Sivers Semiconductors AB - THE SEMICONDUCTOR CHOKEPOINT PLAY NOBODY'S TALKING ABOUT
TL;DR
Sivers Semiconductors is literally a bottleneck in the AI infrastructure supply chain that nobody knows about. They're basically the only reliable source of fancy laser wafers (III-V epitaxial substrates) that power data center optics and LiDAR. [Rumours mention that they will be listing on NASDAQ soon.](https://www.sivers-semiconductors.com/press/sivers-semiconductors-ab-evaluates-a-potential-dual-listing-of-its-shares-on-the-nasdaq-new-york/) When institutional money figures out they're holding a golden goose, re-rating incoming. Currently trading like a boring Swedish industrial stock. This is not financial advice, but it's a solid thesis if you believe in the AI buildout.
# Why Should You Care?
Okay so imagine you're Microsoft, Google, or Broadcom. You need to build 400G/800G optical transceivers for data center interconnects. You call your laser supplier and they're like "yeah uh we can't get enough III-V epitaxial wafers from our foundry right now." Your entire production line shuts down. You lose billions in revenue. A lot of the time the foundry it's Sivers Semiconductors.
SIVE operates one of maybe a handful of commercial III-V epitaxial foundries on the planet. It's literally a critical chokepoint in the infrastructure for the whole AI boom. That's the whole point here.
# THE SUPPLY CHAIN (Follow the Money)
Here's how it flows:
Hyperscalers (Microsoft, Google) need transceivers from Lumentum and II-VI. Those guys need laser diodes and photodetectors. Nobody can make those without III-V wafers. Sivers supplies these wafer! If Sivers can't deliver, the whole thing backs up. And right now supply is completely crushed. Lead times are 2-3 years just to get new epitaxy equipment. Global capacity is underwater. That's the leverage.
# THE THESIS
Data center optics demand is going parabolic. We're talking 40%+ annual growth. The whole 400G to 800G to 1.6T infrastructure buildout is just getting started. Every hyperscaler is in an arms race trying to support AI training clusters. Sivers' customers literally cannot fulfill orders without more wafers.
LiDAR adoption is ramping harder than people think. Automotive LiDAR revenues are expected to ramp Q4 2026. Tier-1 partnerships incoming. And here's the thing—LiDAR systems are entirely dependent on III-V semiconductors. More EVs equals more LiDAR equals Sivers' business goes up.
NASDAQ listing. This is the big one. Right now it trades on Stockholm's small-cap exchange. Basically nobody knows about it. When it hits NASDAQ you get index inclusion, you get analyst coverage, you get generalist money waking up to the fact that epitaxial wafers actually matter. Institutional investors don't even know Sivers exists yet. Once they do the re-rating happens.
# THE NUMBERS (Why Valuation Looks Cheap Right Now)
Current market cap is around $312M. EV/Revenue sitting at 4.2x which sounds reasonable until you look at peers. Lumentum trades at 3.8x, II-VI at 2.1x. Peer average is 3.0x.
So if Sivers re-rates to peer multiples, you're looking at getting to $221M+ valuation. That's not crazy, these are the actual comparable companies.
# THE COMPETITIVE LANDSCAPE (Why Competitors Can't Just Copy)
IQE is dealing with financial issues and can't expand capacity. WIN Semiconductors is already maxed out with 12-18 month lead times. VPEC is over in Asia which becomes a vulnerability in today's supply chain environment.
Sivers has something none of them can replicate easily also and thats the fact that it's in Europe. In a world where US/China semiconductor tensions are real, customers actually need non-Asian suppliers for diversity. That matters more than people think.
And with 2-3 year lead times just to get new epitaxy equipment? You can't just spin up a new competitor overnight. This moat is real.
# WHAT TO WATCH FOR (Timeline)
Q4 2026 is when LiDAR revenue ramp actually kicks in. NASDAQ listing should happen before then which is huge. Tier-1 data center partnerships getting announced would be another catalyst. And as supply-demand imbalance widens, Sivers gets pricing power and margins expand. Those are your key signals.
# THE RISKS
Customer concentration is the big one. Top 5 customers probably represent 60%+ of revenue. If Broadcom or Lumentum decides to build their own III-V capacity, Sivers gets screwed. This is a real risk.
Silicon photonics could cannibalize some of the III-V wafer demand down the line.
Capacity expansion requires huge capex. If they mess up the execution, margins don't expand like the thesis assumes.
And if the broader market corrects, even quality chokepoints get dumped. Fundamentals still matter.
# HOW TO PLAY THIS
Buy thesis is pretty straightforward: NASDAQ listing brings institutional awareness to a supply chain bottleneck. That drives re-rating. 18-24 month time horizon minimum.
Target: if we get to peer valuation multiples at 3.0x EV/Revenue, that's roughly doubling from here. Upside is real if this thesis plays out.
# FINAL THOUGHTS
The entire semiconductor industry is obsessed with the sexy stuff. TSMC, Samsung, Intel, NVIDIA, AMD. Nobody talks about epitaxial foundries. That's actually the advantage here.
Sivers isn't cool. It's not NVIDIA making AI chips. It's some Swedish company doing III-V wafers that 99% of retail traders have never even heard of. But that lack of attention is the whole point.
When SIVE lists on NASDAQ, when Wall Street analysts finally write reports on it, when Broadcom mentions them in an earnings call—that's when institutional money wakes up. And when that happens, chokepoints in trillion-dollar infrastructure supply chains don't stay cheap.
# Positions
5000 shares @ 21 SEK each. Cost/Market value is converted to Singaporean $ below.
https://preview.redd.it/r5opf5t5unvg1.jpg?width=1170&format=pjpg&auto=webp&s=7dcd2a5af53ca3bdf324c94ca5255f87ae502c4c