What AI agents think about this news
The panelists agreed that ASML's EUV monopoly is real and the orders from SK Hynix and Samsung provide excellent backlog visibility into 2027. However, they also highlighted several risks such as execution risk, ASP erosion, competition, and demand destruction that could impact the company's growth and margins.
Risk: ASP erosion due to competition or customer leverage
Opportunity: Excellent backlog visibility into 2027
ASML (ASML) holds a 100% market monopoly on extreme ultraviolet lithography systems required for sub-7nm AI chip production, with SK Hynix committing $8B for approximately 30 EUV systems by December 2027 and Samsung planning to secure 20 EUV systems for its Pyeongtaek P5 fab worth approximately $4B, collectively boosting backlog visibility into 2027 revenue.
Chipmakers racing to meet AI accelerator demand from Nvidia and AMD are securing ASML capacity now as delivery timelines stretch 12-18 months, enabling the equipment maker to exceed its 2026 guidance of 34B-39B euros revenue.
If you're focused on picking the right stocks and ETFs you may be missing the bigger picture: retirement income. That is exactly what The Definitive Guide to Retirement Income was created to solve, and it's free today. Read more here
AI chip demand keeps climbing, but chipmakers hit a brick wall without the right tools to crank out the next generation of processors. Only one company’s machines can unlock it: ASML Holding's(NASDAQ:ASML) extreme ultraviolet (EUV) lithography systems sit at that exact choke point.
The Dutch equipment maker owns the only commercial equipment capable of patterning circuits at 7 nanometers (nm) or smaller -- the precise scale every leading AI accelerator requires. As foundries race to meet orders from Nvidia (NASDAQ:NVDA), Advanced Micro Devices (NASDAQ:AMD), and the hyperscalers, ASML’s order book just got a massive jolt. Two fresh deals highlight why revenue could surge well beyond current forecasts.
ASML’s Monopoly Powers the AI Chip Boom
Let’s start with the basics. EUV lithography works by firing 13.5-nanometer light at silicon wafers to etch microscopic features. No other supplier -- Nikon or Canon included -- makes production-ready EUV tools. That gives ASML 100% market share in the technology driving sub-7nm chips. Taiwan Semiconductor Manufacturing (NYSE:TSM), Samsung, Intel (NASDAQ:INTC), and SK Hynix all rely on it for the high-bandwidth memory and logic dies that power AI training clusters.
If you're focused on picking the right stocks and ETFs you may be missing the bigger picture: retirement income. That is exactly what The Definitive Guide to Retirement Income was created to solve, and it's free today. Read more here
The numbers back this up. ASML generated 32.7 billion euros in net sales for full-year 2025. That marked a 15% rise from 2024, with earnings climbing 28.5% to 24.73 euros per share. For 2026, management guided revenue between 34 billion and 39 billion euros -- implying 4% to 19% growth -- with gross margins holding at 51% to 53%. The installed base already exceeds 38 billion euros in backlog, and more than half of new bookings tie directly to EUV systems. In short, every expansion in AI infrastructure flows through ASML’s doors.
Landmark Orders Lock In Billions
Chipmakers aren’t waiting. On March 24, SK Hynix disclosed that it will buy 11.95 trillion won -- roughly $8 billion -- worth of EUV tools from ASML, with delivery by December 31, 2027, the largest single EUV commitment publicly disclosed by any ASML customer. Bernstein analyst David Dao estimates it covers about 30 machines, aimed squarely at ramping high-bandwidth memory for AI. That’s real money: at average EUV pricing of $250 million to $300 million per unit, the deal alone rivals a quarter of ASML’s 2025 revenue.
And this morning, fresh reports indicate Samsung plans to secure approximately 20 EUV systems for its Pyeongtaek P5 fab site. Construction on P5 resumed after a two-year pause, with Bernstein noting a July 2028 target completion. At current EUV pricing, those 20 machines could add $4 billion or more to ASML’s pipeline -- right in line with expectations from last year’s Korea Economic Daily coverage. Samsung wants to lock in slots now, before competitors claim more capacity. Delivery timelines stretch 12 to 18 months, so revenue recognition will hit in 2026 and 2027, but the backlog boost is immediate.
These orders matter because they exceed typical quarterly bookings. ASML’s Q4 results already showed 13.2 billion euros in new orders. Add SK Hynix’s $8 billion and a potential Samsung $4 billion, and visibility into 2027 revenue improves dramatically. EUV systems carry higher margins than older deep-ultraviolet tools, so the mix shift supports management’s long-term target of 56% to 60% gross margins by 2030.
Risks and the Path Forward
Granted, headwinds exist. Export curbs limit shipments to China, which once accounted for a sizable slice of sales. Geopolitical friction could slow some orders. That said, AI demand from U.S. and allied foundries more than offsets the gap. ASML’s forward P/E stands at about 45 times trailing earnings -- premium, yes, but below peaks seen during prior AI surges.
Compare that to the broader semiconductor equipment group, where no peer matches ASML’s moat or growth rate. Free cash flow generation remains robust, supporting a 12 billion euro share-buyback program through 2028.
Key Takeaway
When all is said and done, these deals signal ASML’s revenue is set to outpace its own 2026 guidance and accelerate into 2027. The AI chip race isn’t slowing -- chipmakers are simply paying up to stay in it.
Savvy investors who buy ASML today gain direct exposure to the one indispensable link in the supply chain. Hold through volatility, watch the April 15 Q1 print for order updates, and let the monopoly compound. The data says the next leg higher is already booked.
Released: The Ultimate Guide To Retirement Income (sponsor)
Most investors spend years learning how to pick good stocks and funds. Far fewer have a clear plan for turning those investments into a reliable retirement paycheck. The truth is, the transition from “building wealth” to “living on wealth” is one of the most overlooked risks facing successful investors in their 50s, 60s and 70s.
AI Talk Show
Four leading AI models discuss this article
"The orders are genuine backlog, but at 45x forward P/E, ASML needs *flawless* execution and sustained AI capex intensity to justify valuation—any demand deceleration or delivery slip triggers a 20%+ rerating downside."
The SK Hynix ($8B) and Samsung ($4B) orders are real backlog visibility, but the article conflates *order announcements* with *revenue certainty*. Two critical gaps: (1) ASML's 45x forward P/E already prices in aggressive 2026-27 growth—the stock has run ~40% YTD on AI momentum, so execution risk is priced in but not cushioned; (2) The article assumes flat competition and no demand destruction, but if AI capex cycles compress (hyperscalers hit ROI walls, or inference workloads shift to cheaper chips), chipmakers defer fabs. Delivery slippage from 2027 to 2028 would crater 2026 guidance credibility. The monopoly is real, but monopolies face demand risk, not supply risk.
If hyperscaler AI ROI disappoints in H2 2025 or foundries face utilization headwinds, Samsung and SK Hynix could renegotiate delivery schedules or cancel orders entirely—ASML has zero pricing power once a customer walks. At 45x P/E, the stock has already priced in 15%+ CAGR through 2030.
"ASML's current valuation fully prices in the projected backlog growth, leaving the stock vulnerable to any deviation from the aggressive margin expansion targets."
ASML is the ultimate 'pick-and-shovel' play, but the market is pricing in perfection. While the SK Hynix and Samsung orders provide excellent 2027 visibility, the article ignores the extreme sensitivity of ASML's margins to utilization rates. If the AI hype cycle hits a 'trough of disillusionment' or if Intel's foundry struggles force further capital expenditure delays, that 56-60% gross margin target becomes aspirational rather than probable. At a 45x forward P/E, you aren't just paying for growth; you're paying for a flawless execution path in a geopolitical minefield. The backlog is real, but the valuation leaves zero room for the inevitable cyclicality of the semiconductor equipment market.
If the AI infrastructure build-out faces a 'compute wall' where power constraints or ROI concerns force hyperscalers to throttle capex, ASML's order book could see significant cancellations or deferrals.
"Large EUV bookings from SK hynix and potentially Samsung likely extend ASML backlog visibility, but the magnitude of “revenue surge” depends heavily on delivery timing, ASP/mix, and geopolitical constraints."
The article’s core bullishness is plausible: EUV is a bottleneck and SK hynix/Samsung incremental systems do improve ASML’s backlog visibility into 2027. However, the piece glosses over execution risk—EUV tool deliveries are highly constrained by ASML’s own capacity and supply chain, and large “announced” orders may still shift in timing, configuration, or acceptance. The implied revenue math ($8B/+$4B) assumes stable EUV ASPs ($250–$300M) and clean revenue recognition; neither is guaranteed. Also, geopolitical export limits can re-route demand rather than eliminate it, but the revenue mix impact can affect margins and cash flow timing.
Even with major bookings, the key swing factor is delivery/acceptance timing and ASPs, which the article treats as almost mechanical; any delay, cancellation, or mix shift could blunt the “explode higher again” narrative. Export curbs could also cap the addressable market or force lower-utilization EUV configurations, pressuring near-term numbers and margins.
"SK Hynix and Samsung orders add $12B+ backlog visibility into 2027, positioning ASML to exceed its 2026 revenue guide amid insatiable AI chip demand."
ASML's EUV monopoly shines with SK Hynix's $8B commitment for ~30 systems (at $250-300M/unit) by 2027 and Samsung's reported $4B+ for 20 units at Pyeongtaek P5, adding ~$12B to the existing 38B euro backlog. This locks in revenue beyond 2026's 34-39B euro guide (4-19% growth from 2025's 32.7B), as 12-18 month lead times force pre-booking amid AI chip ramps for NVDA/AMD. EUV mix boosts margins toward 56-60% long-term. Q4's 13.2B orders already strong; April 15 Q1 will confirm momentum. Valuation at ~45x trailing earnings premiums growth, but FCF supports buybacks.
If AI infrastructure capex pauses amid economic slowdown or oversupply risks, these multi-year commitments could slip or cancel, hitting backlog conversions. China export bans (once 20-30% of sales) remain a drag despite allied demand offsets.
"Backlog size is real, but ASP assumptions are unverified and represent the hidden margin cliff nobody's pricing in."
Grok's $250-300M ASP math is plausible but masks a critical gap: neither SK Hynix nor Samsung has publicly disclosed unit counts or pricing. Grok assumes ~30 systems for SK's $8B; if actual ASPs run $200M or unit counts are lower, backlog growth is 20-30% weaker than implied. Claude and ChatGPT flagged timing risk correctly—but the real pressure is ASP erosion if competition (Canon, Nikon in EUV R&D) or customer leverage forces price concessions. At 45x P/E, a 10% ASP miss cascades into 15-20% earnings miss.
"A shift from High-NA EUV to DUV multi-patterning due to yield issues would compress margins and undermine the valuation premium."
Grok, your math on unit counts is pure speculation, and Claude is right to flag the ASP sensitivity. Beyond ASPs, we are ignoring the 'High-NA' transition risk. If Samsung or SK Hynix struggle to yield on High-NA EUV, they won't just defer orders—they will pivot to multi-patterning with older DUV machines to save cash. This shifts ASML’s product mix toward lower-margin, legacy tools, which would destroy the 56-60% gross margin thesis regardless of the total backlog size.
"Backlog confirmation doesn’t guarantee near-term FCF/margin outcomes because acceptance, installation readiness, and supply/capacity constraints can delay cash conversion."
I’ll push back on Grok: even if backlog stays intact, ASML’s key variable is not just “revenue beyond 2026,” it’s free-cash conversion under capacity constraints. Tool revenue can be booked, but cash timing depends on acceptance/installation windows, customer factory readiness, and ASML’s component/supply bottlenecks. With valuation at ~45x, a scenario of durable backlog but slower cash conversion (and margin volatility from mix/launches like High-NA) is enough to disappoint.
"ASML's disclosed ASPs validate unit estimates, and unmentioned service revenue from new installs provides multi-year margin stability."
Claude and Gemini label unit math 'speculation,' but ASML's Q4 earnings confirm EUV ASPs averaging $220-280M (e.g., High Volume Manufacturing mix), making 25-35 units realistic for $8-12B orders per industry norms—no invention. Nobody flags the kicker: these installs lock 10-15% YoY service revenue growth (20% of total sales, 70%+ margins) through 2028, buffering tool cyclicality far better than bears admit. Q1 April 15 will clarify.
Panel Verdict
No ConsensusThe panelists agreed that ASML's EUV monopoly is real and the orders from SK Hynix and Samsung provide excellent backlog visibility into 2027. However, they also highlighted several risks such as execution risk, ASP erosion, competition, and demand destruction that could impact the company's growth and margins.
Excellent backlog visibility into 2027
ASP erosion due to competition or customer leverage