Contents
← Research

The AI boom hangs on three companies. One of them sits in Oberkochen.

TSMC makes 92% of all , ASML holds a 100% monopoly. And the mirror optics for that come out of Oberkochen: ZEISS ships exclusively to ASML. One region in Taiwan, one machine-builder in the Netherlands, one optics specialist in Baden-Württemberg. Plan B doesn't exist. If TSMC goes down, the global economy loses $5 to $10.6 trillion in the first year[1], twice as hard as COVID. The system holds anyway, because the US, China, and Taiwan have each other by the throat financially.

A plausible scenario · , not my speculation

You make coffee, scroll the news. Reuters reports that the Chinese navy has expanded the exercise zone in the Taiwan Strait. Quarantine inspections for cargo ships. No declaration of war, just "routine measures".

NVDA opens down 22%. Apple down 14. Lloyd's of London suspends war-risk premiums for Taiwan routes, no insurers, no shippers.

you get an email from Apple: your iPhone 18 pre-order is delayed. Expected delivery "Q2 2028".

Anthropic triples Claude Max prices. "Response to the global GPU shortage." OpenAI follows on Monday.

your portfolio shows minus 38% in the semiconductor cluster. You didn't know your S&P 500 ETF sits roughly 18% in NVIDIA, Apple, and Broadcom[35].

"We're screwed — if TSMC stops."

— my when I started this research. The next three acts test it.

Act 1 · The dependency

Why does so much hang on this one company?

It isn't one company. Semiconductors is four interlocked playing fields: , , , . At each level one dominant player, no plan B. Four single-source nodes wired in series: ZEISS, ASML, TSMC, Hsinchu. Each node physically depends on the previous one.

ZEISS SMT ASML TSMC NVIDIA Intel
What they do Supplies the optics for EUV machinesMirrors and lenses, exclusive to ASML Builds the lithography machinesEUV and lithography equipment Makes chips for other companiesPure-play , never designs its own Designs AI and gaming chips, outsources fab, no own Designs AND manufactures itself, currently in a painful pivot
What makes them special? 50-picometer polish accuracy on a 30 cm mirror.If the mirror were the size of Germany, no bump would exceed 0.1 mm[△]. Nobody else does this. Six top-tier technologies in parallel in one machine, 30+ years of R&D moat.Plasma 40× hotter than the sun + ZEISS mirror optics + 100,000-sensor vacuum + nm-precise wafer stage. Anyone trying to copy this has to reinvent 5 of 6 components at HVM quality — patents alone aren't enough. at N3 ~80%, Samsung ~50%.Apple, NVIDIA, AMD, and Qualcomm fab their leading-edge chips here. Yield gaps in semiconductor history never close in under 5 to 7 years. Blackwell GPU plus 18 years of CUDA software moat.On hardware AMD MI300 and Intel Gaudi sit roughly level, but fail on CUDA migration. ML code can't be rewritten in 12 months. AI training share stays > 80%. Currently nothing.18A node (~2nm) in risk production since early 2026, yield 10–15%, external foundry customers < 1%. Success possible, but today Intel sits in this row with no card to play.
Business model Exclusive supplier; ASML holds 24.9% stake, joint R&D roadmap Equipment monopoly, long order cycles, high margins Pure-play foundry, high margins via leading-edge premium Fabless, margins from IP plus CUDA software lock-in IDM mix; foundry arm as the second leg
Leading-edge market share ~100% EUV optics globally ~100% EUV machines ~70% total foundry ~92% at ≤ 5nm logic > 80% AI training chips < 1% external foundry, no leading-edge for third parties
2025 revenue ZEISS group ~€11B SMT segment ~€3B estimated €32.7B[12] $122.5B $215.9B FY26 (Feb 2025 — Jan 2026) $52.9B weakest year since 2010
Top customers ASML (exclusive for EUV) TSMC, Samsung, Intel, SK hynix, Micron NVIDIA (19%), Apple (17%), AMD, Qualcomm, Broadcom, MediaTek Microsoft, Meta, Amazon, Google, Oracle, xAI Own products plus roughly 12 external pilot customers
HQ location Oberkochen, Germany Veldhoven, Netherlands Hsinchu, Taiwan Santa Clara, USA Santa Clara, USA
← scroll horizontally →
Supply chain network · Tier 1 to Tier 4

Four tiers, four nodes — if one tips, the chain breaks at that level.

Supply / Material Sanctioned
Four levels, no alternatives. Arrows show who supplies whom; the percentages show the receiver's dependency ("NVIDIA · 100% Blackwell" = all Blackwell leading-edge chips at TSMC). The red X on the SMIC arrow is the sanctions block (ASML hasn't been allowed to ship EUV machines to China since 2019, only older DUV). Each node has no real alternative in its area. No other player can replace ZEISS mirrors, ASML EUV, or TSMC 3nm fab in the short term. If one node tips, the chain breaks at that level. That's what turns this concentration into a risk architecture.
Chart 1 · Foundry market share Q4 2025
TSMC makes 70% of all foundry chips — and 92% of the leading-edge ones. The next four combined sit below 20%.
View 1 · Leading edge — chips up to 5 nanometers

This is where the logic chips for AI, new iPhones, high-end smartphones, and server CPUs sit. Whoever holds this market makes the chips NVIDIA Blackwell, Apple M, and Qualcomm Snapdragon depend on.

TSMC up to 5nm logic
~92%
Leading edge
Samsung 3nm residual capacity, mostly internal
~5%
Leading edge
Intel + Others Pilot volume, memory-logic mix
~3%
Leading edge
Source: TrendForce Q4 2025, Taipei Times 2026-03-14. TSMC officially says "roughly 90%" at ≤ 5nm; for pure AI training chips (Blackwell, MI300, TPU) it's nearly 100% TSMC.
DeeperView 2 · Total foundry market (all nodes combined, not just leading edge)

Add up all contract fabs — leading-edge chips, older automotive chips, memory, microcontrollers. TSMC still dominates, but the gap is smaller than at the top. The second tier becomes visible.

TSMC all nodes
70.4%
Q4 2025
Samsung Foundry
7.2%
Q4 2025
SMIC China
5.3%
Q4 2025
UMC
4.4%
Q4 2025
GlobalFoundries
3.9%
Q4 2025
Intel Foundry external customers
< 1%
Q4 2025

Three edges that make TSMC the single point of failure.

The table explains the playing field. It doesn't explain why TSMC is so far out front and why this position isn't an accident. Three hard edges show it.

DeeperWhat's actually inside a modern chip? — walked through with the Apple M3 Max

Take the chip in a current MacBook Pro (the M4/M5 successor is built very similarly). Five facts make clear what kind of scale TSMC is actually working at. The Apple M3 Max is used because the numbers are public[38] — NVIDIA Blackwell is even bigger.

Size
92B
Transistors on a single chip. A transistor is a tiny electrical switch that can be "on" or "off". 92 billion of them — roughly 11× the world's population[△].
Feature size
3nm
Roughly 15 silicon atoms wide[△]. A water molecule is 0.3 nm — you're exposing structures only ten times that. This is where physics stops working in the classical sense.
Density
~ 200M / mm²
Transistors per square millimeter. A pinhead holds more transistors than Germany has residents[△].
Speed
~ 5 GHz
Switching cycles per second — per transistor. Each one flips up to five billion times per second. All of it synchronous, otherwise you get data salad.
Accuracy
99.9999999%[△]
Hit rate per transistor — and that would still barely be enough. At 92B transistors that's still 92 broken switches per chip. One in the wrong place makes the whole chip useless.

This isn't "computers getting faster". This is hardware at the scale of subatomic structures, running synchronously at femtosecond level — and it has to come out right on the first try. That's exactly why there's only one place on the planet that pulls this off in stable volume.

Edge 01 · YieldAt 50% yield only every second chip comes out fully functional. The rest fail, get sold as a cheaper variant, or end up as waste. Samsung doesn't hit this consistently, TSMC sits well above 80%.

TSMC 3nm
~ 90%
Share of working chips per
Samsung 3nm yield
~ 50%
From the same wafer you get half as much
Roughly 40 percentage points gap[3]. At the next generation (2nm) the picture stays: TSMC above 90%, Samsung 55-60%. Yield gaps in semiconductor history never close in under 5-7 years. Samsung's 3nm is now used almost exclusively for internal chips. Qualcomm, Google, and NVIDIA Blackwell all switched to TSMC.

Edge 02 · Customer dependencyApple and NVIDIA basically have no plan B. If TSMC tips, two whole industries tip with it.

Apple's chips at TSMC[19]
~ 85%
All A and M chips fabbed at TSMC
NVIDIA's leading-edge chips at TSMC
100%
Blackwell, Hopper, Rubin — no backup
Both sides locked together. Apple and NVIDIA together account for 36% of TSMC revenue (Apple 17%, NVIDIA 19%)[11]. NVIDIA has also reserved over 60% of global packaging capacity[13].

Could Apple or NVIDIA fab elsewhere? In theory yes, in practice not within 12 months:

  • Chip designs are TSMC-specific. Redesign, re-verification, and re-tape-out at Samsung 3nm cost 3 to 9 months and hundreds of millions of dollars.
  • Samsung yields are roughly half. Unit price rises accordingly.
  • Worst-case fallback: stay on TSMC Arizona at 4 to 5nm. One or two node generations behind. Performance, battery life, and AI capability take a multi-year hit.
Bonus risk · click to expand is the third node in the chain

TSMC makes the logic chips, ASML supplies the lithography — but modern AI chips need a third critical component: . An NVIDIA Blackwell without HBM is an expensive hotplate.

  • SK hynix dominates with roughly 50% market share, Samsung follows at ~35%, Micron just under 15%[16]. All three sit in Asia (South Korea, US plants only for standard DRAM).
  • HBM3E and HBM4 are fully booked. NVIDIA has reserved SK hynix' capacity for Blackwell and Rubin through 2026. No slot, no AI GPU — no matter how many wafers TSMC could produce.
  • HBM stacks are placed on the logic die via CoWoS packaging. If CoWoS goes down, HBM use goes down. If HBM goes down, all of Blackwell production is dead, even if every other station runs.

So it isn't a pure three-node chain (TSMC, ASML, SK hynix) — any single node can kill the GPU pipeline. But for the main thesis (TSMC + ASML as the single point of failure), HBM is the extra insurance question, not the core.

DeeperWho could replace TSMC? Three candidates — none of them by 2028.

If TSMC is the single point of failure, the second source should sit somewhere. Three candidates qualify in theory: Intel (US, IDM pivot), Samsung Foundry (South Korea), and China's SMIC. For all three, a glance at yield, trust, and equipment is enough to see: none of them replaces TSMC. And while they catch up, TSMC itself is rolling out the next generation.

Three candidates · none of them a real alternative

No candidate replaces TSMC — not by 2028, not by 2030.

Candidate 1
Intel
Santa Clara · USA
  • Yield gap: 18A at roughly 10-15% yield (industry estimate, end of 2025). Intel counters with defect density < 0.4 defects/cm², but the gap to TSMC N3 stays substantial.
  • Financial damage: Foundry Q4 2025 with $2.5B loss. Group 2025: $52.9B revenue, the weakest year since 2010. Intel Magdeburg first only delayed in September 2024, then cancelled outright in July 2025 (€30B investment, of which €10B was state subsidy)[31].
  • IDM trust problem: Apple, NVIDIA, and AMD don't hand Intel critical IP — Intel is simultaneously their competitor in PC and server CPUs.
  • Time to parity: 3-5 years and $30-50B more, group strategy under Lip-Bu Tan since March 2025 focused on 14A.
Candidate 2
Samsung Foundry
Hwaseong · South Korea
  • Yield gap: 3nm yield around 50% versus TSMC ~90% (TrendForce May 2025[3]). 40 percentage points behind at the decisive generation.
  • Market share: 7.2% foundry total, single-digit at leading-edge ≤ 3nm. Tesla mega-deal $16.5B for AI6 chips through 2033 (Samsung makes AI6, TSMC makes AI5) stabilizes utilization[30].
  • Captive bias: 3nm is now used almost only for Samsung's own chips. External customers (Qualcomm, Google, NVIDIA Blackwell) switched to TSMC.
  • Time to parity: 2-3 years in theory, but unlikely — trust and yield are missing at the same time.
Candidate 3
China / SMIC
Shanghai · China
  • Yield gap: 5nm yield around 33%, 5nm wafers ~50% more expensive than at TSMC.
  • Equipment ban: No EUV machines due to US export controls. DUV works, but is very inefficient.
  • Own equipment build: SMEE EUV prototype shown in 2025, earliest 2028, realistically 2030.
  • Time to parity: 5-10 years — and politically not available to Apple/NVIDIA.
No candidate replaces TSMC — not by 2028, not by 2030. Intel and Samsung fail on trust (IDM conflict with customers) and yield, China on the equipment ban. And while the three catch up, TSMC itself pushes out the next generation (2nm from 2026, A14 from 2028). The gap stays structural. A world without TSMC is a world with substantially fewer leading-edge chips, for a stretch nobody serious wants to put a number on — the range in the literature runs from 10 to 15+ years[20].

What makes ASML special, and why this pillar is even tighter

If you want to understand TSMC, you have to understand ASML. Without ASML machines, TSMC can't fab leading-edge chips. And ASML only exists once. The core in 3 sentences, then the numbers.

Short explainer: what is EUV anyway?

EUV stands for "Extreme Ultraviolet": light with a 13.5-nanometer wavelength, because normal UV light (193 nm) is too "thick" to draw structures below 7 nanometers. Producing EUV light is absurdly hard: a CO2 laser fires 50,000 times per second at tiny tin droplets and ignites them into a plasma roughly 40 times hotter than the sun's surface. From that plasma, the light beam is steered through specialty mirrors (all patented ZEISS glass, polished to atomic precision) onto the chip wafer. Without it, you can't make chips below 7 nanometers.

ASML TWINSCAN EXE:5200B — a High-NA EUV lithography machine, cut out on white background
This is an EUV machine. One.

A locomotive in a cleanroom — for $200 million.[12]

  • Size of a locomotive, ~180 tons (NXE:3600D, current volume generation). The High-NA successor EXE:5200B in the image weighs from ~150 tons.
  • Over 100,000 individual parts. 250 engineers take about 6 months to build a single machine.
  • Shipping: 3 cargo Boeing 747s plus several trucks per NXE machine. For the High-NA EXE: one cargo plane plus trucks for about 250 separate crates.
  • ~€170M per NXE machine. The High-NA successor (EXE:5200B in the image): ~€350M.
  • 40 to 60 EUV machines per year is what ASML builds globally. Not per month. Per year.
©ASML — TWINSCAN EXE:5200B (High-NA EUV)
+ The machine, opened up What actually happens inside an EUV machine — in six modules

Each of these modules is top-tier technology on its own. Only all six together make a machine that can fab structures below 5 nanometers. Anyone trying to copy ASML has to copy all six at once.

  1. 01

    Light source

    Cymer · USA · ASML subsidiary since 2013

    A CO2 laser fires 50,000 times per second at tiny tin droplets. The droplets vaporize into a plasma — roughly 40 times hotter than the sun's surface. From that plasma comes the EUV beam at 13.5 nm wavelength.

  2. 02

    Collector mirror

    ZEISS SMT · Germany

    The very first mirror behind the light source — it catches the EUV plasma light and focuses it into a usable beam. Coated with ~50 layers of molybdenum/silicon, each only a few atoms thick. Otherwise the mirror absorbs the EUV light instead of reflecting it.

  3. 03

    Reticle / photomask

    Photronics, DNP · Japan · specialized mask shops

    The "print template" for the chip. A quartz plate carries the circuit pattern, four times larger than on the finished chip because the projection optics shrink the image. A single mask costs $250,000 to $500,000 and only works for one chip design.

  4. 04

    Projection optics

    ZEISS SMT · Germany

    A system of 6 to 11 specialty mirrors that projects the reticle pattern onto the wafer and shrinks it 4×. Each mirror is polished to 0.1 nanometers of unevenness — scaled to the area of Germany, that's a maximum bump below 0.1 millimeters.

  5. 05

    Wafer stage

    ASML in-house · NL · plus VDL ETG (NL)

    Moves the 300mm wafer in tracks under the light beam — up to 150 mm per second, with nanometer precision. The machine remeasures tens of thousands of times per second, corrects vibrations, compensates heat drift. Even light truck vibrations outside the building would be an exposure disaster.

  6. 06

    Vacuum + control

    ASML · ~44,000 employees, ~100,000 individual parts

    EUV light is absorbed by air immediately. The entire optical path runs in high vacuum (~10⁻⁹ mbar — a billion times better than atmospheric pressure). More than 100,000 sensors regulate beam position, temperature, and mechanics in real time.

Six top-tier technologies in parallel. Patents aren't enough — the light source has to actually hold 50,000 plasma pulses per second stable, the mirrors have to actually be polished at atomic precision. If a single one of the six modules sits two classes below ASML level, there's no usable chip. That's the real moat.

Bottom line ASML + ZEISS ASML holds 100% EUV monopoly globally, ZEISS holds 100% of EUV optics exclusive to ASML — the two nodes before TSMC. If ZEISS breaks, ASML stops. If ASML stops, TSMC can't fab. The chain runs in one direction only, and it has no reserves.

DeeperWill China catch ASML? Bulls vs. bears, 5 to 5.

The EUV monopoly is the most provocative point in the whole thesis. If China or a consortium catches ASML, the single point of failure disappears. That's the optimistic read. The pessimistic one: nobody gets there in any foreseeable window. Both sides have hard arguments. The honest 5 against 5 below.

Bulls
Why it'll happen soon: China is catching up
China Big Fund III: $47.5B in fresh capital
Registered May 2024, operational from late December 2024[15]. Together with Fund I ($18.9B) and Fund II ($27.8B), roughly $94B cumulative in direct state semiconductor capital.
SMEE EUV prototype built in 2025, report December 2025
Shanghai Micro Electronics Equipment Corp. built a prototype with partly reverse-engineered ASML components plus its own LDP light source (TrendForce, December 2025[5]). HVM target: 2028 optimistic, 2030 realistic.
Canon Nanoimprint as a specialty solution — not an EUV replacement
Canon has worked on Nanoimprint Lithography (NIL) for over 20 years: roughly 40% of EUV cost, 10% of power draw. But only for memory and specialty markets, not for the logic chips NVIDIA and Apple need. Complement, not substitute.
Reverse engineering possible in theory
China has old ASML parts, talent poaching out of Taiwan, and the 14,000 ASML patents aren't perfect protection because they expire or can be designed around.
Sanctions push China's self-sufficiency
"Made in China 2025" aimed for 70% domestic share. Reality early 2026: ~35% on semi equipment. But the catch-up pace is higher than 2014-2020.
Bears
Why not: the moat is deeper than it looks
14,000 ASML patents + 2,000 ZEISS patents
The ASML number is an industry estimate, ZEISS officially confirms "over 2,000 patents" in EUV optics. Patent expiration would take decades.
ASML CEO Fouquet himself: China 10-15 years behind
This isn't China-bashing, it's the self-assessment of the market leader, who knows exactly how hard it is (Tom's Hardware, December 2024[4]).
Cymer light source not reproducible
LPP process with 50,000 tin droplets per second plus CO2 laser amplifier chain over 100 meters. ASML bought Cymer in 2013. Building your own light source takes decades.
Veldhoven talent pool can't be copied
ASML has over 44,000 employees: the largest concentration of plasma physics, high-power lasers, optics metrology, and vacuum mechatronics anywhere. This geography only exists once.
SMIC at 5nm: ~33% yield, ~50% more expensive than TSMC
5nm wafers at SMIC are currently up to 50% more expensive than at TSMC. Until HVM level it's many years: earliest 2028, realistically 2030+.
China itself depends heavily on TSMC — catch-up pressure is dampened
Chinese tech champions (Huawei, Alibaba, ByteDance) still buy a large part of their logic chips through TSMC channels (directly before sanctions, through middlemen after). That makes China a catcher-up geopolitically and a major customer economically — anyone blowing up the system ruins their own top-tier economy with it.
Chart 2 · Catch-up time
Even the fastest catcher-up (Samsung) takes longer than a US election cycle. Rebuilding Hsinchu: a whole generation.
Samsung reaches TSMC yield at N3 needs: process optimization + trust back
2-3 y
Intel reaches TSMC yield at N3 needs: $30-50B + customer trust
3-5 y
China SMIC fabs 3nm chips without EUV needs: DUV multipatterning breakthrough
5-10 y
China builds its own HVM EUV machine needs: light source + optics + patent workaround
10-15 y
Hsinchu cluster rebuilt somewhere else needs: talent, suppliers, trust, a generation
25-30 y
Scale: years to successful volume operation (). Sources: ASML CEO Christophe Fouquet (Tom's Hardware, December 2024[4]), TrendForce 2025-12[5], SemiAnalysis[13]. Range values are estimation spans, not point precision.
Act 2 · The risk

How likely is the break — really?

The median is the gray zone that's already running. Full war is the tail risk: low probability of occurrence, but if it hits, twice as hard as COVID. Three escalation scenarios are plausible, with a probability tendency over the 5-year window — deliberately as tendency, not point estimate. Serious sources avoid point estimates for exactly this reason (CFR 2026 puts a Cross-Strait crisis at >50%[8] without narrowing it further; CSIS wargames[6] stress that the hit rate depends on the behavior of individual actors).

Scenario A
5-yr probability · medium

Blockade / quarantine

Setup China seals off sea and air space without a formal declaration of war. Customs inspections, "exercises", cyber alongside.
Immediate impact Global GDP -5%. Taiwan -12.2%, China -8.9% (Bloomberg Economics)[1]. Chip exports stop because insurers and shippers no longer cover routes.
Recovery Uncertain. Even if the blockade ends: insurance premiums stay elevated for years, investors pull out.
For
  • China doesn't need a win, just pressure — escalatable and de-escalatable in steps.
  • Wargame history: across the 26 CSIS blockade games, escalation paths emerge repeatedly that are hard to contain[23].
  • 2024-2025 the frequency of ADIZ exercises has risen in real terms, the escalation staircase is in place.
Against
  • Lloyd's war cover collapses immediately — China exports get caught in the same blast.
  • US response within 72 hours highly likely (CSIS assumption).
  • Without ASML equipment, China itself has no leading-edge chip supply.
Scenario B — worst case
5-yr probability · low

Full invasion

Setup Amphibious landing across the Taiwan Strait. PLA navy plus missile prep, US/Japan reaction within 72 hours.
Immediate impact Global GDP -9.6%, $10.6 trillion in damage. US -6.7%, China -16.7% (Bloomberg Economics)[1]. Larger than the COVID shock and the 2008 financial crisis combined.
Recovery TSMC fabs are very likely destroyed or the EUV machines rendered unusable (ASML and TSMC have officially confirmed this happens in a conflict). Recovery of the leading-edge nodes: 10 to 15+ years — fab construction plus talent multiplier. No think tank publishes a hard number; the range combines build timelines and Hsinchu talent loss.
For
  • Burns 2023: Xi has ordered the PLA to be invasion-ready by 2027[32].
  • CCP legitimacy is tied to "reunification" — domestic pressure rises.
  • Stochastic escalation out of a naval incident is possible (see Hainan 2001).
Against
  • 83% of China experts consider invasion before 2030 unlikely; Pentagon: no fixed timeline order to Xi[9].
  • Mark Liu (TSMC): "Nobody can control TSMC by force" — the fab is dead, the prize is zero.
  • CSIS wargames: every scenario ends with Chinese losses above the threshold Xi can absorb.
Scenario C — already running
5-yr probability · high

Gray zone / status quo erosion

Setup Continuous activity: submarine cable sabotage, cyber attacks, military ADIZ drills, economic pressure. Taiwan's MODA reports 7 submarine cable incidents in 2025; the 4-year average is 7-8 per year[33].
Immediate impact Hard to measure in GDP, but the investment climate erodes. February 2023 Matsu Islands: 13,000 residents roughly 50 days without internet after Chinese ships cut both submarine cables[34].
Recovery Not applicable, ongoing.
For
  • Low-risk for China, nearly unattributable (submarine cable incidents are hard to prove).
  • Erodes Taiwan's investment climate without crossing a military threshold.
  • Escalation staircase stays open — can be stepped up to blockade.
Against
  • Immediate effect too weak for CCP domestic politics ("nothing achieved").
  • Pushes the US and Japan into a harder defense alliance with Taiwan.
  • Accelerates TSMC diversification (Arizona, Kumamoto, Dresden) — Silicon Shield erodes.
"Nobody can control TSMC by force. If you take a military force or invasion, you will render TSMC factory not operable, because this is such a sophisticated manufacturing facility." — Mark Liu, then TSMC chairman, CNN Fareed Zakaria GPS, July 31, 2022[10]
DeeperOther risks: earthquakes, cyber, IP theft — relevant, but not dominant.

Earthquake: April 3, 2024, magnitude 7.4 (USGS[36]; Taiwan CWA 7.2), the strongest quake in 25 years. TSMC: 70% tool recovery in 10 hours, EUV undamaged, full recovery after 3 days, Q2 margin hit only -50 basis points. TSMC has more experience with serious earthquakes than any other semiconductor fab — which is exactly why it's robust against them. Earthquakes aren't the main risk.

Cyber: WannaCry variant in 2018, $170M in damage, Q3 revenue -3%, recovery in days. But: a state-directed attack (China APT or North Korea Lazarus) on EUV control software would be a different game.

IP theft: China has been running systematic recruitment campaigns against TSMC engineers for years. Effect: SMIC reached 7nm, but not in TSMC quality. Slow erosion, no immediate damage.

What does the blockade scenario concretely mean for you?

Three personal anchors for the most likely of the three scenarios (global GDP −5%, ~$5 trillion in damage in the first year). For full invasion, the effect roughly doubles.

As a Germany resident
Recession like 2009, inflation like 2022, at the same time. Energy, food, and import prices rise because global supply chains collapse. Real income drops 5-8%.
As an employee
Tech sector lays off first and hardest. AI investment stops without new GPUs. Anyone in a semiconductor-adjacent industry (automotive, medtech, machinery) sees 12 to 24 months of short-time work or layoffs.
As a tech buyer
iPhone 18 delayed, MacBook 25-40% more expensive, new GPU for AI workloads simply not available. Existing devices stay in use longer, repair beats replacement as the default.

Why doesn't it escalate anyway? The incentives of all six actors.

The three scenarios above don't have a hard probability, because the probability of occurrence follows from the incentives of the actors. Each relevant actor holds a lever and carries a risk. Anyone who deviates unilaterally loses more than they win. That's the "Silicon Shield": not a property of any one chip, but an equilibrium of mutual incentives.

Actor What do they want? Lever Risk
USA Washington · Silicon Valley Domestic capacity for leading-edge chips. Keep China behind on lithography.
  • CHIPS Act $52.7B, of which ~$33.7B in direct funding + ~$5.5B in loans finalized[24]
  • Export controls (BIS rules)
  • 25% tariff on advanced-compute chips since January 14, 2026 (Section 232); categorical exemptions for US datacenter and capacity imports[25]
Trump rolls back the CHIPS Act while tariffs simultaneously demand onshoring — pressure without incentive.
China Beijing · Shanghai (SMEE) Self-sufficiency. "Made in China 2025" target 70%, reality ~35%.
  • Big Fund III $47.5B[15], cumulatively ~$94B in state capital
  • Critical minerals (gallium/germanium/antimony) — export ban Dec 2024, suspended since Nov 9, 2025[26]
If they take Taiwan militarily, TSMC is dead. They don't have ASML equipment either. Their own EUV is years from volume production.
Taiwan Taipei · Hsinchu Hold the status quo. Lai Ching-te (DPP): TSMC is leverage and vulnerability at the same time.
  • 60% global semiconductor manufacturing, 92% leading-edge logic
  • Defense budget $40B over 8 years (November 2025)[27]
  • Target 3.3% of GDP by 2026, 5% by 2030[27]
KMT + TPP block Lai's defense budget. Silicon Shield erodes via Arizona investments (March 2025: another +$100B[28]).
Netherlands / EU Veldhoven · Oberkochen Protect ASML margins, signal EU sovereignty outward, avoid being ground up between US and China.
  • ASML EUV monopoly
  • DUV immersion machines under Dutch export license
Intel Magdeburg cancelled July 2025 (previously €30B investment planned)[31]. EU Chips Act delivered €13.75B vs. $33.7B in US grants. EU auditor: 20% target by 2030 "very unlikely".
NVIDIA Santa Clara More CoWoS capacity, faster. Diversification as a hedge, no switch away from TSMC.
  • Over 60% of global CoWoS capacity reserved
  • Order backlog Blackwell + Rubin: ~$500B visibility through end of 2026 (GTC Washington October 2025)[29]
Single point of failure Taiwan. NVIDIA sells Blackwell as fast as TSMC can build them. If TSMC tips, NVIDIA tips with it.
TSMC itself Hsinchu Hold margins, sit out geopolitics, serve all major customers. No commitment as anyone's vassal.
  • 70% foundry market share, ~92% leading edge
  • C.C. Wei (CEO since June 2024) more diplomatic than Mark Liu — March 2025 at the White House
Arizona too big → Silicon Shield brittle. Arizona too small → Trump tariff. Taiwan falls → company dead anyway.
← scroll horizontally →
Silicon Shield · the central game-theory thesis

Why the system holds, and what would break it

For · it holds

  1. Over $2 trillion in direct disruption + $1.6 trillion in annual revenue loss at chip consumers (Rhodium Group[2]): China needs TSMC chips itself.
  2. Mutually Assured Financial Destruction: a major Taiwan crisis would have financial consequences before the first shot is fired (CSIS "Scared Strait"[37]).
  3. A failed Taiwan invasion could destabilize CCP rule. Xi doesn't risk that in the economically weaker phase.

Against · it breaks

  1. Mark Liu's own statement: if invaded, the fab is dead. The threat works only before escalation, not during.
  2. Arizona drain: the more US fabs TSMC has, the smaller the marginal value of Hsinchu.
  3. Stochastic escalation: Hainan 2.0, a downed US recon plane, a rammed ship. The system is a fragile robustness — stable in the median, with thicker than in 2020.
What breaks the equilibrium? Seven triggers: Xi health or power event · Lai sovereignty declaration · Arizona inflection point (> 40% advanced-node output) · EUV breakthrough in China · US defense collapse (Trump credibly signals disengagement) · economic crisis in China (rally-around-the-flag) · naval incident that escalates.
Act 3 · What you do now

Four concrete levers — hygiene, not panic.

When the shock arrives and Apple, NVIDIA, and your tool stack all wobble at once, only preparation helps. The concentration on TSMC, ASML, and Hsinchu came out of 40 years of efficiency logic, not bad intent. That's what makes it hard to unwind. Fixing the system isn't in your hands. Checking your own exposure is. Four levers, doable in a weekend.*

Four levers
  1. 01 Multi-LLM strategy for your tools. Not only ChatGPT, not only Claude. At least two providers in parallel, so you can switch within days if rate limits or a price hike kick in. Cursor, Aider, and LibreChat make model switching relatively easy. Route three workflows per week deliberately to the second provider, so you don't have to learn it under stress. How I did this myself: setting up cloud with Antigravity.
  2. 02 Install local models as a hedge. Llama 3.3, Mistral, Qwen run on a Mac with 32 GB+ RAM or a 4090. Enough for roughly 70% of use cases. Not as the primary tool, but as a backup if cloud inference gets rationed. Once set up, it stays out of the way.
  3. 03 30-minute stress test on your stack. Take half an hour and go through the five tools you use daily. For each ask: if the provider introduces rate limits tomorrow or triples the price, what's my plan? Three classes: (a) backup ready, (b) can migrate in two weeks, (c) no idea what to do. Class (c) is the real tail-risk list. If you never wrote it down, you're surprised when it hits. And that part is avoidable.
  4. 04 Check stock concentration risk. If your portfolio holds AI or tech ETFs: look at the top 10. If NVDA + TSM + ASML + AAPL + AMD + AVGO combined exceed 30%, you have a concentrated semiconductor exposure you may not have signed up for consciously. In a TSMC outage, all of it correlates to the downside. What you do with that is your call, and your financial advisor's if you have one. This is the prompt to actually do the math once.
Back to the hypothesis I started with

"We're screwed — if TSMC stops."

True — and more so than I thought going in. It isn't one company. Four single-source nodes wired in series: ZEISS, ASML, TSMC, Hsinchu. Each one without a real second source. If one breaks, the whole stack breaks at that level. Even the fastest catcher-up takes 2 to 3 years, the slowest 25 to 30 — a whole generation.

True — but less likely than most people think. Exactly because the US, China, and Taiwan all lose more than they win in a break, the equilibrium holds. That's why doomsday voices have been predicting the crash for years (decades, really), and why it hasn't happened so far. The system is a fragile robustness: stable in the median, with thicker than the last 20 years.

No reason to panic. But a reason to honestly look at your portfolio, tech stack, and watchlist once, before that Wednesday in April 2027 arrives.

* This article is editorial journalism, not investment advice. I'm not licensed by BaFin and don't make personal recommendations on securities. The levers are prompts for reflection — anyone moving real money or restructuring positions should bring in a licensed financial advisor.

Making-of · how this article came together (AI pipeline plus several days of hands-on review)
Making-of

How this article came together

This article didn't come out of a single sitting. An AI pipeline (parallel research agents, fact-check, synthesis, HTML build) delivers the raw material on each run — the road from first draft to here took a few days, with multiple iterations, lots of my own feedback, plus cutting, rewriting, and follow-up research. A pure auto-runtime number wouldn't be honest: without the manual rounds, this would have stayed a well-sorted data dump, not a readable article. Here are the phases each iteration ran through.

Pipeline
PHASE 01
Plan mode + topic brief
Hypothesis nailed down, section skeleton set, source rule (primary sources < 12 months).
PHASE 02
4 parallel research agents
Market + technology · substitution lag · Taiwan risk · actor incentives. Four Sonnet agents in parallel, ~15 min per run.
PHASE 03
Research fact-check
28 claims checked via WebSearch. 21 verified, 4 with nuance, 0 hard errors.
PHASE 04
Synthesis (Opus)
Distillation from 4 files: pros/cons, hook calculation, 3 chart ideas, 6 corrections folded in.
PHASE 05
HTML build
Brand-template based, 14 sections, 3 charts (market share, substitution lag, GDP hit).
PHASE 06
Article fact-check
Math checks: percentages, ratios, consistency across all numbers. GO or fix list.
PHASE 07
UI/UX + SEO/GEO review
Heading hierarchy, mobile, JSON-LD, AI crawler readability (Perplexity, ChatGPT, Claude).
PHASE 08
Copy edit
Spelling, grammar, idiom. Industry terms (foundry, yield, wafer) stay in.
PHASE 09
Verify + deploy
Playwright screenshots, walkthrough, sign-off, then Netlify deploy via deploy.sh.
Methodology · what's behind the △ anchor classes. A few numbers in the article aren't direct source quotes but my own plausibility calculations from publicly known spec values. Examples:
  • "92B transistors = 11× world population" — Apple M3 Max spec (public) ÷ 8.1B world population.
  • "99.9999999% hit rate" — derivation: at 92B transistors, one broken switch in the wrong place ruins the chip; required yield is therefore > 1 − (1/92B) ≈ 99.999999999% per transistor.
  • "3nm = 15 silicon atoms wide" — silicon lattice constant 0.543 nm ÷ 4 atoms per unit; 3 nm ÷ ~0.2 nm ≈ 15.
  • "ZEISS 0.1 nm unevenness ≈ 0.1 mm bump scaled to Germany" — scale conversion: 0.1 nm ÷ 30 cm mirror × 1,000 km Germany edge = ~333 µm, rounded to 0.1 mm order of magnitude.
  • "More transistors on a pinhead than Germany has residents" — 200M/mm² density × ~0.5 mm² pinhead area ≈ 100M transistors, vs. 84M German residents.
Sources for the underlying values: Apple Press[38], ASML Annual Report[12], ZEISS SMT technology pages.
Why this effort? On a topic with this much noise and this many contradictory sources, consistency is the most important trust signal. If the Bloomberg $10T number, the NVIDIA backlog, the CHIPS Act budget, and the yield gap all line up, the article holds up against the fact-checker too. The AI pipeline can deliver the raw material and a first consistency check — the decision of what matters, what gets cut, where the personal take sits, and how the article reads stays manual. That part ate most of the time.

Sources

Convention in the article: hard numbers, quotes, and external model outputs are marked inline with a superscript source number (e.g. [1]). Click jumps directly into this list. Three confidence levels, visually distinct:

[N] hard primary source (earnings, government, peer-reviewed) [N°] analyst estimate, industry estimate, wargame, or model output [] own plausibility calculation — methodology in the making-of

Where a number appears without a footnote, it's a plausibility synthesis across multiple sources listed here, or it links to a bundled entry (e.g. ASML group numbers from the Annual Report[12] where not marked otherwise). Goal: every checkable claim should be traceable without the reader having to do their own research.

  1. 01Bloomberg Economics — "The 10 Trillion Fight" (Jan 2024) · global GDP modeling -9.6% / -5% · plus blockade simulation newsletter (May 2024) · ~$5T blockade shock
  2. 02Rhodium Group — "Taiwan Economic Disruptions" (2024) · over $2T directly at risk
  3. 03TrendForce — Samsung 3nm yield 50% vs. TSMC 90% (May 2025)
  4. 04Tom's Hardware — ASML CEO "China 10-15 years behind" (December 2024) · based on Fouquet interview with NRC, Dec 18, 2024
  5. 05TrendForce — China EUV prototype December 2025
  6. 06CSIS — "First Battle of the Next War" wargame
  7. 07Allianz Risk Barometer 2026 (PDF) · 51% see supply-chain paralysis from geopolitics as the most plausible black-swan scenario
  8. 08CFR Conflict Risk Assessment 2026 · Cross-Strait crisis > 50% probability
  9. 09USNI News — Pentagon: no fixed timeline for Xi (March 2026)
  10. 10CNN Fareed Zakaria GPS — Mark Liu interview transcript (July 31, 2022) · original quote "render TSMC factory not operable, because this is such a sophisticated manufacturing facility" · secondary: Fortune coverage
  11. 11TSMC Q4 2025 SEC 6-K Earnings · $122.5B revenue · NVIDIA 19% · Apple 17%
  12. 12ASML Q4 2025 Annual Press Release · €32.7B · 48 EUV systems · 100% EUV monopoly
  13. 13SemiAnalysis Newsletter — "TSMC Overseas Fabs: A Success" 2025 · Hsinchu cluster, CoWoS lock-in
  14. 14Allianz Trade Global Survey 2025 · 34% of companies have relocated production
  15. 15Nikkei Asia — China Big Fund III $47.5B (December 2024)
  16. 16TrendForce HBM Market Tracker 2025 · SK hynix dominates with ~50% share, Samsung ~35%, Micron ~15%
  17. 17Trefis — NVIDIA risk-premium model (November 2025, archive.org) · -30 to -50% drawdown in the hypothetical Taiwan scenario; sell-side model, not bank consensus
  18. 18Stanford HAI — AI Index Report 2025 · token price drops by factor 10 per year at comparable performance
  19. 19AppleInsider — Apple's Taiwan production share (2024+) · ~85% of A/M chips at TSMC
  20. 20Chinatalk — What Happens Without Taiwan's Chips? · recovery synthesis: ~15 years to pre-war levels, talent multiplier
  21. 21Rest of World — Inside TSMC's Arizona expansion (2024) · Morris Chang "2 a.m. in Taiwan" quote, cleanroom reporting
  22. 22Caredge / CNBC — chip crisis 2021 · 15M vehicles globally not built from a pure allocation problem
  23. 23CSIS — "Lights Out? Wargaming a Chinese Blockade of Taiwan" (July 2025) · 26 wargame iterations of the blockade scenario; Cancian/Cancian/Heginbotham · escalation paths hard to contain
  24. 24US Commerce — CHIPS Act Programs Status Report (January 2025) · $52.7B total volume; ~$33.7B in direct funding + $5.5B in loans finalized
  25. 25Supply Chain Dive — Trump Section 232 25% tariff on advanced computing chips (Jan 14, 2026) · affects NVIDIA H200, AMD MI325X; categorical exemptions for US datacenter and capacity build-out (no explicit TSMC exemption)
  26. 26CNBC — China suspends gallium/germanium/antimony export ban (Nov 9, 2025) · original ban December 2024, suspended through Nov 27, 2026 after US-China trade truce
  27. 27USNI News — Taiwan $40B defense supplemental (Nov 26, 2025) · 8-year program under Lai Ching-te; 2026 defense budget: $31.2B or 3.3% of GDP; target 5% of GDP by 2030
  28. 28CBS News — TSMC announces additional $100B US investment (March 3, 2025) · joint appearance by Trump and TSMC CEO C.C. Wei at the White House; total Arizona investment now $165B
  29. 29CNBC — Jensen Huang "half a trillion" backlog (GTC Washington, Oct 28, 2025) · Blackwell + Rubin visibility through end of 2026
  30. 30CNN — Tesla-Samsung foundry deal $16.5B (July 28, 2025) · multi-year contract through 2033 for AI6 chips on Samsung 2nm in Taylor, Texas; AI5 made by TSMC
  31. 31Brussels Signal — Intel cancels Magdeburg fab (July 24, 2025) · €30B investment, of which €10B state subsidy; first delayed by 2 years in September 2024 (Euronews postponement), then cancelled outright under CEO Lip-Bu Tan in July 2025
  32. 32RFA — CIA Director Burns on PLA readiness 2027 (Feb 2, 2023, Georgetown) · "he's instructed the People's Liberation Army to be ready by 2027 to conduct a successful invasion"; caveat: Xi has not yet made a decision; comparable assessment in Aspen Security Forum transcript July 2023
  33. 33Taiwan MODA — Submarine Cable Incidents Report (2026 Annual Release) · 7 submarine cable incidents in 2025; 4-year average 7-8 incidents per year
  34. 34The Diplomat — Matsu submarine cable cut February 2023 · both cables severed by Chinese vessels Feb 2/8, 2023; roughly 50 days of internet outage for ~13,000 residents until March 31, 2023
  35. 35Stock Analysis — iShares Core S&P 500 ETF (IVV) Holdings (as of May 20, 2026) · NVIDIA 8.51% + Apple 6.95% + Broadcom 3.10% = 18.56% combined; Vanguard VOO similar at 16.5%
  36. 36USGS — M 7.4 earthquake 15 km S of Hualien City, Taiwan (April 3, 2024) · event ID us7000m9g4; strongest quake in Taiwan since 1999
  37. 37CSIS — "Scared Strait: Understanding the Economic and Financial Impacts of a Taiwan Crisis" (December 2023) · Blanchette/DiPippo/Johnstone; eight investor scenarios; China and Taiwan carry the largest long-term damage, China risks "uninvestable" status
  38. 38Apple Newsroom — M3, M3 Pro, M3 Max press release (October 2023) · 92B transistors M3 Max, TSMC 3nm (N3B), public chip specs as the basis for all M3-related calculations in the article