Vik's Newsletter

Vik’s Newsletter

HBM: Lower Die-Count Stacks and Custom Base-Die Performance

Expert Interview conducted on November 6, 2026.

Vikram Sekar's avatar
Vikram Sekar
Nov 18, 2026
∙ Paid

This interview was conducted on the FUNDA platform under its X KOL Creator Program on November 6, 2026. The transcript has been screened for compliance by FUNDA’s team. The summary, key takeaways, and transcript reproduced below are presented verbatim from the reviewed document.

Editorial annotations, if added, are mine alone and will be clearly identified as “[Vik’s annotation]”. No annotations have been inserted into the compliance-reviewed material below. Annotation blocks are the author’s own commentary and fall outside FUNDA’s compliance review.

VIK’S ANNOTATION

Why did I conduct this interview?

Memory has seen the lion’s share of bearishness in the recent AI reversion. Even recently 13Fs filed by popular investors have shown trimming of positions in memory stocks. Regardless of short term sentiment, I want to understand where memory is headed long term.

There has been talks of de-specing HBM in next generation accelerators, at least in certain variants. Others have argued that increasing capacity alone is insufficent when there is limited bandwidth to access all that capacity.

To increase capacity from where we are today, we will inevitably need hybrid bonding. Base dies in HBM are getting increasingly custom, and the question is where does the value in HBM naturally flow in this case, who designs the base die, and where does memory makers’ moat lie in the future.

This interview was designed to answer as much of this as possible. Expert opinions help shape some understanding, but not all. I hope this helps you develop your perspectives as well. I will provide my own commentary alongside the expert’s for more context.


By reading this post, you agree to the terms and conditions. Also see the full ethics statement.

If you’re new, check out the About page. A lot of readers expense the subscription to this newsletter as it helps their professional work. Group subscriptions (3+) are 20% off. If you have any questions, reply to this email and let me know!

For the full post, upgrade to a paid subscription

Check out the Semi Doped podcast, and our daily free newsletter with latest semi news. For institutional research, contact me at SemiExponent.


Speakers

Expert: Former senior executive · Memory manufacturing

FUNDA Client: Vikram Sekar

Disclaimer

The expert interviewed for this meeting is an industry expert. All data cited herein, including but not limited to growth rates, customer scale, transaction volume, and regional breakdowns, as well as forward-looking projections, are personal estimates and recollections provided by the interviewee based on prior industry experience. They do not represent official disclosures by the relevant companies or their affiliates, nor do they constitute any material nonpublic information. These meeting notes are for reference only and do not constitute investment advice or a basis for business decisions. The views expressed herein represent only the personal opinions of the respective contributors at a specific point in time and do not represent the position of FUNDA. As an independently operated platform, FUNDA maintains its position as a neutral third party and makes no express or implied warranty regarding the accuracy, completeness, or applicability of the views expressed in these notes. Given the rapid changes in market conditions and technology trends, these notes reflect market conditions as of the period before the publication date, and the relevant information may become outdated over time. FUNDA has no obligation to update these notes in real time.

The content that follows is compliance reviewed. Highlighted annotations are mine.

This is not investment advice. Please do your own due diligence.

Summary

• Fixed package height increases die fragility: The industry expert stated that total HBM package height remains limited to 775 microns. As a result, average die thickness must decrease from approximately 50 microns for an 8-layer stack to 35 microns for a 12-layer stack, and to below 25 microns for a 16-layer stack. Thinner dies are more susceptible to cracking or breaking under the mechanical pressure of thermocompression bonding.

• Historically wide gap in bonding yields: According to the industry expert’s recollection, approximately a year and a half ago, the single-layer bonding yield of the TC-NCF process was in the mid-to-high range. Because this process must be repeated 8 times, cumulative losses significantly reduced the overall yield of an 8-layer stack. During the same period, the MR-MUF process was less sensitive to layer count because the primary bonding was not performed during the thermocompression step.

User's avatar

Continue reading this post for free, courtesy of Vikram Sekar.

Or purchase a paid subscription.
© 2026 Vikram Sekar · Privacy ∙ Terms ∙ Collection notice
Start your SubstackGet the app
Substack is the home for writing, podcasting, and media founders.