in High-Volume Production of HBM4 Designed for ...
ad-hoc-news.deMicron's Fully Booked HBM Pipeline Meets a Market Still ...
seekingalpha.comMicron Technology: HBM Sold Out For 2026, Wall Street Is ...