Earnings season is fast approaching. As always, I’ll provide in-depth coverage of all the companies I include here at Stocktrades Premium.
However, this week works out perfectly because we had two companies featured here (and core to the AI space) reporting earnings.
So instead of just reiterating and providing commentary on these two companies’ earnings reports, I want to intertwine them into a complete newsletter that speaks on AI demand.
If there is one thing we can take from both companies’ earnings reports, it is that the AI spending boom is nowhere near being over.
Let’s dig right into it. In terms of portfolio moves, I won’t get into that too much. I made a routine weekly purchase of Foundational Stock Brookfield Corporation (TSE:BN).
Bull List Stocks TSMC and ASML Highlight The AI Spending Boom is Not Over
TSMC and ASML just reported results that show chip demand isn’t cooling off. In fact, it might be accelerating.
ASML brought in a record 13 billion euros in new bookings last quarter, more than double the previous year. The more shocking thing is that these bookings were more than double what analysts had expected. On the year, the company logged more than 28B euro in new bookings, a 50% jump from the previous year.
Bookings represent the value of new sales contracts signed within a specific period. From there, they get added to the backlog, which is a cumulative amount of bookings.
Because ASML’s bookings had declined for 2 straight years, there were fears that the AI cycle had slowed down. In reality, it just looks like ASML’s customers were waiting for a bit of reassurance from customers on their end before being willing to spend more.
TSMC’s capital spending plans for 2026 remain at the high end of guidance. Why are they spending so much? Because they simply cannot keep up with demand. Clients want more chips, and they want them now.
Taiwan Semiconductor mentioned capital expenditures in the next 3 quarters will be higher than the previous 3 years, combined.
Everyone likes to talk about the GPU companies (the NVIDIAs of the world). However, I’ve specifically targeted what I call the “picks and shovels” plays of AI in TSMC and ASML.
Why? Because the supply chain underneath the GPUs is what matters. ASML makes the lithography equipment that prints the world’s most advanced chips, and TSMC manufactures those chips at scale.
When ASML’s CEO says customers are more confident about AI sustainability, it means these companies see the buildout as being sustainable enough that they’re going to order machines and chips years in advance.
This is a good thing for the two major AI stocks featured here, ASML and TSMC.
Blowout Earnings from TSMC and ASML
TSMC reported results that beat expectations across the board.
If you were to draw a conclusion from this quarter regarding TSMC that didn’t necessarily mention any actual numbers, it would be that the company has told you to stop worrying about whether or not we are in an AI bubble and instead worry about where the chips are going to come from to satisfy the demand.
Revenue for the year jumped 36% to $122 billion, earnings increased by 45%, and operating margins hit a staggering 50.5%. The shift to artificial intelligence is now blatantly obvious, as High Performance Computing (HPC) now makes up over half of the business. Pre-pandemic, this accounted for around 35% of revenue.
The chart below gives you a good idea as to how fast this segment is expanding.
The most telling part of the conference call was management’s admission that they are “nervous” about the massive investment step-up in terms of capital expenditures, but only because they feel the heavy burden of being the reason the world is bottlenecked when it comes to AI development.
ASML’s Story is Equally Compelling
For members who quietly accumulated ASML since its addition to the Bull List at the $700 level, it was a reasonably frustrating year or so. However, the recent price action goes to show you how patiently accumulating and holding high-quality companies can pay off.
In just 6 months, the stock price has doubled.
As I mentioned, ASML sells the lithography machines that companies like TSMC need to build the best chips for these AI companies. Strong bookings mean companies like TSMC are racing to add capacity, not pulling back. That’s a direct indicator that orders will remain strong as the industry expands to meet AI needs.
The company’s 2030 production targets are around 600 DUV machines and 90 EUV machines. If you look to the chart below, the company produced half of that in 2025. That is some large-scale growth.
The company also recognized revenue on the first commercial High-NA EUV (EXE:5200B) system delivered to Intel. Why is this important? Well, these new machines come at double the cost of their old ones (the yellow line in the chart above).
Bottlenecks For The Entire Industry are in High End Capacity
This area is going to require some explanation. I will make it as simple as possible.
Think of a traditional chip like a single, flat LEGO brick. For years, the goal was just to make that single brick more complex. But AI chips are too big and powerful to be one brick. They are a collection of different parts that need to talk to each other at exceptionally fast speeds.
CoWoS (Chip-on-Wafer-on-Substrate) is essentially the glue and high-speed wiring that creates these. Without this specific packaging, you just have a pile of expensive parts that aren’t interconnected.
This is the bottleneck slowing down the entire AI industry. TSMC’s factories are busy, but their packaging lines (the CoWoS part) are the actual choke point. They simply cannot produce fast enough. Nvidia, a single chip producer, has 60% of worldwide packaging capacity booked for 2026. Mind-blowing numbers.
This is where a portion of TSMC’s capital expenditures are going. They plan to double packaging capacity by the end of 2026.
The story for ASML is now a proven success. There should be extensive demand for these machines for the foreseeable future. The real story here is TSMC.
If TSMC’s new facility buildouts hit even a small delay, the entire AI world slows down. The next two years aren’t about who can design the best chip, but who has a reserved seat at TSMC’s packaging table as they continue to expand packaging capability.
So what does this all mean as shareholders? This creates a powerful hedge: ASML secures the long-term infrastructure, while TSMC manages the short-term delivery. What this ensures is that as long as the world needs GPUs, they must pass through your portfolio’s gatekeepers.
AI Spending Is Not Just NVIDIA Anymore
For the longest time, this was one of the most fascinating charts in investing. In pretty much the snap of a finger, Nvidia turned from a company providing GPUs for gaming systems into an AI giant, supplying data center compute for hyperscalers. Data center revenue jumped 1600% in just 3 years.
However, the AI infrastructure buildout has moved well beyond NVIDIA’s GPUs and data centers. In fact, some would call this old news. This space is moving ridiculously fast, there is zero doubt.
What started as a rush for compute chips has evolved into a full-scale infrastructure overhaul. NVIDIA remains central to AI, don’t get me wrong, but the buildout around it is widening, fast.
Power supplies, liquid-cooling systems, contract assembly, grid expansion, and so much more. In areas you’d simply never think were possible, either.
An example? Bull List Stock Exchange Income Corporation (TSE:EIF) has practically doubled over the last year.
It just so happens the medivac and remote area flyer also has a segment that produces composite matting that is widely used in remote construction and power grid expansion. They’re so busy in this department they’ve had to fast-track a new $60M facility to create more.
Back to the tech side, TSMC is fabricating chips for multiple AI customers, not just NVIDIA. Hyperscalers like Amazon, Google, and Microsoft are designing their own equipment, and they’re all relying on TSMC to get them done. That diversification gives TSMC steadier demand even if one customer pulls back. This probably wouldn’t have been the case 3 years ago. ASML benefits from the same dynamic.
The spending isn’t just about training models anymore. Workloads are expanding, which requires different chip architectures and more memory capacity. Big Tech is committing hundreds of billions to AI infrastructure, and those dollars are filtering down to a wide variety of companies.
Geopolitical Tailwinds are a Massive Incentive
National security is now at the center of everything. We’re in the midst of a new space race here. Except obviously it has nothing to do with space, and everything to do with AI.
Countries want top-of-the-line access to everything, and they do not want even a shred of possibility that that access is disrupted.
TSMC’s Arizona investments are a pretty clear signal of this. The company has pledged to build six fabs in Phoenix. About 30% of its future 2nm capacity will come out of the U.S.
The U.S. CHIPS Act is providing subsidies and loans to make a lot of this happen. Without those subsidies, TSMC wouldn’t bother building top-of-the-line nodes outside Taiwan.
Why? Production costs in Arizona can be 10-30% higher than in Taiwan. There are more labor regulations in the US and Canada, people require higher wages, materials are required to be shipped long distances, and overall it’s just more expensive to operate a business here than in Taiwan.
If TSMC had planned to do this without the intervention of the US government, a lot of investors would probably see this as money ill-spent. However, when you have the US government behind you, anything is possible.
Geopolitical priorities are now steering economic choices. Export controls, tariffs, and industrial policy are all being utilized. The U.S. started restricting exports of advanced chips and lithography tools to China back in 2022. This is why ASML can only sell China DUV machines. These are cheaper pieces of equipment, technologically inferior to the EUV.
Without EUVs, China can’t make the most advanced AI chips. The U.S. teamed up with the Netherlands and Japan to block China’s access.
Now, AI infrastructure spending is spreading out. TSMC’s Arizona buildout isn’t just chips. It’s got advanced packaging as well, the bottleneck I had explained above. That helps build a local supply chain for companies like Apple, AMD, and Nvidia.
Overall, there are no deeper pockets than governments who want top-end technology at their disposal, which is ultimately bullish for both TSMC and ASML.
Why the Market May Still Be Underestimating the AI Buildout
The reaction to TSMC’s addition to the Bull List in the $290 range was lukewarm at best. I did get a lot of replies asking why I would highlight the company when it was at all-time highs and up so much.
However, in my opinion, I really think a lot of investors are underestimating how long this spending cycle will last. When TSMC commits to $52–56 billion in capex for 2026, they are not doing so guessing that CAPEX will result in sales. They’re doing so because clients are pounding at the door for them to increase capacity.
Valuation multiples, in my opinion, haven’t caught up to the reality of sustained demand. TSMC’s growth expectations keep getting revised upward.
Of course, there are real risks. Geopolitical tensions, supply chain bottlenecks, and a chance that the AI investment boom goes wrong on the opposite end of the spectrum.
While the demand for the physical infrastructure (chips and fabs) is currently through the roof, the risk lies in whether the companies at the other end of the spectrum, the software and service providers, can actually turn all this compute power into a profitable business.
Until we figure that out, the picks and shovels (TSMC, ASML) will continue to get paid.
