In partnership with
New way to crush the market by trading in your sleep (from Millionaire Publishing)
Global semiconductor sales in 2025 reached a record high of $791.7 billion, a 25.6% increase over 2024. The AI memory supercycle isn't coming. It's already here.
But here's what most investors are missing.
The real story isn't Nvidia's GPU backlog or TSMC's latest node shrink. It's memory. Specifically, the quiet formation of a three-nation consortium that just took its first public steps toward ending the near-total monopoly that Samsung, SK Hynix, and Micron hold over AI memory chips.
And it has direct implications for every semiconductor ETF in your portfolio.
Z-Angle Memory
On February 3, 2026, at Intel Connection Japan, SoftBank subsidiary SAIMEMORY unveiled a new memory architecture called Z-Angle Memory (ZAM), the first credible challenger to HBM in years.
It's an AI-focused alternative to HBM for data centers, co-developed by Intel and SoftBank's SAIMEMORY.
Here's the deal: SoftBank founded SAIMEMORY in December 2024, specifically to develop AI-optimized memory.
Intel signed on as the primary technical partner, bringing decades of chip-stacking expertise and its Embedded Multi-Die Interconnect Bridge (EMIB) technology. And now, Taiwan's PSMC (Powerchip Semiconductor Manufacturing Corp.) has officially joined as the manufacturing partner, alongside Japan's Shinko Electric Industries, to handle pilot production and eventual mass manufacturing.
This is a clearly defined three-player structure:
Saimemory (SoftBank): IP management, design, and capital. SoftBank has also invested $2 billion directly in Intel, which makes this partnership deeply interlinked commercially.
Intel (INTC): Memory architecture and stacking know-how. Its first re-entry into memory since the 1980s.
PSMC: Trial production and manufacturing. Taiwan's foothold in AI memory, something the island has lacked despite dominating logic chip production through TSMC.
The global semiconductor industry posted its highest-ever annual sales in 2025, nearly hitting $800 billion, and global sales in 2026 are projected to reach roughly $1 trillion.
Why ZAM Matters
Early projections are ambitious. According to reports, ZAM targets:
2–3× the capacity density of current HBM3E
Up to 512 GB per chip (vs. ~192 GB for current HBM stacks)
40–50% lower power consumption
~40% lower production cost than equivalent HBM
The architecture differs from traditional HBM, rather than purely vertical die stacking, ZAM uses a Z-Angle approach that Intel claims improves heat dissipation and interconnect density simultaneously. Intel's EMIB packaging tech reduces latency at the die interfaces.
The full-scale commercial rollout target is 2029, with working prototypes expected by fiscal year 2027–2028. Intel confirmed this timeline in a blog post following the Saimemory announcement.
HBM Market
Right now, the HBM market is effectively controlled by three players: SK Hynix (~50%), Samsung (~30%), and Micron (~20%). SK Hynix is Nvidia's primary HBM supplier, which means Nvidia's AI accelerator roadmap is tightly coupled to one Korean memory giant.
That's a supply-chain concentration risk that hyperscalers and AI chip designers want to reduce. A viable ZAM standard, backed by Intel's architecture and PSMC's manufacturing capacity, could give Nvidia and other AI chipmakers a second source and potentially a better-performing one.
That's not a small development. Samsung and SK Hynix share prices already moved on the announcement, Intel surged 5% in overnight trading after the initial Intel-SoftBank partnership news broke, and SoftBank climbed 3.13% in the same session.
How seriously do you take ZAM as an HBM challenger by 2029?
What This Means for Geopolitics
The US-Japan-Taiwan formation here isn't accidental. This is "friendshoring" in semiconductors, a democratic-ally tech bloc explicitly building AI infrastructure supply chains that don't depend on either China or South Korean memory giants. It fits squarely within the broader CHIPS Act logic and the broader allied tech bloc that's emerged post-2022.
For ETF investors, the direct exposure map looks like this:
Intel $INTC: Holds 5.0–5.6% weight across the major semiconductor ETFs (SMH, SOXX, SOXQ). Intel has been a laggard for years. ZAM is the kind of structural catalyst that changes a stock's narrative, not overnight, but over a multi-year period. Watch $INTC weight carefully.
Micron $MU: At 6.3–6.4% weight in SMH and SOXX, Micron is the US incumbent in memory. ZAM doesn't directly help Micron, but it does validate the AI memory investment thesis broadly. And if ZAM enables a larger market for AI memory, Micron benefits from the secular demand growth even if ZAM captures share from Korean players.
How much does the US-Japan-Taiwan "friendshoring" angle affect your semiconductor thesis?
Meet America’s Newest $1B Unicorn
A US startup just hit a $1 billion private valuation, joining billion-dollar private companies like SpaceX, OpenAI, and ByteDance. Unlike those other unicorns, you can invest.
Over 40,000 people already have. So have industry giants like General Motors and POSCO.
Why all the interest? EnergyX’s patented tech can recover up to 3X more lithium than traditional methods. That's a big deal, as demand for lithium is expected to 5X current production levels by 2040. Today, they’re moving toward commercial production, tapping into 100,000+ acres of lithium deposits in Chile, a potential $1.1B annual revenue opportunity at projected market prices.
Right now, you can invest at this pivotal growth stage for $11/share. But only through February 26. Become an early-stage EnergyX shareholder before the deadline.
This is a paid advertisement for EnergyX Regulation A offering. Please read the offering circular at invest.energyx.com. Under Regulation A, a company may change its share price by up to 20% without requalifying the offering with the Securities and Exchange Commission.
Bottom Line
ZAM is not going to disrupt HBM next quarter. But the formation of a credible US-Japan-Taiwan memory consortium, backed by Intel's architecture, SoftBank's capital, and PSMC's manufacturing is the most significant structural development in AI memory in years.
The AI memory supercycle is real. Global semi sales confirm it. The question now is who captures the next layer of value — and whether a new memory standard can break the three-player oligopoly that's bottlenecked AI infrastructure for the better part of a decade.
Watch the 2027 prototype milestone closely. That's when ZAM goes from vision to evidence.
For now, the semiconductor ETFs with meaningful INTC and MU exposure, SMH, SOXX, and SOXQ, are the most direct ways to access this theme through a diversified vehicle.
When investing in semiconductor themes like AI memory, which ETF approach do you prefer?
Subscribe to ETF Alert for real-time market news.
We track the trends that move billions, before they hit mainstream headlines.
What's catching investor attention today: Trump’s Tariffs. Gold at $5,000. Tech Down 20%. The Rotation Is Real. Most Investors Are Still Sleeping On It
Disclaimer: This is not financial or investment advice. Do your own research and consult a qualified financial advisor before investing.