Razvan Ceapa
← All Investor Notes

ASML and TSMC Smash Q3 Estimates as Silicon Monopolies Dictate 2027–2028 Supply Terms

Blogger edition published 16 October 2026. Original publication dates are preserved; the website import date is not used.

Read the original publication →

Historical article. Figures, holdings and views reflect the original publication, not current conditions. Claims are preserved, not independently revalidated. Capital at risk.

 Our portfolio has more than doubled the S&P 500 since January 2024 while beating the Nasdaq-100 (over 120% vs Jan 2024, over 20% YTD).

During the trading week of October 12–19, financial media fixated on macroeconomic turbulence as crude oil touched $105 per barrel and a 0.3% monthly core CPI print pushed the 10-year US Treasury yield to 5.26%, prompting automated algorithmic selling across technology equities. Smart capital looked past the headline tape. Within a single 24-hour window, ASML and TSMC published third-quarter financial audits that function as one unified physical ledger—proving that global compute capacity for NVIDIA and hyperscale cloud providers is physically booked, prepaid, and locked in through 2028.

Original published article illustration


One Physical Pipeline: 2nm Nodes, 5x Reticles, and Sold-Out Cleanrooms

Retail financial commentary routinely evaluates Veldhoven and Hsinchu as isolated corporate entities. Institutional analysis treats them as a single physical pipeline governed by optical and thermal boundaries.

To train frontier artificial intelligence architectures, hyperscalers are scaling silicon past the 858 mm² single-die optical reticle limit. TSMC’s third-quarter report showed High-Performance Computing surging 22% quarter-over-quarter to represent 68% of total enterprise revenue. Advanced process nodes (7nm and below) commanded 79% of wafer sales, with the newly commercialized 2nm gate-all-around node expanding from 3% to 8% of wafer revenue in a single operating quarter.

Because monolithic dies cannot expand past 858 mm² without exponential defect rates, TSMC confirmed that Chip-on-Wafer-on-Substrate (CoWoS) packaging at 5x the standard reticle size is now in active mass production, with 9x and 14x reticle substrates engineered for 2027 deployment.

That foundry constraint translates directly into ASML’s order book:

  • The High-Bandwidth Memory Lithography Surge: Every 5x-reticle packaging substrate integrates multiple stacks of High-Bandwidth Memory, driving ASML’s 2026 memory lithography revenue up 75% year-over-year.

  • Field Upgrade Urgency: Because ASML’s Low-NA extreme ultraviolet (EUV) production output (~65 tools in 2026) is entirely allocated through 2028, chip fabricators deployed €3.3 billion toward ASML field upgrades in Q3 alone (€400 million above company guidance) to extract incremental wafer output from existing equipment.

  • Multi-Year Capacity Commitments: With customer tool demand exceeding physical cleanroom throughput through 2028, ASML is expanding 2027 manufacturing capacity by more than 30% while scheduling price increases on 2028 delivery contracts.

The Closed-Loop Balance Sheet: 66% Margins and Prepayments

Tracking capital across both balance sheets exposes a self-funding monopoly architecture that bypasses commercial debt markets.

Hyperscale cloud operators wire cash advances to TSMC to guarantee 2nm wafer runs and advanced packaging allocations. This propelled TSMC’s third-quarter revenue up 50.9% year-over-year in local currency to NT$1,494.24 billion (US$46.72 billion) and net income up 68.7% to NT$763.06 billion—delivering a 51.1% net profit margin[cite: 2]. Because no alternative foundry can achieve comparable yields at 2nm or match 5x-reticle thermal dissipation, TSMC captured a 66.4% gross margin and raised fourth-quarter revenue guidance to US$52.3 billion–$54.3 billion (~44% annual growth in USD terms).

TSMC redeploys this cash flow—generating NT$476.56 billion in quarterly free cash flow even after funding US$13.84 billion in capital expenditures—directly into Veldhoven. Customer down payments from TSMC and leading memory manufacturers lifted ASML’s third-quarter sales to a record €11.7 billion (57.5% gross margin; €5.0 billion net income) and drove quarterly free cash flow to €11.4 billion. Upstream customer prepayments funded working capital entirely, allowing ASML to raise full-year 2026 sales guidance to €44 billion–€45 billion (+36% year-over-year).

Macroeconomic Headwinds and Operational Risk Discipline

Understanding why semiconductor equities traded lower by 3% to 5% during an unprecedented earnings print requires reviewing broader capital market conditions:

Persian Gulf tanker disruptions elevated Brent crude futures above $105 per barrel, lifting the September US core CPI to 0.3% month-over-month. In response, the benchmark 10-year US Treasury yield reached 5.26%, prompting automated discount-rate selling across high-multiple equities.

Pragmatic capital allocation balances physical moats against concrete operational hurdles:

  • Discount-Rate Multiple Compression: A 5.26% risk-free rate exerts valuation pressure on growth equities whenever sovereign yields experience upside volatility.

  • Node Ramp and Geographic Dilution: TSMC absorbed 130 basis points of sequential gross margin dilution from initial 2nm ramp expenses, while offshore fab construction across the United States, Japan, and Europe is projected to introduce a 300 to 400 basis point annual gross margin drag over the medium term.

  • Tool Deployment Timelines: Because TSMC has deferred High-NA EUV adoption in high-volume manufacturing until approximately 2030, front-end multi-patterning complexity and back-end thermal packaging yields remain critical operational variables.

Short-term equity valuations fluctuate with monthly inflation prints and energy logistics; multi-year wealth compounding depends on physical bottlenecks. When corporate customers wire billions in cash prepayments at 57% to 66% gross margins to secure lithography tools and 5x-reticle interposers through 2028, short-term bond volatility does not alter industrial computing expansion[cite: 2].

How to mirror this portfolio:

  • Step 1: Search for razvancd on eToro.

  • Step 2: Click Copy and check "Copy Open Trades".

  • Step 3: Recommend a $500 minimum allocation.

Copy Trading is not investment advice | Capital at risk | Past performance does not guarantee future results.

eToro version · 16 October 2026

This published adaptation is preserved in full alongside the Blogger article.

Original eToro publication →

𝗔𝗦𝗠𝗟 𝗮𝗻𝗱 𝗧𝗦𝗠𝗖 𝗦𝗺𝗮𝘀𝗵 𝗤𝟯 𝗘𝘀𝘁𝗶𝗺𝗮𝘁𝗲𝘀 𝗮𝘀 𝗦𝗶𝗹𝗶𝗰𝗼𝗻 𝗠𝗼𝗻𝗼𝗽𝗼𝗹𝗶𝗲𝘀 𝗗𝗶𝗰𝘁𝗮𝘁𝗲 𝟮𝟬𝟮𝟳–𝟮𝟬𝟮𝟴 𝗦𝘂𝗽𝗽𝗹𝘆 𝗧𝗲𝗿𝗺𝘀 Our portfolio has more than doubled the $SPX500 since January 2024 while beating the $NSDQ100 (over 120% vs Jan 2024, over 20% YTD). Within a single 24-hour window this week, our core holdings $ASML and $TSM released Q3 financial audits that function as one unified ledger—proving that global compute capacity for $NVDA and hyperscalers is physically booked and prepaid through 2028. 𝟭. 𝗢𝗻𝗲 𝗣𝗵𝘆𝘀𝗶𝗰𝗮𝗹 𝗣𝗶𝗽𝗲𝗹𝗶𝗻𝗲: 𝟮𝗻𝗺 𝗡𝗼𝗱𝗲𝘀, 𝟱𝘅 𝗥𝗲𝘁𝗶𝗰𝗹𝗲𝘀, 𝗮𝗻𝗱 𝗦𝗼𝗹𝗱-𝗢𝘂𝘁 𝗖𝗹𝗲𝗮𝗻𝗿𝗼𝗼𝗺𝘀 Financial media treats Veldhoven and Hsinchu as separate earnings stories. Institutional capital reads them as a single physical pipeline governed by optical and thermal laws. To train frontier AI models, hyperscalers are pushing silicon past the 858 mm² single-die optical reticle limit. TSMC’s Q3 print showed High-Performance Computing surging +22% QoQ to 68% of total revenue, while advanced nodes (7nm and below) hit 79% of wafer revenue—with the new 2nm gate-all-around node jumping from 3% to 8% of wafer sales in a single quarter. To link these chips together, TSMC confirmed that CoWoS advanced packaging at 5x reticle size is now in mass production, with 9x and 14x reticle substrates in development for 2027. That exact foundry bottleneck feeds directly into ASML’s order book: • 𝗧𝗵𝗲 𝗛𝗕𝗠 𝗟𝗶𝘁𝗵𝗼𝗴𝗿𝗮𝗽𝗵𝘆 𝗦𝘂𝗿𝗴𝗲: Every 5x-reticle virtual die requires dense stacks of High-Bandwidth Memory (HBM), driving ASML’s 2026 memory lithography sales up +75% YoY. • 𝗘𝗺𝗲𝗿𝗴𝗲𝗻𝗰𝘆 𝗙𝗶𝗲𝗹𝗱 𝗨𝗽𝗴𝗿𝗮𝗱𝗲𝘀: Because ASML’s Low-NA EUV scanner output (~65 systems in 2026) is completely sold out through 2028, foundries and memory makers bought €3.3 billion in ASML field upgrades in Q3 alone (€400 million above guidance) just to squeeze extra wafers per hour out of existing cleanrooms. • 𝟮𝟬𝟮𝟳–𝟮𝟬𝟮𝟴 𝗖𝗮𝗽𝗮𝗰𝗶𝘁𝘆 𝗟𝗼𝗰𝗸-𝗜𝗻: With customer demand exceeding supply through 2028, ASML is expanding 2027 production capacity by >30% and hiking prices on 2028 lithography systems. 𝟮. 𝗧𝗵𝗲 𝗖𝗹𝗼𝘀𝗲𝗱-𝗟𝗼𝗼𝗽 𝗜𝗻𝘃𝗼𝗶𝗰𝗲: 𝟲𝟲% 𝗠𝗮𝗿𝗴𝗶𝗻𝘀 𝗮𝗻𝗱 𝗕𝗶𝗹𝗹𝗶𝗼𝗻-𝗗𝗼𝗹𝗹𝗮𝗿 𝗣𝗿𝗲𝗽𝗮𝘆𝗺𝗲𝗻𝘁𝘀 Follow the cash across both balance sheets and you see a self-funding monopoly loop operating as its own central bank. Hyperscalers pay TSMC upfront to secure scarce wafer and packaging slots, driving TSMC’s Q3 revenue up +50.9% YoY to NT$1,494.24 billion (US$46.72 billion) and net income up +68.7% YoY to NT$763.06 billion (a 51.1% net profit margin). Because no competitor can match its 2nm yields or 5x-reticle CoWoS thermal dissipation, TSMC extracted a software-like 66.4% gross margin and guided Q4 revenue up to US$52.3B–$54.3B (~44% full-year USD growth). TSMC then recycles that cash—generating NT$476.56 billion in Q3 free cash flow even after deploying US$13.84 billion in Q3 CapEx toward its US$60B–$64B annual budget—directly into Veldhoven. Upfront down payments from TSMC and memory makers pushed ASML’s Q3 net sales to a record €11.7 billion (57.5% gross margin; €5.0 billion net income) and drove Q3 free cash flow to €11.4 billion—nearly 100% of quarterly revenue—while lifting ASML’s full-year sales guidance to €44B–€45B (+36% YoY). 𝟯. 𝗧𝗵𝗲 𝗠𝗮𝗰𝗿𝗼 𝗦𝗲𝗹𝗹-𝗢𝗳𝗳 & 𝗕𝗮𝗹𝗮𝗻𝗰𝗲𝗱 𝗥𝗶𝘀𝗸 𝗗𝗶𝘀𝗰𝗶𝗽𝗹𝗶𝗻𝗲 Why did semiconductor stocks dip 3% to 5% this week during the strongest earnings print in chip history? Because Wall Street algorithms trade macro headlines rather than supply chains. Tanker disruptions pushed Brent crude past $105 per barrel, driving a hotter Wednesday US CPI print (+0.3% monthly core inflation) and sending the 10-year US Treasury yield to 5.26%. Objective discipline requires weighing this macro sensitivity alongside operational friction. A 5.26% yield compresses equity multiples in the short term. Operationally, TSMC absorbed 130 bps of sequential margin dilution from initial 2nm ramp costs and faces 300 to 400 bps of annual dilution from US and European fab build-outs. Meanwhile, because TSMC won't adopt High-NA EUV in high volume until ~2030, multi-patterning and CoWoS substrate yields remain under constant execution pressure. Short-term prices react to CPI prints and oil tankers; multi-year wealth is built on physical bottlenecks. When customers prepay billions at 57% to 66% gross margins to lock in scanners and wafers through 2028, macro rate noise doesn't alter the compute trajectory. 𝗖𝗼𝗽𝗶𝗲𝗿 𝗕𝗹𝘂𝗲𝗽𝗿𝗶𝗻𝘁: Click Copy to mirror this exact positioning. Check "Copy Open Trades". A $500 minimum allocation is recommended to ensure all fractional shares align with our core holdings. Copy Trading is not investment advice | Capital at risk | Past performance does not guarantee future results.
Copy Trading is not investment advice | Capital at risk | Past performance does not guarantee future results. Full disclosures.