Here’s what to watch on October 15—and why ASML’s earnings release could be the next green light for the AI-chip rally.
The AI trade is having another “prove-it” moment. On October 15, ASML Holding (ASML)—the bottleneck supplier whose extreme ultraviolet (EUV) and deep ultraviolet (DUV) systems help produce the cutting-edge chips fueling the data center buildout—reports Q3 results. Because the company sits at the choke point of advanced chip production, its orders, backlog, and guidance tend to foreshadow demand (and its sustainability). If ASML confirms re-accelerating demand tied to AI compute buildouts, it supports the ongoing melt-up across chipmakers and AI platforms.
We just got a taste of that momentum: Advanced Micro Devices (AMD) shares exploded (+28% on the day as I write this) after unveiling a multi-year deal to supply OpenAI with next-gen Instinct GPUs and granting OpenAI warrants for up to a 10% stake—fresh evidence that hyperscale AI capex is not easing. If ASML helps confirm that story with solid bookings, investors get another piece of evidence that the investments hitting the headlines are flowing further down the food chain.
[text_ad]
If Nvidia (NVDA) (and Taiwan Semi) sells picks and shovels, ASML sells the forges that help build them.
The Setup Going into Earnings
ASML’s July quarter (Q2 2025) delivered €7.7B in sales (it’s based in the Netherlands, so all its financials are measured in euros) and a 53.7% gross margin, alongside €5.5B in net bookings (about €2.3B of that EUV). Management guided Q3 sales to €7.4–€7.9B with 50–52% gross margin and reiterated ~15% full-year 2025 revenue growth with ~52% margins.
October’s earnings release should be instructive for Wall Street, as it will show whether ASML has been under- (or over-) estimating demand. It’s also another opportunity for management to set the table heading into the end of the year and into 2026.
Why it matters: ASML is effectively a referendum on whether AI demand is spilling from GPUs into sustained foundry and memory investments. In other words, are these (often conditional) multi-billion-dollar (and sometimes multi-hundred-billion-dollar) deals between hyperscalers and data center builders fostering the kind of longer-term commitments from chipmakers that Wall Street would expect to see if the buildout bull market still has a long runway?
Four Confirming (or Cautionary) Indicators
1) Bookings velocity, especially EUV.
A step-up in EUV bookings would signal that the tidal wave of data center capacity bookings is translating to purchase orders today. ASML’s Q2 already showed EUV bookings at €2.3B; any meaningful improvement here would signal conviction in the achievability and sustainability of the data center buildout. Conversely, if demand stalls out or declines, it would show that the companies actually building these GPUs and high-performance chips may not be expanding operations to support the multi-year demand that these latest deals imply.
The 10-gigawatt deal between Nvidia (NVDA) and OpenAI, for instance, is a $100 billion deal that will play out over five years and will require “millions of GPUs.” (Appears to be 6 million based on the math below.)
For some context, in the Q1 2026 earnings call in July, Nvidia’s management indicated that major cloud players were “deploying nearly 72,000 GPUs per week.”
The GB200 (Grace Blackwell chipset) NVL72 server rack uses 72 Blackwell chips and draws 120 kilowatts (at the low end of estimates) of power.
10 gigawatts of data center demand translates to 83,000 racks (each drawing 120 kilowatts) and 6 million Blackwell chips (72 chips per rack). That’s over a year and a half of deployment (at 72,000 GPUs per week) for the OpenAI deal alone. Someone has to produce those chips.
2) Installed Base strength.
ASML’s Installed Base Management was over €2B last quarter; resilience here suggests fabs are pushing assets harder for AI-driven utilization, which supports tool vendors and indirectly corroborates GPU-driven demand.
3) 2026 visibility.
Management has been constructive on 2025 but has kept 2026 somewhat closer to the vest. Confident projections for 2026 could be the strongest confirmation for bulls arguing this is a multi-year, not one-off, cycle. On the flip side, if management remains cautious with forward guidance, it raises questions about sustainability. It’s not a death knell if guidance is soft or uncertain, but it’s a yellow flag.
4) Demand validates newsflow.
The AMD–OpenAI deal (and Nvidia’s ongoing wins) are top-down demand signals: More AI clusters require more cutting-edge wafers. If ASML’s order book reflects the top-down demand signals, it serves to support the narrative being painted by the headlines.
Why ASML Is the Bellwether for Confirming the Bull Thesis
Unlike chip designers, ASML sells the picks and shovels for the most advanced nodes—there is no EUV alternative. That monopoly status means its backlog and lead times are a clean read on how aggressively the world is building advanced capacity. When ASML is busy, the entire compute supply chain is busy.
What to Look for on October 15
- Bookings at or above Q2, with robust EUV mix and positive commentary.
- Resilient Service revenue (Installed Base growth), indicating high utilization and upgrades.
- Tone on 2026 edging from “uncertain” toward “constructive.”
A report that checks most of those boxes likely stretches the AI rally beyond GPU headlines. Conversely, if bookings and outlook underwhelm, it raises questions about the industry and could easily drive rotation away from chip names until the next AI catalyst lands.
ASML’s October 15 report won’t tell the full story, but it’s an important gut-check for the AI rally and an opportunity to see whether the potentially circular, eye-catching deals between chip firms, data center builders, hyperscalers and mega-cap tech companies are playing out in the real world and not just the headlines.
If ASML’s earnings blow expectations out of the water, it answers a lot of questions that have been bubbling up in the minds of investors, analysts, and Wall Street.
[author_ad]
