Taiwan Semi builds chipsets for other companies like Nvidia, AMD and Qualcomm. It does so in its highly expensive, highly complex chip fabrication plants. These are called “fabs” for short and arguably provide one of the deepest moats in public markets.

Taiwan Semi beat revenue estimates by 1% & beat revenue guidance by 3.1%. Its 16.5% 3-year revenue compounded annual growth rate (CAGR) compares to 9.8% last quarter and 13.5% two quarters ago.

For the full year, TSM sees “close to 30%” Y/Y revenue growth. This is a raise vs. “slightly better than mid-20%” growth guidance offered last quarter. It also changed annual CapEx guidance from $31 billion to “slightly higher” than $30 billion.

TSM trades for 23x forward earnings. EPS is expected to grow by 30% in each of the next two years.

Advanced technology represented 69% of wafer revenue vs. 59% Y/Y. This was powered by 11% Q/Q growth within its high performance computing (HPC) bucket. HPC is now 51% of overall sales vs. 52% Q/Q and 42% Y/Y. Internet of Things (IoT) growth also materially boosted advanced technology results. Specifically, that bucket enjoyed 35% Q/Q expansion and a rise from 6% to 7% of total wafer revenue.

For non-AI demand, leadership called the environment stable. For AI specifically, it sees 2024 processor revenue contribution tripling to 15% of total. Nvidia’s boom is TSM’s boom too. When asked about durability of GenAI demand, leadership reminded us that it works with all of the companies making AI chips, frequently communicates with them, and sees demand remaining very strong. When asked about GenAI return on investment (ROI) for clients, CEO C.C. Wei offered tangible examples of the technology driving considerable efficiency and cost gain internally. Those improvements are transferable to other companies.

“We continue to observe extremely robust AI-related demand from our customers throughout the second half of 2024. This is leading to increasing overall capacity utilization rates for our leading-edge 3-nanometer and 5-nanometer process technologies. One key customer said the demand right now is insane and that it's just the beginning... it will continue for many years.”

CEO C.C. Wei

This was the most impressive piece of the report to me, and more context is needed. As a reminder, ramping up new nano-chip technology always leads to gross margin dilution. It takes time for economies of scale to eventually power margin parity between new products and older ones. Last quarter, TSM also told us that shifting some N5 production capacity to N3 would be a 100-200 bps GPM headwind for 2024. This change and new tech scaling perfectly coincided with electricity inflation in Taiwan. All of these GPM challenges were amplified by expansion to higher-cost regions for another 200-300 bps of GPM impact.

With all of this context, 350 bps of GPM expansion and the large beat were certainly notable highlights. Where did the outperformance come from? Per the team, higher capacity utilization (CEO quote above fares well for that continuing) and successful “cost improvement efforts” were the two primary sources. The headwinds described above have not improved at all, these positive items just overcame the obstacles. The strong Q4 GPM guide was due to the same factors.

On the CapEx side, budget allocations remain the same as TSM invests ahead of future demand. 75% of earmarked CapEx is for advanced process technologies and 15% is for specialty technologies. Complementary Metal-Oxide-Semiconductor (CMOS) is a type of specialty technology that TSM offers. It’s an IC used in logic processes to help assign functions. 

The remaining 10% of the CapEx budget is for the advanced packaging, testing and mass marketing bucket; this makes up a little more than 10% of overall revenue. Last quarter, TSM raised its total addressable market (TAM) from $115 billion to $230 billion for the foundry business. That was based on including packaging and testing in the equation. Packaging includes storing and integrating chips with thermal protection, maintenance and connectivity tools too. Traditional foundry services make up the actual creation of an IC or chip for a customer. TSM had a 28% market share of this extended TAM definition in 2023, which it saw rising in 2024 as of last quarter. No update this quarter. Based on the quality of this team, I would take that as a reiteration; they would tell us if previous forecasts were worsening.

All of these projects make TSM less dependent on Taiwan for running its business. Considering rather heated geopolitical tensions between Taiwan and China, that’s very important. In Arizona, coordinated government support led to “strong progress” over the last several months. CEO C.C. Wei continues to envision 3 fabs in the state, with clean rooms 2x the size of its Taiwan locations. The first factory actually began producing N4 chips with “highly satisfactory results and a very good yield” in April. It expects production there to scale by early 2025. The second and third factories will “utilize more advanced technologies” and will enter volume production in 2028 and 2030, respectively.

Government support in Japan was also called strong. It has a specialty tech factory gearing up for volume production this quarter. Another fab is planned there to support HPC demand, with production hopefully starting by the end of 2027. In Europe, construction on the Germany fab has begun, with volume production scheduled for the end of 2027 as well.

Leadership acknowledged the “fragmented globalization environment” and elevated costs in its newer geographies. That will mean lower margins in its new plants vs. its old ones for now. At the same time, TSM knows it can rely on its “manufacturing technology leadership and its large manufacturing base.” With these strengths, it fully expects to remain the “most efficient and most cost-effective manufacturer” in all of the regions where it conducts business. So lower margin than its most mature facilities… but still higher margin than everyone else.

We didn’t hear much about its newest N2, N2P and N16 (1.6 nanometer) roadmaps. I thought I’d include last quarter’s commentary on these important projects to keep them fresh on investor minds. These will be future growth drivers. Last quarter, N2 had been progressing well on performance, density and yield gains vs. the N3 platform. N2P was in very early stages of development, but already with incremental performance gains vs. N2. N2P is scheduled for volume production during the 2nd half of 2026, along with its newest N16 technology. N16 is for HPC use cases. This comes with a separate power rail with backside power delivery (first of its kind). Traditionally, power is transferred to transistors from the front of the chip. By moving it to the back, energy efficiency and heat management are both improved. N16 unsurprisingly offers considerable performance gains vs. N2P. It’s always iterating.

This was an elite quarter and guide from an elite team and company. It is optimally positioned to capture the large GenAI opportunity through relationships with Nvidia and AMD. This business is masterfully executing with no slowdown currently in view. The only thing that makes me worry here is geopolitics. That’s really the risk, as it has been for years. Congratulations to shareholders on the wonderful results.

Netflix needs no introduction.

Netflix beat revenue estimates by 0.6% & beat guidance by 1.0%. Foreign exchange neutral (FXN) growth outperformed by a larger margin, meaning Netflix overcame stronger-than-expected currency headwinds to deliver the revenue beat. It also beat 4.0 million net subscriber add estimates by 1.1 million or 27.5%.

Netflix raised $1.8 billion in debt during the quarter in its first investment grade bond offering. This $1.8 billion in cash will pay down more expensive debt on the balance sheet.

Engagement health is vital for Netflix, just like for other consumer subscription businesses. This is what motivates lower churn, more word-of-mouth growth and more pricing power. It’s why Spotify is expanding into audiobooks and podcasts; it’s why Duolingo is expanding into music and math; it’s why Netflix is expanding into live sports and gaming. When consumers use you more frequently, they stick around for longer and deliver more value. This quarter, per Nielsen, Netflix secured more top 10 titles (ranked by hours viewed) than all other streamers combined. To be fair, it doesn’t have a legacy media arm to cannibalize a portion of streaming demand, but this is undeniably impressive.

Household viewership per day remained above two hours despite paid sharing restrictions being implemented. Restrictions mean fewer people per account and so represents a headwind here. While household viewership did fall a bit Y/Y due to this headwind, when accounting for the unique item, growth remains firmly positive. For evidence, owner households (excluding previous sharers) saw average hours viewed rise Y/Y.

Like last quarter, leadership talked up an opportunity to grow screen time market share beyond the roughly 10% clip in mature markets. It sees the depth and breadth of its library as unmatched, while sports and gaming further enrich it. This is also why it has no interest in bundling with other streamers, like we’ve seen from its competition. 

Netflix raised prices in a few European markets and Japan recently, with intentions to hike pricing in Spain and Italy this quarter. NFLX has arguably its best content slate in years lined up for 2025 as we move away from strikes and its volume recovers. New seasons of Stranger Things, Wednesday, Squid Game, NFL games and several more hit titles coming will likely lead to more price hikes in 2025. Leadership wouldn’t confirm that but did hint at it.

Netflix also phased out its cheapest ad-free plan (Basic plan) in the USA and France as planned. These two markets join the UK and Canada in terms of not having access to that tier. So far, the change is going well and it will soon make the same move in Brazil. Per CFO Spencer Neumann, the Basic plan was creating more complexity than value for consumers. This was addition by subtraction, and also works to push more viewers to its ad-supported tier, as it pushes for critical mass there. Speaking of which…

Netflix has two key priorities when it comes to advertising: building consumer scale to be attractive to ad buyers and giving those buyers more of what they need to target and measure performance.

First, on consumer scale, things are progressing quite nicely. Ad-supported plans rose 35% Q/Q (still a small base but great progress), with revenue roughly doubling Y/Y. Furthermore, 50% of all new subscribers in eligible markets went with ad-supported packages vs. 45% Q/Q. It’s on track to “reach critical subscriber scale for advertisers in all 12 ad countries next year.” Without the needed base of eyeballs, it doesn’t matter how good Netflix’s content is or how good its new tools are. Eyeballs must come first, and they are. The progress here led to it netting a 150% Y/Y rise in commitments during this year’s upfront ad sales. This met expectations.

To really optimize this business and properly drive monetization, it will readily tell you it has work to do on debuting more buyer tools. That leads us to priority number 2, which is seeing slower progress than priority 1. As a result, NFLX ad impression growth is outpacing impression sales and ARM is suffering a bit. That will continue while this immature business grows up. In terms of milestones to focus on, the debut of its in-house first party ad tech platform is a key step in its journey. That will start in Canada and be released across the globe next year.

Another big piece of catering to advertisers is adding integrations with their preferred platforms. This makes it easier for these buyers to access inventory with their trusted campaign managers. Knowing this, it recently integrated with The Trade Desk and Google’s DV 360 programmatic buy-side platforms. These partnerships are going well so far. Leadership was asked if they thought using these vendors would be temporary as it tries to build a walled garden. All it said is that it views the relationships as very positive. More demand means higher impression value and higher impression booking rates.

Overall, Netflix continues to expect ad revenue to double in 2025 and to grow into a material part of the business by year’s end.

Netflix wants to do a better job with converting traffic to paid users. It thinks there’s a lot of optimizing to be done, and that work is underway. Heading into 2025, it will launch its new TV homepage, which “lays the foundation” for more rapid interface split testing down the road. Netflix feels they needed this change to stay ahead of the curve. It will also upgrade its signup page in a bid to juice conversion rates further. These are changes that can be made to drive growth without spending any more money on marketing.

Some were expecting Netflix to entertain changing its pay-up-front model for content creators. That’s not the case. It’s allowing some who prefer different compensation methods to have more of what they want, but the de facto method is still paying up-front. This, per the team, takes the pressure off of creators and frees them to focus on their actual work. It also helps the company attract world-class talent.

Another fantastic quarter with nothing to pick at and much to praise. This is arguably the highest quality consumer software subscription on the planet aside from maybe Amazon and Costco. It continues to outperform its sector by a wide margin, continues to successfully expand into advertising, continues to broaden its content offering and continues to win. The 2025 guidance points to more of the same, while the planned content slate is ripe for delivering more subscriber growth upside. As a subscriber, I can’t wait. What a turnaround from this leadership team over the last 2-3 years and what another great showing from them.