ASML and TSMC Q2 Recap: Time to Chase or Time to Wait?
This article breaks down ASML and TSMC’s Q2 results, focusing on catalysts, margin strength and upcoming headwinds/tailwinds.
Earnings season has officially kicked off! It’s handy that two of the most important companies for the semiconductor industry report at the start, as it sets the tone for expectations for the rest of the industry. I recently discussed my expectations for Q2 results, so feel free to check it out. Without further ado, let’s get into it.
ASML
Financials
I’d expect most readers to be familiar with ASML, but if not, check out my AI industry guide and Q1 article for a refresher/intro. The story so far this year/back end of last year has been a slowdown in bookings, restricted sales to China and a lack of current catalysts. But amidst the gloomy cloud surrounding ASML, it remains a powerful company with strong top and bottom-line growth. I had quite a negative outlook on Q1 results, but feel better with the Q2 figures, although it still wasn’t amazing. However, the nature of ASML’s business is cyclical, and it’s only natural for dips to occur, especially when a new technology is still a work in progress (more on this later).
Total net sales came in at €7.69bn, just above the consensus estimate of €7.66bn. The figure was down 1% QoQ but up 23% YoY, and slightly ahead of the midpoint of management’s €7.4 to €7.9bn guidance range. System sales made up €5.6bn of the total sales, while the Installed Base Management segment contributed €2.1bn. IBM continues to perform well, up 8% YoY, and now accounts for ~27% of total revenue. Gross margin for the quarter came in at 53.7%, down slightly from Q1 but still comfortably ahead of guidance. The figure was helped by a favourable mix, some one-offs, and lower-than-expected tariff impacts. These positives were partially offset by the inclusion of High-NA system revenue, which is still margin dilutive at this stage, given the early nature of the technology and higher production costs. High-NA, or high numerical aperture EUV, is ASML’s next-generation lithography platform designed to enable 2nm and sub-2nm nodes. For context, sub-2nm nodes don’t yet exist, whilst TSMC will soon begin shipping 2nm nodes. The tools are larger, more complex, and not yet fully industrialised, meaning each system is costlier to build and service. As a result, while the margin held up well, it could have been even higher without the impact from High-NA. The rollout of these High-NA machines are the next awaited catalyst for ASML, but management has been very strict and stern on the fact that this technology is still a work in progress and that it’s not expected to be their best-selling product until further down the line.
One of the main takeaways from the Q2 sales figures is the improvement within IBM. The segment brought in €2.1bn this quarter, up 8% YoY, and has already reached around €4.1bn for H1, putting it on pace to comfortably surpass last year’s full-year total of €6.5bn. This strength has been driven by both service revenue and an uptick in upgrade activity, especially related to the NXE:3800E. ASML completed a large number of field upgrades this quarter to bring older systems up to the final 220 wafers per hour configuration, and expects to complete the rest through 2025. IBM remains a key area of consistency and margin support during a time when system sales remain lumpy.
On a side note, if you’re interested in really understanding how these machines work, I’d highly recommend you check out ASML’s website as they have very cool and interactive visual displays of how the machines work.
Net bookings came in at €5.54bn, up from €3.94bn in Q1, marking a clear rebound from earlier in the year. EUV bookings were €2.3bn, with the remainder coming from DUV systems. The mix was dominated by logic, which made up 84% of total bookings this quarter, compared to just 16% for memory. That split isn’t unusual in this part of the cycle, but it’s particularly pronounced right now and reflects what’s happening across the broader industry. Logic refers to chips used in computing and processing tasks, such as CPUs, GPUs, and custom AI accelerators. Demand in this area has remained strong as hyperscalers continue to invest in AI infrastructure. These customers are focused on leading-edge nodes, which continue to drive orders for both EUV and, increasingly, High-NA systems. Memory, on the other hand, includes DRAM and NAND, which are more tied to end-device volumes and cyclical inventory dynamics. After a period of oversupply, memory producers are still cautious, and that is showing up in their capex decisions. The logic-heavy mix supports the view that AI remains the main tailwind for lithography demand. Management also confirmed that the High-NA system shipped this quarter hasn’t yet been booked, so the reported figure slightly understates the true demand backdrop. Overall, bookings were solid and directionally encouraging, though still below the highs seen in 2022.
Another key positive from the latest results has been the capital returns. Historically, investors have shown frustration at the timing of ASML’s share buybacks, often pointing out that purchases peaked during market highs and slowed during corrections. But the company appears to be taking a more consistent and opportunistic approach this cycle. ASML’s ability to steadily return capital speaks volumes about the strength of its business model and balance sheet. The company returned €2.2bn to shareholders this quarter, split between €1.4bn in buybacks and €0.8bn in dividends. That brings total buybacks under the current 2022 to 2025 programme to €5.8bn, with €6.2bn still left to deploy. The pace has been steady, and management has shown no signs of pulling back. With €7.2bn in cash and short-term investments on hand, ASML remains in a position to support both High-NA investments and shareholder distributions. Capital returns rarely grab headlines but quietly support the equity case, especially during periods of operational transition.
On the regional front, China remained a major contributor, accounting for 27% of system sales, flat compared to Q1. The biggest shift came from Taiwan, which rose sharply to 35% of total shipments, up from 16% last quarter. This likely reflects increased demand tied to 2nm and advanced-node investments. South Korea, on the other hand, dropped from 40% to 19%, partly due to front-loaded capex earlier in the year, but also linked to Samsung’s weaker earnings. Samsung reported a sharp decline in operating profit, driven by ongoing challenges in its memory and foundry divisions, which have likely weighed on equipment spend. The regional mix can be volatile quarter to quarter, and therefore, I’d take it with a pinch of salt, but Taiwan’s resurgence and Korea’s pullback tell a clear story about where the momentum currently sits.
Price Action
On the day of the results, shares closed roughly 6% lower and have stayed within that candle to end the week. At this point, I am starting to think about where I can add. Although there isn’t a clear catalyst at this stage and the risks are clear, ASML remains a critical business at the forefront of cutting-edge tech. Beyond this, they are highly profitable, with industry-leading margins, a dominant market position, and long-term demand drivers that remain intact. The transition to 2nm and eventually High-NA nodes will take time, but ASML is uniquely positioned as the only supplier capable of enabling that shift. While near-term volatility in bookings and regional demand may persist, the fundamentals haven’t changed. I’m not in a rush, but with the stock pulling back and sentiment resetting, I’m starting to view this as a potential opportunity to gradually build exposure, as I only currently have a small position (as a % of my portfolio).
Also it’s important to note that ASML’s revenue and performance are seasonally much stronger in the 2nd half of the year, especially Q4. Looking at the seasonality chart below, focusing on the dotted white line, on average since 2015, ASML has provided positive returns mid-october onwards.
The next point at which I’d be looking to add to my position would be in the 615-545ish range. I’m not day trading this so I’m not too fussed about being 100% precise with an entry, although I would of course prefer to be as accurate as possible. This all depends on the momentum though. As we come up to Trump’s tariff deadline with the EU, a deal not being reached could push ASML aggressively lower. If that does happen and the shares tank through my specified area then I will just wait for things to calm down a bit.
Outlook
Looking ahead, ASML reiterated its full-year 2025 guidance, with revenue expected to grow around 15% and gross margin trending towards 52%, implying a stronger second half. But when it comes to 2026, management struck a more cautious tone. On the call, they acknowledged that visibility is limited and that it’s too early to call it a growth year. That doesn’t necessarily imply a downturn, but it does reflect uncertainty around customer capex, particularly as the industry moves from pilot High-NA deployments to real production volumes. Much of 2026 will depend on how quickly demand ramps at the leading edge and how long the memory side of the market stays muted.
There are still a few clear risks to keep in mind. The export landscape remains fluid, with further restrictions on China-bound DUV shipments always a possibility. A meaningful portion of system sales still comes from mature-node demand in China, so any escalation there could weigh on top-line growth. There’s also the tariff tensions between the EU and the US. On the demand side, memory remains soft, and it’s unclear how soon that segment will reaccelerate. There’s also the risk that High-NA adoption takes longer than expected to scale or faces delays on the customer side. Finally, valuation is still elevated relative to historical averages, which leaves little room for disappointment if orders stall again. None of these are new issues, but they continue to hang over the stock and help explain why the market is struggling to find a strong bid for now.
Now, onto one of ASML’s largest customers.
TSMC
Well, well, well. A record net income quarter. TSMC posted Q2 net income of $13.5bn, up 61% YoY and the highest in the company’s history. Revenue came in at $30.1bn, up 17.8% QoQ and 36% YoY, also ahead of expectations. It was a clean beat on all key metrics, and the quality of growth stood out. The 3nm ramp continues to gain traction, now contributing 24% of total wafer revenue, up from 15% in Q1. Advanced nodes, defined as 7nm and below, made up 74% of wafer sales, reinforcing TSMC’s lead at the cutting edge. High-performance computing remains the core growth engine as demand remains strong. HPC revenue rose 14% QoQ and now accounts for 60% of total revenue, up from 45% a year ago. This is being driven by AI infrastructure demand, custom silicon for accelerators, and expanded advanced packaging capacity. Smartphone revenue also held up well, rising 7% QoQ as seasonality kicked in. Other segments like IoT and Automotive were more muted, with Auto down 4% QoQ, reflecting softer demand.
Gross margin came in at 58.6%, slightly down from 58.8% in Q1 but better than expected considering the currency pressure. The New Taiwan dollar has appreciated 12% year-to-date, and that’s a headwind because almost all of TSMC’s revenue is booked in US dollars, while around 75% of its cost base, including labour, utilities, and local procurement, is in NT dollars. So when the NT dollar strengthens, TSMC receives the same amount of USD revenue, but those dollars convert into fewer NT dollars, making local costs effectively more expensive. This mismatch reduces reported profitability, especially at the gross margin level. In Q2 alone, FX shaved about 180 basis points off margin, and management expects another 260 basis point hit in Q3 if the currency trend continues. The underlying business is performing well, but currency translation is creating a drag on reported financials.
TSMC kept full-year capex unchanged at $28 to $32bn, with spending still heavily weighted toward advanced capacity expansion. The Arizona and Kumamoto sites remain on track, although ramp timelines continue to carry cost and execution risk. Management was clear that investment is aligned with long-term structural demand rather than near-term fluctuations, but they also emphasised that spending will remain disciplined. The overseas fabs are a strategic move to localise critical chip production, but they come with a trade-off, which is higher labour, operational costs and less mature supply chains that may run into issues. That’s already showing up in gross margin dilution and will continue to weigh for the next few years as production ramps up. But I feel that this margin dilution is already baked into the price, so unless their margins get hit unexpectedly, I don’t expect lower gross margins to impact share prices. Even so, management reiterated that these investments are necessary to support geopolitical diversification and customer co-location demands.
Price Action
The market reaction was somewhat muted, however, TSMC has made a sharp recovery since the April crash and is trading far into the territory of ATHs. There is no doubt that TSMC is an excellent company and extremely profitable. I think it’ll continue to trade higher for at least the next 2 years, but at one point the growth cycle will slow down. Remember that the semiconductor industry is highly cyclical, and when this initial buildout phase slows, the growth won’t be as explosive. I’m not trying to dampen the mood here but it’s just inevitable. I’m not saying that TSMC will go to 0 when this happens, it simply opens up the opportunity to add more to a position in preparation for the next cycle. But anyway, that is a problem for another day.
Having added TSMC to my portfolio in size at around $160 after the April crash, I have no interest in adding at these highs. I do plan to add to my position, but simply not at these levels. Buying at ATHs isn’t necessarily a bad thing, as if you had bought at ATHs in January, you’d still be in the money now. It’s just a personal preference of mine to not add at ATHs. The ~28x P/E ratio (as of Friday’s close) doesn’t concern me at this stage either, so I would understand if investors were buying TSMC at this stage.
Outlook
TSMC raised full-year revenue growth guidance to 30%, up from the mid-20s previously, reflecting how strong the first half has been and the ongoing pull-through from AI and custom compute demand. However, management also flagged a possible sequential decline in Q4, which suggests they’re leaning cautious heading into year-end. That’s not a sign of weakness, just an honest read of the demand curve after a front-loaded year. Even so, the overall tone remains constructive as customers aren’t pulling back, and major ramps are still underway.
A big part of the story going forward will be 2nm. The company reiterated that it’s on track for risk production in 2025 and volume production in 2026. This will be TSMC’s first node to introduce gate-all-around (GAA) transistor architecture, which is a major shift from FinFET and should bring better power and performance efficiency. Backside power delivery will be introduced in later versions, further boosting performance. If TSMC executes the 2nm rollout the way it has with 3nm, it will cement its edge into the second half of the decade. The customer interest is already there. From what’s been said publicly and hinted at on the call, demand for early 2nm capacity looks strong, especially from hyperscalers and mobile clients. There’s always some risk around yield and cost at these levels, but TSMC’s execution record makes it easier to take those risks in stride. From an investor perspective, this is the bridge to the next growth phase, and I’m interested to see how it plays out.
Final Thoughts
I think ASML is in a more fragile place than TSMC, which is pretty clear to see. The visibility beyond 2025 is murky, with some hesitancy around High-NA adoption timelines, whereas TSMC, on the other hand, feels much more solid. The growth is real, the margins are still strong despite currency and fab pressure, and the roadmap is moving forward exactly how you'd want it to. I think TSMC is the easiest choice when choosing to invest, as with ASML, there are many more headwinds to consider but it’s also important to remember that without ASML, TSMC would not be able to operate, so one shouldn’t underestimate ASML. In terms of the AI buildout, TSMC results have reiterated that the demand is going nowhere, though it’s possible the pace of buildout slows a bit toward the back end of the year.
Thank you for reading this article. I hope you found it insightful! I’m always happy to discuss, so if you have any questions, feel free to reach out. For more articles like this, subscribe (please)! And if you think any friends/family/colleagues would find this article interesting, feel free to send it across.
Disclaimer: This article is for informational purposes only and should not be considered financial or investment advice. The views expressed are my own and based on publicly available information, market trends, and personal analysis. I own both ASML and TSMC shares, which may change at any time. Readers should conduct their own research and consult a financial professional before making any investment decisions.