Is Samsung Foundry Dead? What the Obituaries Miss About Its 2nm Order Book

This article is for informational purposes only and is not financial advice. TheGatBull may earn a commission from some links at no cost to you — see our disclosure and full disclaimer.

Type “Samsung foundry” into any search bar and the autocomplete finishes the sentence for you: dead. The obituary writes itself — TSMC now controls 72.3% of the global foundry market and Samsung 6.5%, an 11-to-1 revenue gap that widened over the past year. Here’s the short answer to the question in the title: half right, half wrong. Samsung lost the market-share war, and that verdict is unlikely to reverse soon. But “dead” collides with an awkward set of facts: a foundry division that just booked the largest contract in its history — $16.5 billion from Tesla — has Nvidia queued up next, and is expected to post its first quarterly profit in roughly four years as early as Q3 2026. Defeat and death are different diagnoses. This article separates them with numbers. This is not financial advice.

🎩 Under the Gat — Before anything else, the one line US readers miss: you cannot buy “Samsung Foundry.” TSMC (TSM) is a pure-play foundry — its stock is the foundry. Samsung Foundry is one division buried inside Samsung Electronics (KRX: 005930), which doesn’t even have a US-listed ADR — a London GDR (SMSN) and OTC trading are the only routes. And what moves that stock right now isn’t foundry at all; it’s memory and HBM. What it means for investors: “the foundry is dead, so avoid Samsung” may simply be an argument about the wrong address — check the address before acting on the obituary.

The myth: “TSMC won, so Samsung Foundry is finished.”
The data: market share supports the myth — TSMC 72.3% vs Samsung 6.5% in Q1 2026, an 11x revenue gap (TrendForce). But the order book, yields, and profit line push back.
Where the myth is right: on scale, share, and leading-edge yield credibility, Samsung is a distant second — and the gap grew (7.7% a year ago, 6.5% now).
Where it’s wrong: “dead” is an exaggeration. 2nm (SF2) yields in the 55–60% range (reports conflict), Tesla’s $16.5B and Nvidia’s inference-chip orders, and a possible Q3 2026 quarterly profit all say “bottomed,” not “buried.”
For investors: market share is a lagging indicator. The inflection that matters for the stock is the foundry’s loss-to-profit swing — but pure exposure is impossible. Buying Samsung Electronics means buying memory, phones, and foundry as a bundle.
Timing: Samsung reports Q2 2026 preliminary earnings tomorrow, July 7. Consensus: operating profit around ₩85 trillion (~$56B), up 18–19x year over year — almost all of it memory. The foundry commentary is this debate’s next data point.

All ₩/$ conversions at USD/KRW ≈ 1,530 (July 3, 2026), unless dated otherwise.

The conventional wisdom: “TSMC won, so Samsung Foundry is dead”

The consensus on Reddit and X’s semiconductor threads is simple. TSMC has effectively monopolized leading-edge volume from Apple, Nvidia, and AMD. Samsung, the story goes, nearly had to outsource its own Exynos chips. Add the April 2026 report that Samsung’s 2nm yields sat around 55% — below the ~60% threshold usually needed for stable mass production — and that Qualcomm might take its next flagship to TSMC (TrendForce, citing Busan Ilbo), and the picture looks clean: the number two dies slowly.

The headlines lean the same way. A June 2026 piece built on TrendForce’s Q1 data (Sammy Fans) put it bluntly: “TSMC is now 11x bigger than Samsung’s chip foundry business.” Against numbers like that, arguing feels foolish.

Now look at what the data actually says — the share table and the order book disagree

Here’s the core of this myth-bust: if you stop at one table (market share), you’ve seen half the picture. Three sets of numbers have to sit side by side.

① Market share — the myth’s evidence

Q1 2026 foundry market: TSMC revenue ~$35.86B / 72.3% share (up from 70.4% the prior quarter) vs Samsung ~$3.2B / 6.5% (TrendForce).
Samsung’s share fell from 7.7% (~$2.89B) a year earlier. Revenue inched up; the market grew faster.
The gap isn’t closing — it’s widening. → The myth is right here.

② Yields — the rebuttal (with an honesty caveat: reports conflict)

Samsung’s 2nm (SF2) gate-all-around yields reportedly reached the 55–60% range in November 2025. Since then, a March 2026 report said yields “topped 60%” while an April 2026 report put them at ~55%, below the mass-production threshold. The direction is up; the level is disputed. The stated next target: 70%.
What is confirmed: Samsung started mass production of first-generation 2nm in Q4 2025, targets second-generation SF2P production in the second half of 2026, and has publicly guided to double-digit foundry revenue growth and improved profitability this year (4Q25 results).
The counterargument stands too: to win top-tier customers of the Qualcomm and Apple class, Samsung still trails TSMC’s mature 2nm yields, reportedly in the 60–70% range.

③ Orders and the profit line — the rebuttal’s strongest evidence

Tesla: $16.5 billion (about ₩22.8 trillion at announcement) through the end of 2033 — the largest contract in Samsung Foundry’s history. AI6 chips for Full Self-Driving, the Dojo supercomputer, and Optimus robots, built on 2nm-class process at the Taylor, Texas fab. Elon Musk called the figure “just the bare minimum,” and later reports say Tesla is moving to more than double AI6 volumes.
Nvidia: Jensen Huang confirmed at GTC 2026 that Samsung will manufacture an Nvidia AI inference chip, with revenue contribution expected from the second half of 2026. Samsung had already joined Nvidia’s NVLink Fusion ecosystem in late 2025.
2nm-related orders for 2026 were reported in January to be heading for 30%+ growth over 2025 (TrendForce); by June, Digitimes had raised that estimate to roughly 130%.
After roughly four years of losses, the foundry division may post a quarterly profit in Q3 2026. (One honest asterisk: foundry chief Han Jin-man says a full-year profit is realistic in 2028, not 2026. A quarter and a year are different claims.)

🎩 Under the Gat — The hands writing the obituary and the hands signing the purchase orders belong to different people. The market reads Samsung Foundry’s share table and delivers a eulogy; Tesla and Nvidia read the same industry and signed contracts. Neither is stupid. What they’re buying isn’t “the number one foundry” — it’s the most credible leading-edge alternative to TSMC, a second source their supply chains can’t live without. Geopolitics and supply-chain insurance set that price, not market share. What it means for investors: the order book prices Samsung’s strategic scarcity; the share table prices its competitive defeat. Both are real. Only one of them is new information.

Myth vs. reality, line by line

The myth (“it’s dead”)	The reality (the data)
TSMC took all the leading-edge volume	True. 72.3% vs 6.5% share in Q1 2026, an 11x revenue gap — and widening (TrendForce)
Samsung can’t get yields up	Partly true, dated. Yields recovered to the 55–60% range — but reports conflict, and the 60%+ stability bar and 70% target aren’t cleared yet
Big customers are abandoning Samsung	False. Tesla signed the largest deal in the foundry’s history; Nvidia is next. Qualcomm’s exit is a report, not a fact
The foundry is a bottomless money pit	Changing. After ~4 years of losses, a Q3 2026 quarterly profit is in sight — full-year profit targeted for 2028 (foundry chief)
Samsung Foundry isn’t worth buying	Wrong address. You can’t buy it as a pure play. It’s bundled inside Samsung Electronics — and that stock is currently a memory/HBM story

The Korea-specific point US readers miss: there is no pure play

Want pure foundry exposure? The TSM ADR is the direct route — a pure foundry, where the stock price is the foundry business. Samsung has no such ticker. Samsung Electronics (KRX: 005930) has no US-exchange ADR at all — a London GDR (SMSN) and OTC trading are the only offshore routes (we mapped every door for US investors here). And whatever you buy, you get the world’s largest memory business, plus smartphones, displays, and the foundry — one basket, no substitutions.

That’s why the force moving Samsung’s stock right now isn’t the foundry — it’s the memory supercycle. Tomorrow’s (July 7) Q2 preliminary print carries a consensus operating profit of roughly ₩85 trillion (~$56B), up 18–19x from a year ago, on DRAM and NAND prices that jumped around 60% during the quarter. That is an HBM and memory story, not a foundry story (we wrote about the memory thesis from the inside). In this stock, the foundry is close to a free option: if it swings to profit, that’s a bonus. If it doesn’t, memory carries the stock anyway.

🎩 Under the Gat — So what does this mean for investors? Avoiding Samsung Electronics because “the foundry is dead” is making a decision about the wrong division. The foundry is neither this stock’s main risk nor its main engine — the memory cycle is the body; the foundry turnaround is the tail. Don’t read the headline; read which division moves the P&L. — a view, not advice.

The verdict

Verdict: “dead” is an exaggeration. The precise statement is: “Samsung lost the leading-edge share war — and is turning the corner at the bottom.” Writing the foundry’s obituary off the share table is using a lagging indicator to forecast the future. But the opposite exaggeration — “Samsung is about to catch TSMC” — has no data behind it either; even Samsung’s own foundry chief dates a full-year profit to 2028. The truth sits in between: a defeated number two that may be about to print its first profitable quarter in four years. For the stock, the axis of the investment case is memory, not foundry — the foundry turnaround is an option that adds value only once confirmed.

Where each side can be wrong

Risk to the bull case: if 2nm yields keep missing the bar top-tier customers require, Qualcomm- and Apple-class volume stays at TSMC — April’s “55%, Qualcomm may walk” report is exactly that scenario. And even a Q3 profit leaves a long road to the 2028 full-year target; the timing is hostage to yields and fab utilization, quarter by quarter.
Risk to the bear case: betting on “dead” means avoiding Samsung Electronics — and missing both the memory/HBM upside and the free foundry option.
For both: Samsung Electronics moves far more on the memory cycle, FX, and HBM competition (SK Hynix, Micron) than on anything the foundry does (the chip-flows story we covered here). Share, yield, and order figures above are media-reported estimates verified as of July 6, 2026 — recheck against primary sources on your trade date.

FAQ

Can Samsung Foundry really catch up to TSMC?
On market share, the data says no — not any time soon (an 11x revenue gap, still widening). But “catching up” and “making money as the industry’s second source” are different goals. The second is in sight: a quarterly profit is possible in H2 2026, while Samsung’s own foundry chief points to 2028 for a full-year profit.

How do I invest in Samsung Foundry?
There is no pure-play ticker. Exposure comes only indirectly and bundled, through Samsung Electronics (KRX: 005930) — which has no US-listed ADR; you’d go through the London GDR (SMSN) or OTC. If you want pure foundry exposure, TSMC (TSM) is the direct route — a comparison, not a recommendation.

How big are the Tesla and Nvidia orders?
Tesla’s deal runs through the end of 2033 at $16.5B (~₩22.8T at announcement) — the foundry’s largest ever, and Musk called it “the bare minimum.” AI6 will be built at the Taylor, Texas fab on 2nm-class process. Nvidia’s inference chip is expected to contribute revenue from H2 2026. Large — but foundry revenue is still far smaller than memory inside Samsung’s total.

What should I watch in tomorrow’s (July 7) earnings?
Not the headline. The ~₩85T (~$56B) consensus operating profit, up 18–19x, is almost entirely memory and HBM. For this debate, the tell is the foundry division commentary — yields, utilization, and the timing of the profit swing.

Searched “South Korea foundry market” and landed here?
The essentials: at the leading edge, Korea’s foundry industry is effectively Samsung alone (plus legacy fabs like DB HiTek), and Samsung Foundry is not a listed company — it lives inside Samsung Electronics. TSMC dominates global share (72.3% vs 6.5%), but the 2026 story is the pivot: 2nm production, a US fab in Taylor, and record orders from Tesla and Nvidia.

Frequently Asked Questions

Can Samsung Foundry really catch up to TSMC?

On market share, the data says no — not any time soon. TSMC’s foundry revenue is roughly 11x Samsung’s, and the gap widened over the past year. But “catching up” and “turning profitable as the industry’s second source” are different goals. The second one is in sight: analysts expect a quarterly profit as early as Q3 2026, while Samsung’s own foundry chief points to 2028 for a full-year profit. This is not financial advice.

How do I invest in Samsung Foundry?

You can’t buy it directly. Samsung Foundry is a division inside Samsung Electronics (KRX: 005930), which has no US-listed ADR — only a London GDR and OTC trading. Buying Samsung means buying memory, smartphones, displays, and foundry as one bundle. For pure-play foundry exposure, TSMC (TSM) is the direct route. A comparison, not a recommendation.

How big are the Tesla and Nvidia orders for Samsung?

Tesla’s deal runs through the end of 2033 at $16.5 billion (about ₩22.8 trillion at announcement) — the largest contract in Samsung Foundry’s history, and Elon Musk called that figure “just the bare minimum.” The AI6 chip will be made on 2nm-class process at the Taylor, Texas fab. Nvidia’s AI inference chip is expected to start contributing revenue in the second half of 2026.

What should I watch in Samsung’s July 7 earnings?

Not the headline. The consensus operating profit of roughly ₩85 trillion (~$56B) — up 18-19x year over year — is almost entirely a memory and HBM story. For the foundry debate, watch the foundry division commentary: yields, fab utilization, and the timing of the profit turnaround.

What does the South Korean foundry market look like?

At the leading edge, it is effectively Samsung alone (plus legacy players like DB HiTek). Samsung Foundry is not a listed pure-play — it lives inside Samsung Electronics. TSMC dominates global share (72.3% vs 6.5%), but the 2026 story is Samsung’s pivot: 2nm production, a US fab in Taylor, Texas, and record-sized orders from Tesla and Nvidia.