Latest / Deep Values: Value Investing Deep Dives / TSMC Q3 Financial Deep Value Analysis
Transcript
- 0:00Welcome to the deep dive. We take that dense research
- 0:03stack you have and well, we cut right to the chase, finding the
- 0:06most valuable stuff for you today.
- 0:08We've got a pretty critical mission getting inside the
- 0:11engine room of really the modern tech world.
- 0:15We're talking about Taiwan Semiconductor Manufacturing
- 0:17Company, TSMC. We need to figure out its
- 0:20financial health, it's competitive shield, and maybe
- 0:23most importantly, is this incredible company actually a
- 0:26good value right now? Yeah, and that's vital.
- 0:29Our sources, they're really clear on this.
- 0:31TSMC isn't just, you know, some big tech firm.
- 0:34They actually call it a single point of failure for the whole
- 0:36global tech supply chain. It's that serious?
- 0:39Wow so basically if you use a smartphone, care about AI, or
- 0:42just want to understand global tech flows, you.
- 0:44Absolutely have to get a handle on TSMC.
- 0:46We need to look at their strengths, sure, but also the
- 0:48huge risks they're navigating. And we're definitely armed for
- 0:51this today. Got a detailed Q32025 financial
- 0:54report right here. It covers well everything from
- 0:57the latest earnings all the way to a deep 10 year valuation
- 1:00analysis done sort of Buffett style.
- 1:02Exactly. And our goal here is to take the
- 1:04market temperature of TSMC. It's this phenomenal business
- 1:08totally dominates cutting edge chips, but is it priced for
- 1:11reality or is it priced, you know, for absolute perfection
- 1:14forever? We'll use this rigorous
- 1:16financial analysis to find out. OK, let's jump straight into
- 1:19that engine room then the recent results because these Q2 2025
- 1:24numbers, they are just staggering revenue hit NT $933.8
- 1:29billion, that's about what US $30 billion up almost 39% year
- 1:35over year. 39% and the net income surged over 60% compared
- 1:39to last year. Yeah, these aren't small bumps.
- 1:41This is like monumental growth on what was already a massive
- 1:44company. And what's really interesting is
- 1:46why the source of that acceleration points right to
- 1:48their tech advantage. Management specifically said it
- 1:51was continued robust AI and HPC related demand.
- 1:54HPC, high performance computing. Right.
- 1:56The chips powering the cloud data centers, the whole AI
- 1:58explosion we're seeing. Exactly.
- 2:00And that demand means they're running their most advanced
- 2:02lines, the 5mm and three nanometer nodes, really hard,
- 2:05high utilization. High utilization is probably an
- 2:07understatement. Probably.
- 2:08Yeah, the sources, they have a great breakdown.
- 2:11Those cutting edge 3 nanometer chips, they alone brought in 24%
- 2:15of TSMC's total wafer revenue in Q2.
- 2:18Wow, just one node type is almost 1/4 of their
- 2:21manufacturing income. It's incredible and you see this
- 2:24shift in their overall business mix too.
- 2:26You do, I mean TSMC is a pureplay foundry, right?
- 2:29They make chips under contract. That's the core business.
- 2:32But looking at the revenue sources over the last year or
- 2:34two, HPC that high performance computing is now clearly the
- 2:37main driver. It was 51% of revenue in 2024.
- 2:41It's completely overtaken smartphones, which dropped to
- 2:44what, 35%? Exactly.
- 2:45And this mastery of the AI Adidas center chip market,
- 2:48that's really what's fueling that incredible 58.6% gross
- 2:52margin, they reported. And that margin, that number
- 2:55brings a street to their competitive Moat.
- 2:57How can they command that kind of pricing power?
- 3:00The research points to four key things, sort of interlocking
- 3:03pieces. First, it's technological
- 3:05leadership. They're always first, right?
- 3:06Consistently first to market with the new nodes.
- 3:09They hit 300m commercially back in 2022.
- 3:11They just stay one, maybe 2 steps ahead of everyone else.
- 3:14OK, what's #2? Scale and execution.
- 3:17They have this Gigafab model. Think enormous factories,
- 3:20incredibly efficient places like Fab 18 in Taiwan.
- 3:25They can churn out over 100,000 wafers a month.
- 3:28The economies of scale there are just immense rivals can't really
- 3:31match. It unbelievable scale OK third.
- 3:343rd is something less tangible, but maybe even more important.
- 3:37Customer trust. Since way back in 1987, they've
- 3:40stepped religiously to this no competition pureplay foundry
- 3:44policy. Meaning they promised not to
- 3:46compete with their own customers.
- 3:47Exactly. The founder, Doctor Morris Chang
- 3:50built the whole company on that idea.
- 3:52Our word is our bond. So giants like Apple, NVIDIA,
- 3:55they trust TSMC with the most secret chip designs, knowing
- 3:58TSMC won't suddenly become a rival.
- 4:00That neutrality is gold. Makes sense.
- 4:02The final piece of the MO. The big one, capital
- 4:05requirements, just building a new leading edge fab costs well
- 4:08it starts at $10 billion, often more like 20 billion.
- 4:10Plus TSMC has the cash to spend big continues.
- 4:13Their CapEx was nearly $36 billion back in 2022.
- 4:16So that kind of spending just keeps series new competitors
- 4:19out. It sounds like an absolutely
- 4:20unbreachable fortress. It does, doesn't it?
- 4:23But you know, even the strongest fortresses can have
- 4:25vulnerabilities. If we connect this to the bigger
- 4:28picture, you start to see potential cracks.
- 4:31They're incredibly resilient, sure, but they're not totally
- 4:34immune to market cycles. Like what?
- 4:37Well, short term stuff, the semiconductor market is famously
- 4:40cyclical, right? Demand goes up and down, you get
- 4:43inventory corrections like they saw in early 2023 and just
- 4:47normal seasonality. Q1 is usually slower, Q3 is the
- 4:50peak. OK.
- 4:51Those are sort of normal business risks, but the sources
- 4:54highlighted bigger structural dangers, didn't they?
- 4:57Absolutely. The biggest long term risk, the
- 4:59one that keeps policymakers and analysts up at night, is
- 5:02geopolitical risk. It's all about their heavy
- 5:04concentration in Taiwan. Western governments openly call
- 5:08TSMC's dominance a single point of failure for the world's tech
- 5:11supply. And that's not just talk, right?
- 5:13It actually impacts big investors.
- 5:15I mean, Warren Buffett. Exactly.
- 5:17Think about Buffett buys great companies holds forever.
- 5:20That's his whole thing. But the sources concern after
- 5:23buying into TSMC, he sold relatively quickly and the main
- 5:26reason he cited geopolitical risk, even though he called it
- 5:30one of the best managed companies.
- 5:32That tells you something. Yeah, it really underlines how
- 5:34serious that non financial risk is perceived to be.
- 5:38OK. What about other pressures,
- 5:39Competition costs? Right, while TSMC has the lead
- 5:43now, Samsung and Intel are throwing billions, literally
- 5:46billions trying to catch up. They're both targeting 2
- 5:49nanometer production by 2025, maybe 2026.
- 5:52It's this constant, incredibly expensive race.
- 5:55And are there cost pressures internally too?
- 5:57I remember reading something about the US fabs.
- 5:59Yes, that's a key point from the research.
- 6:01TSMC themselves noted their new Arizona fab will have
- 6:04considerably higher operating costs than their fabs back in
- 6:07Taiwan. Maintaining those amazing
- 6:09margins when you're forced to build expensive, maybe less
- 6:12efficient capacity overseas, that's a huge, huge challenge
- 6:15going forward. Definitely something to watch,
- 6:17though it is worth mentioning their management seems very
- 6:19stable. Doctor Cece Way now chairman and
- 6:23CEO. He's seen as continuing that
- 6:25disciplined execution. First culture from the founder.
- 6:29That continuity helps, and they are unusually transparent,
- 6:32providing those monthly revenue reports.
- 6:34It helps build trust even when discussing these frankly massive
- 6:38risks. OK.
- 6:38So we've established it's an amazing business, but with some
- 6:41pretty significant risks. Let's pivot now to the investor
- 6:44perspective. Let's run this fabulous
- 6:46enterprise, as Buffett might say, through his legendary
- 6:49checklist. Does it qualify as a wonderful
- 6:52company at a fair price? All right, let's do it.
- 6:54Looking at the wonderful company part, using those 10 year
- 6:57metrics from the report, TSMC scores really, really well.
- 7:00Let's start with profitability and efficiency.
- 7:02Gross margin comfortably above 40%, usually in the 45% to
- 7:06almost 60% range. That's pure pricing power.
- 7:10OK. Impressive margins.
- 7:11What about how efficiently they use capital like return on
- 7:14equity, Roe, especially for business that spends so much?
- 7:17Their 10 year average Roe is around 27% and now remember
- 7:22Buffett often looks for maybe 15% as a threshold.
- 7:26TSMC is almost double that. It shows outstanding use of
- 7:29shareholder money. Double the threshold, OK.
- 7:31And the balance sheet you mentioned, it's a fortress even
- 7:33with that huge CapEx spend. It absolutely is minimal
- 7:36reliance on debt. Their debt to equity ratio was
- 7:39super low like .25 in 2024. And interest coverage, how much
- 7:43operating income covers interest payments?
- 7:45It was over 125 times. They generate enormous cash
- 7:48compared to their debt. Wow.
- 7:50OK. And my personal favorite, the
- 7:51one that really shows management skill, capital allocation, how
- 7:55good are they at reinvesting the profits they don't pay out?
- 7:58This number is genuinely remarkable.
- 8:00Their return on retained earnings over the last decade is
- 8:02about 32%. 32% so for every dollar they kept in the
- 8:06business. They generated 32 cents of
- 8:08additional annual earnings. That kind of return on
- 8:10reinvestment for a company this huge is almost unheard of.
- 8:13It proves fantastic execution. OK, so it definitely passes the
- 8:17wonderful business test with flying colors.
- 8:19Now here's the tricky part for value investors.
- 8:22Where does it fail the strict Buffett criteria?
- 8:25Right. It generally fails on two key
- 8:27areas that are really fundamental to Buffett's classic
- 8:30approach. First, plain old valuation.
- 8:33TSMC just rarely, if ever, trades at the kind of low price
- 8:37to earnings ratio. Historically, it's averaged
- 8:39around 21X or low price to book ratio, currently about 7.4 X
- 8:43that value investors hunt for. You usually don't get to buy
- 8:46this quality on the cheap unless something really bad seems
- 8:49imminent. OK, valuation's the first
- 8:51hurdle. What's the second failure point?
- 8:53Yeah, mentioned it relates to the business model itself.
- 8:55Exactly. Capital intensity.
- 8:57It fails the test looking at depreciation relative to gross
- 9:00profit, and 2023 depreciation ate up something like 45% of
- 9:04their gross profit. 45% and Buffett prefers.
- 9:07Much, much lower. He loves asset light businesses
- 9:10where that number is tiny, maybe low single digits, Well, TSMC
- 9:14makes massive gross profits. That 45% figure shows that
- 9:17nearly half of it has to get plowed right back in just to
- 9:20maintain the equipment and stay ahead technologically.
- 9:23It's like a very profitable treadmill, but it's still a
- 9:25treadmill. That really crystallizes the
- 9:27dilemma, doesn't it? Fantastic business, incredible
- 9:29returns, but always hungry for more capital and rarely cheap.
- 9:33So let's bring it home with the valuation What is the current
- 9:35market price? Up near $300 lately, actually
- 9:38imply let's look at that discounted cash flow.
- 9:41The DCF analysis from the research.
- 9:43OK. The DCF models are super
- 9:45sensitive to the growth rate you plug in for future free cash
- 9:49flow FCF. The current market price is
- 9:51essentially a bet on a specific future growth pass.
- 9:54Right. So let's walk through the
- 9:55scenarios they laid out. What if we're conservative?
- 9:57Say only 5% FCF growth per year going forward, which seems
- 10:02honestly really low given the AI boom.
- 10:04If you assume only 5% growth, the calculated intrinsic value
- 10:08per share is only around 85 dollars miles below where it
- 10:11trades today. OK, $85 is clearly not what the
- 10:13market thinks. What about a more nominal base
- 10:17case? Maybe 15% growth, closer to
- 10:19their historical average? Right.
- 10:21The nominal base case using 15% annual FCF growth gives you an
- 10:25intrinsic value of about $175 per ADR. Now if you're a
- 10:29cautious investor, especially with the geopolitical stuff and
- 10:32you want a 25% margin of safety. Which Buffett certainly would.
- 10:35Then your buy price based on that $175 value would be around
- 10:39$137 and you can see why Berkshire buying near $70 back
- 10:43in 2022 look like such a huge cushion, OK.
- 10:46But 137 dollars is still way below the $300 level.
- 10:50So to justify today's price, you have to get really optimistic,
- 10:53right? All the way to the optimistic
- 10:54case. This scenario assumes extremely
- 10:56fast, sustained 25% FCF growth with every year for a whole
- 11:00decade. Only that kind of aggressive
- 11:02forecast gets you an intrinsic value around $360.00 or so.
- 11:05That's even in the ballpark of the current market price.
- 11:07So, bottom line for the listener, what does this DCF
- 11:09exercise tell us? It tells us that today's market
- 11:12price is basically baking in that very optimistic 25% growth
- 11:16for a decade scenario. Yes, the company's fundamentals
- 11:19are strong. They deserve a premium,
- 11:20absolutely. But at this price, there's
- 11:23almost no margin of safety left, no room to absorb a hiccup and
- 11:26demand tougher competition or, crucially, any kind of
- 11:30geopolitical shock. The market seems to have priced
- 11:33in perfection. OK, let's wrap this up.
- 11:35Final summary for you listening. TSMC's financials are incredibly
- 11:38resilient. It's tech mode is wide,
- 11:40management is strong. It's a foundational piece of the
- 11:43global economy, but the current stock price reflects near
- 11:46universal belief in continued absolute breakneck growth,
- 11:50leaving very little room for error or unexpected trouble.
- 11:52Exactly. We've seen it's this high
- 11:54return, brilliantly executed company, but that comes packaged
- 11:57with being inherently high cost, needing constant reinvestment
- 12:01and facing these huge external risks, mainly geopolitics around
- 12:04Taiwan. Which leaves us with maybe one
- 12:07final provocative thought for you to consider.
- 12:09We know the massive capital spending required is a barrier
- 12:12to entry. That's clear.
- 12:13But here's a question. Can TSMC replicate its
- 12:16incredible execution and cost efficiency from Taiwan in its
- 12:19new, much more expensive overseas fabs like the one in
- 12:22Arizona over the next 10 years? If they can't quite match that
- 12:26Titanese magic abroad, those phenomenal margins we talked
- 12:29about today could face some serious long term pressure and
- 12:32that would make today's price of perfection look even riskier
- 12:34down the road.