Latest / Deep Values: Value Investing Deep Dives / 【13F】Dan Bin's Q2 Billion Dollar AI Pivot
Transcript
- 0:00So I want you to imagine a master architect right?
- 0:02Like a a guy who has spent his entire 30 year career building
- 0:07these very austere classical stone monuments.
- 0:11OK, I'm picturing it. He is just utterly obsessed with
- 0:14the durability, predictability tradition.
- 0:18He builds structures that are designed to look exactly the
- 0:20same a century from now, right? Build to last forever.
- 0:23Exactly. But then one day you look at his
- 0:25newest blueprint and he's suddenly building this like 100
- 0:29story neon skyscraper made entirely of experimental gravity
- 0:34defying glass. I mean, honestly, I would
- 0:37immediately assume someone hacked his computer right?
- 0:39Or perhaps that he sees some fundamental shift in the laws of
- 0:44physics that the rest of the structural engineering world is
- 0:46just completely missed. And that is the exact puzzle
- 0:49we're decoding today in the financial world.
- 0:51We're looking at what happens when one of the most famous
- 0:54disciplined traditional value investors, someone the legendary
- 0:57for holding the most boring, durable consumer staples, just
- 1:01suddenly dumps his entire US portfolio.
- 1:04Just completely clears it out. Yeah, and dumps it right into
- 1:07the most aggressive, cutting edge artificial intelligence and
- 1:11tech hardware stocks on the planet.
- 1:13It's wild. It really forces a complete re
- 1:16evaluation of how conservative institutional capital is
- 1:20navigating this whole technological transition.
- 1:22It does, because it challenges the very definition of what a
- 1:25safe asset actually is. So today we are doing a deep
- 1:28dive into a really fascinating document.
- 1:31We've got the Q12026 Form 13F filing for the Oriental Harbor
- 1:36Investment Master Fund. Managed by the renowned Chinese
- 1:38fund manager Dan Binn. Exactly.
- 1:41And we're going to explore his $1.133 billion U.S. equity
- 1:44portfolio. Look at how he has completely
- 1:46redefined Warren Buffett's classic investment playbook for
- 1:50the AI edge. And you know what his extreme
- 1:53bets mean for the future of technology.
- 1:55There's a lot to unpack there. There really is.
- 1:57But before we look at the actual holdings, I want to make sure
- 2:00you, the listener, have the tools to follow this kind of
- 2:02extreme institutional movement yourself.
- 2:05So, to access more advanced researches, follow Gurus's
- 2:08latest movement and leverage our latest Deep Values agents.
- 2:12Welcome to download the Deep Values APP from the App Store.
- 2:16It's highly recommended if you want to track this stuff.
- 2:18Yeah. And if you want to verify the
- 2:19macro environment driving these trades, like to track stock
- 2:23market temperature, welcome to download the iOS app Buffett
- 2:26Indicator from the Apple App Store or Bubble Watch from
- 2:30Google Play. Perfect, and just setting the
- 2:33parameters for today's deep dive.
- 2:34I need to add that all analysis is based on publicly available
- 2:38sources and maybe subject to revisions or differing
- 2:40interpretations. All contents from this channel
- 2:44is for research and entertainment purposes only and
- 2:46does not constitute any investment or financial advice
- 2:49or service. So please conduct your own due
- 2:52diligence before making investment or financial
- 2:54decisions. Absolutely.
- 2:55OK, so let's lay the groundwork here, because to really grasp
- 2:59the magnitude of what Oriental Harbor is buying right now, you
- 3:03have to understand who is doing the buying.
- 3:05Right, the context is everything.
- 3:06Yeah, Dan Bin isn't some Silicon Valley venture capitalist who's
- 3:10just chasing the next shiny object.
- 3:13Historically, he is this fiercely disciplined proponent
- 3:16of Chinese domestic equities. Specifically, durable consumer
- 3:20staples. Yes, he is most famous for
- 3:23holding massive massive positions in Kwai Chao Mutai,
- 3:26which is a premium baiji distiller.
- 3:29Which is really a textbook franchise investment.
- 3:31I mean, Beiji relies on deeply ingrained, predictable consumer
- 3:36behavior, right? It has an incredibly high
- 3:38margins, brand loyalty that spans literally generations.
- 3:41People buy the liquor, they drink it, they go buy more.
- 3:43Exactly. It's a book Coca-Cola playbook,
- 3:45and that Graham and Dodd Buffett style philosophy yielded a 26.3%
- 3:50annualized return since his offshore fund launched in March
- 3:532016. Which is incredible.
- 3:55Right, And across their publicly tracked strategies, they
- 3:57recently saw a three-year return of over 140%.
- 4:01So he built this fortress by betting on human habits that
- 4:04just never change. Which is exactly what makes this
- 4:07new 13 F filing look almost like a clerical error.
- 4:10Literally. Because that $1.133 billion US
- 4:14portfolio has been almost 100% repositioned into the artificial
- 4:19intelligence value chain. And the concentration is just
- 4:22staggering to me. We are talking about 97.56% of
- 4:26the assets sitting in just the top 10 holdings.
- 4:28Right, he's dropped the classical stone monuments and
- 4:31just gone all in on the neon skyscraper.
- 4:33But the interesting part is he claims he is using the exact
- 4:36same architectural principles. Wait, really?
- 4:38How does he justify that? Well, he genuinely believes he's
- 4:41applying the exact same theory. He hasn't abandoned the
- 4:44philosophy of value investing, he's basically evolved it for an
- 4:48era of technological hegemony. OK, explain that a bit more.
- 4:51So traditional value investing looks for low historical
- 4:54valuation multiples, high tangible book value, and
- 4:57predictable consumer demands. Right.
- 4:58Yeah, the boring stuff. Exactly.
- 5:01Dan Ben has taken that framework and adapted it.
- 5:04He's looking at modern computing platforms and hardware
- 5:07bottlenecks as the new unbeatable monopolies.
- 5:11So he's looking for the modern toll bridges.
- 5:13Precisely. But the macro view here is
- 5:15really critical. He doesn't see generative AI as
- 5:19just like a software feature update or a passing theme, the
- 5:23sources note. He views it as a foundational
- 5:25industrial transition that will last 10 to 15 years.
- 5:29He compares it directly to the mobile Internet expansion You
- 5:32know, which drove markets for 13 years.
- 5:34Oh wow, or like the PC hardware build out before that.
- 5:37Exactly. And recognizing that timeframe
- 5:39is really how he justifies his departure from traditional
- 5:42conservatism. He operates on this concept
- 5:45called omission risk. Omission risk, OK, So what does
- 5:47that mean in practice? Dan Bin argues that missing out
- 5:50on the dominant economic beneficiaries of a paradigm
- 5:53shift is far more dangerous than the risk of buying them at a
- 5:55cyclical peak. Wait, so he's arguing that
- 5:58avoiding high growth equity simply because they look
- 6:00expensive on a, you know, traditional spreadsheet that
- 6:04practically guarantees your portfolio's purchasing power
- 6:06will erode over the next decade? That's his thesis, Yeah.
- 6:10If you set out the Internet boom because Amazon's price earnings
- 6:13ratio was just too high in 2012, you lost against inflation,
- 6:17right? So traditionalists fear the
- 6:19bubble, but Dan Bin fears the omission.
- 6:22OK, but how does he manage the downside?
- 6:24The mechanism he relies on for safety in these expensive trades
- 6:27is what he calls valuation. Multiple digestion, right?
- 6:30He isolates companies with absolute pricing power and
- 6:33exceptionally high returns on invested capital.
- 6:35So the thesis is that these companies will grow their free
- 6:38cash flows so aggressively over for the next few years that
- 6:41their future earnings capacity will just compress those sky
- 6:45high multiples down to very safe traditional levels.
- 6:48Exactly. The growth digests the multiple.
- 6:50OK. I have to push back on that
- 6:51though, because that assumes there won't be a single
- 6:54macroeconomic hiccup or a recession or, I don't know, a
- 6:57regulatory crackdown in the next 5 years to derail those
- 7:01earnings. It is definitely an aggressive
- 7:03stance. It's essentially pricing
- 7:05imperfection. I mean, buying tomorrow's cash
- 7:07flows at today's astronomical prices sounds a lot, a lot less
- 7:10like value investing and more like just hoping the music never
- 7:14stops. No, you're right, it does carry
- 7:16massive execution risk. But Dan Bin views these specific
- 7:21AI platforms as insurmountable monopolies.
- 7:24Like actual toll bridges, right? If a company controls the
- 7:28absolute only bridge into the future of digital commerce and
- 7:31enterprise productivity, they can essentially dictate their
- 7:34own revenues if the mode is deep enough.
- 7:36The current price tag is just an optical illusion created by
- 7:39backward looking accounting. OK, let's see how that theory
- 7:42translates into actual capital allocation, because there was a
- 7:45massive, undeniable strategic rotation this quarter.
- 7:48Oriental Harbor completely liquidated their Microsoft
- 7:51position. They sold all 193,527 shares for
- 7:57roughly $94,000,000. That's a 100% exit from one of
- 8:02the most dominant enterprise software companies on earth.
- 8:05And that capital didn't just go to cash, either.
- 8:07It was redeployed into Alphabet, Google's parent company.
- 8:10Yeah, they purchased $79.67 million in the Class A voting
- 8:14shares, which adds to their already massive $345.63 million
- 8:20holding in the Class C shares. And they also hold a leveraged
- 8:23ETF on Alphabet, right? Right.
- 8:24They do. So when you aggregate all of
- 8:26that, Alphabet now accounts for an astonishing 39.80% of his
- 8:30entire portfolio. It's massive.
- 8:32That is $450.95 million. Alphabet is officially the
- 8:37central pillar of his entire thesis, surpassing NVIDIA.
- 8:40Think about the conviction required for that.
- 8:42It's almost unimaginable for a traditional fund.
- 8:44Yeah, I mean, if you are sitting at home listening to this and
- 8:46looking at your own retirement account, imagine allocating
- 8:49almost 40% of your entire net worth to a single search engine
- 8:52company. Well, calling it a search engine
- 8:54company severely understates the structural advantage Dan Bin
- 8:58sees here. You have to look at Alphabet
- 9:00through that toll bridge lens. OK, the core Google search
- 9:03engine processes over 90% of global Internet queries, right?
- 9:08So that creates this impenetrable data network
- 9:11effect. But the real mechanism
- 9:14protecting that toll bridge is their vertical integration in
- 9:17the AI space. Let.
- 9:18Me test a theory on why this Moat is wider than Microsoft's.
- 9:21Because Microsoft has to buy chips from NVIDIA to run open
- 9:24AI's models and then they have to sell that software to
- 9:27enterprises, right? But Alphabet doesn't have to do
- 9:29that. They design their own custom
- 9:32silicon, right? The tensor processing units.
- 9:34Yes, the TP use. And they train their own own
- 9:36proprietary Gemini models on their own world scale data.
- 9:40And then they push those models directly into a distribution
- 9:42network of billions of users through Android and YouTube.
- 9:45Exactly because they own the hardware, the software, and the
- 9:48distribution. They don't have to pay a
- 9:49middleman at any step of the process.
- 9:51So they capture the entire economic rent of the AI
- 9:54transaction. That is the exact mechanism.
- 9:57They are immune to the hardware margin compression that
- 10:00threatens other software companies.
- 10:02Plus, under their current management, they've implemented
- 10:04serious cost discipline and massive share repurchases.
- 10:08So to Dan Bin, Alphabet isn't a speculative tech play.
- 10:11Not at all. It is a classic high quality
- 10:14compounding franchise generating massive free cash flow, and it's
- 10:19trading at a multiple he believes is artificially
- 10:21depressed by antitrust headlines.
- 10:23OK, so he sells Microsoft's enterprise software to double
- 10:26down on Alphabet's vertically integrated toll bridge.
- 10:30But the 13 F reveals he didn't just stay in software.
- 10:33He moved aggressively down the technology stack, away from
- 10:36consumer applications and into the cold, hard physical
- 10:40bottlenecks of the AI revolution.
- 10:42He went straight for the physical infrastructure layer,
- 10:44basically the picks and shovels of the gold rush.
- 10:47Yeah, so let's talk about NVIDIA.
- 10:48It sits as his second largest holding at 19.79% of the
- 10:52portfolio. Now, the casual narrative is
- 10:55just that NVIDIA makes the fastest chips, right?
- 10:57But the sources highlight that the real economic Moat
- 10:59protecting this position isn't the silicon itself.
- 11:02Right, it's the software ecosystem.
- 11:04Exactly. NVIDIA spent nearly two decades
- 11:06establishing CUDA as the foundational programming
- 11:10language for parallel computing and AI development.
- 11:13The mechanism here is massive developer switching costs.
- 11:17Oh, because they'd have to rewrite everything.
- 11:19Exactly. If an enterprise wants to move
- 11:21away from NVIDIA hardware and buy a cheaper chip from a
- 11:24competitor, they have to rewrite millions of lines of code and
- 11:27retrain all their engineers. That friction is almost
- 11:30impossibly expensive. It is.
- 11:32Add in Nvlink, their hardware interconnect system, and
- 11:35Nvidia's selling entire data center scale supercomputers, not
- 11:39just loose components. But Danbin clearly realized that
- 11:42even NVIDIA relies on someone else to physically print their
- 11:45designs, because in this quarter he initiated a brand new $70.87
- 11:50million position in TSMC, the Taiwan Semiconductor
- 11:54Manufacturing Company. Which is 6.26% of the portfolio.
- 11:57Yeah, and this is a direct bet on the ultimate physical
- 12:00manufacturing bottleneck. Because TSMC manufacturers over
- 12:0390% of the world's leading edge silicon chips.
- 12:06Apple, NVIDIA, AMD, none of them actually make chips.
- 12:10They just draw the blueprints. They all physically rely on
- 12:13TSMC's fabrication plants. So the structural pricing power
- 12:16there must be infinite because TSMC requires, what, 30 to $40
- 12:20billion annually in capital expenditures just to maintain
- 12:23their technological edge. Nobody can afford to compete
- 12:26with them. It's an insurmountable barrier
- 12:28to entry. Yeah, if TSMC decides to raise
- 12:31the price of manufacturing a wafer by 15%, NVIDIA and Apple
- 12:35just have to pay it because there is literally nowhere else
- 12:37on earth to get those chips printed at that scale and
- 12:40precision. And they just pass those
- 12:42inflationary costs directly onto the largest tech monopolies in
- 12:45the world. Unbeliev.
- 12:47And to complete this hardware trifecta, the fund initiated a
- 12:49new $24.67 million position in Micron Technology.
- 12:54OK, they manufacture memory chips, right?
- 12:56Specifically, High Bandwidth Memory or HBM 3 and HBM 4, which
- 13:00are physically required to feed data into NVIDIA processors.
- 13:04It prevents bandwidth bottlenecks during AI training.
- 13:07OK, this is where we had a massive contradiction in
- 13:09traditional value theory because Warren Buffett historically
- 13:13despises memory chip businesses. Oh.
- 13:15Absolutely hates them. Because memory is a commodity.
- 13:18It goes through brutal, predictable boom and bust cycles
- 13:20with massive supply gluts and terrifying price crashes.
- 13:24So why would a discipline value investor touch Micron?
- 13:29Because Dan Bin is calculating that the structural undersupply
- 13:32in AI hardware temporarily overwrites that traditional
- 13:35cyclicality. Temporarily.
- 13:37Right. The physical limits on how fast
- 13:39TSMC can package these chips and how fast Micron can yield high
- 13:43bandwidth memory creates this multi year window where supply
- 13:48simply cannot meet the hyperscaler demand.
- 13:51I see. He sees a period of absolute
- 13:53pricing power that will generate enough cash to justify the
- 13:55position before the cycle eventually turns.
- 13:58So the AI revolution really is a gold rush.
- 14:00He's not buying the prospectors, and he's not even buying the
- 14:03shovel makers. He is buying the only factory on
- 14:06earth that possesses the intellectual property to forge
- 14:10the steel for the shovels. That is a perfect way to
- 14:12visualize the hardware bottleneck thesis.
- 14:15I mean, I can follow the logic for Alphabets vertical
- 14:17integration and I understand the structural pricing power of
- 14:20TSMC. But if we dig deeper into this
- 14:2213 F filing, we find allocations that completely break the
- 14:26traditional rules of capital preservation, right?
- 14:29If he's perfectly willing to redefine a semiconductor foundry
- 14:32as a value stock, where else is he bending the rules of
- 14:35traditional finance to capture yield?
- 14:37Well, we have to talk about the leveraged ETF derivatives.
- 14:39Exactly. The portfolio holds 7.22% in
- 14:43TQQQ which is a three times daily leverage leveraged NASDAQ
- 14:46ETF and it holds 2.26% in G GLLA2 times leveraged Alphabet
- 14:51ETF. Right.
- 14:52We are talking about nearly 10% of a billion dollar value fund
- 14:56sitting in highly volatile daily leverage derivatives.
- 15:00Help me untack the mechanics of this because holding these long
- 15:02term usually destroys capital through beta slippage, right?
- 15:05Yes, from a strict value perspective, holding these
- 15:08derivatives is financial heresy. Because these ETFs rebalance
- 15:12their leverage daily, they suffer from mathematical decay
- 15:15and anything other than a perfectly straight line up.
- 15:17Let me break down how punishing that math actually is for anyone
- 15:20listening who hasn't traded these.
- 15:21Let's say you buy a normal $100 stock and it drops 10% today.
- 15:26It's worth $90.00, right? If it recovers and goes up 10%
- 15:29tomorrow, 10% of 90 is only 9. So your stock is now worth $99.
- 15:35You've lost a dollar just to the market.
- 15:37Shop exactly When you apply three times leverage to that
- 15:40daily oscillation. That daily chop completely
- 15:44cannibalizes your capital. In a flat sideways market, a
- 15:47leveraged ETF will bleed out and eventually go to 0.
- 15:51Because the mathematical decay actually scales with the square
- 15:54of the leverage factor. Wow.
- 15:56So a 3X leverage product incurs A volatility drag 9 times larger
- 16:00than the underlying index. Dan Bin understands this
- 16:03continuous time math perfectly. He is systematically accepting
- 16:06that known quantifiable decay to essentially strap a multiplier
- 16:10onto his highest conviction. Ideas.
- 16:12He knows this car leaks oil, but he's only driving it in a
- 16:15straight line for 1/4 mile drag race.
- 16:18That's a great analogy. He believes we are in a
- 16:20unidirectional bull market for tech and he's using the ET FS as
- 16:24a tactical speculative overlay to capture extreme beta.
- 16:28It is entirely tactical and this willingness to explore or non
- 16:32traditional financial architecture.
- 16:34It actually extends to his .27% stake in Circle, which is the
- 16:38issuer of the USDC stablecoin. Yeah, I saw that.
- 16:41It's only a $3.03 million position, but the mechanics
- 16:44behind it are fascinating. They really are.
- 16:47I mean a tiny crypto position for a value investor.
- 16:49What is the fundamental mechanism driving that?
- 16:52Well, Circle operates what is essentially an insurance float
- 16:56model. So when institutions buy their
- 16:58stablecoin, they hand Circle Fiat U.S. dollars.
- 17:01Serbel takes those billions of dollars in cash, invest them in
- 17:04short duration U.S. Treasury bills, and keeps all
- 17:06the yield. Sneaky right?
- 17:08They earn the risk free rate on a massive reserve while paying
- 17:11zero interest to the people actually holding the digital
- 17:14tokens. It's the classic financial float
- 17:17business model. Literally the exact same model
- 17:20Warren Buffett used to build Berkshire Hathaway with GEICO
- 17:23insurance premiums, but applied to digital assets.
- 17:26They are monetizing the plumbing of the crypto ecosystem.
- 17:29Yes, it provides asymmetric upside to to the institutional
- 17:32adoption of blockchain without taking any direct exposure to
- 17:36volatile token prices. OK, let's zoom out and look at
- 17:39the systemic vulnerabilities here, because if you are
- 17:41listening to this and looking at your own portfolio, you might be
- 17:44wondering if your safe index funds are actually exposing you
- 17:47to this exact same concentrated risk.
- 17:50You can't concentrate a billion dollars this aggressively
- 17:53without creating massive architectural flaws.
- 17:56That's the trade off. So what happens if this highly
- 17:58leveraged, highly concentrated bet goes wrong?
- 18:00Well, concentration builds tremendous wealth, but it
- 18:04completely isolates risk. This portfolio is incredibly
- 18:07vulnerable to the CapEx cycle. OK.
- 18:08The CapEx cycle, yeah. Alpha built NVIDIA, Meta, TSMC
- 18:12and Micron make up over 74% of this fund.
- 18:16The entire operation is tied to AI enterprise capital
- 18:19expenditure. Wait, so the CapEx cycle risk
- 18:22means that if Microsoft, Google, or Amazon suddenly realize they
- 18:25aren't making enough subscription revenue from AI
- 18:28software to justify buying more hardware, they just stop buying?
- 18:32Exactly. And if they stop buying, this
- 18:34entire House of Cards collapses from the top down.
- 18:37The hyperscalers are currently spending 10s of billions on
- 18:40infrastructure. If they pause that spending, the
- 18:43compute stack crashes simultaneously, the hardware
- 18:46makers lose their orders, the foundries lose their volume,
- 18:49memory price is crater, and the multiples just collapse back to
- 18:53reality. Because the entire supply chain
- 18:55is feeding the exact same beast. Exactly.
- 18:58And beyond the cyclical risk, there is a terrifying geographic
- 19:01reality to this portfolio. Right.
- 19:03The geopolitical single point of failure by holding nearly 20% in
- 19:07NVIDIA and over 6% in TSMC, the fund is entirely dependent on
- 19:11the physical continuity of the Western Pacific Semiconductor
- 19:14corridor. Any military friction or
- 19:16blockade in the Taiwan Strait immediately halts the global
- 19:19hardware supply. That creates systemic portfolio
- 19:22wide drawdowns that you simply cannot hedge against.
- 19:25No, you can't. So Dandin's portfolio is a
- 19:27master class in high conviction allocation, but it's balancing
- 19:30on a razor thin geographical and cyclical wire.
- 19:34It really requires an iron stomach and just a fundamental
- 19:37belief that the major powers will maintain the status quo out
- 19:42of mutual economic necessity. To bring us all together, you
- 19:45know, we've explored how Dan been a legendary consumer value
- 19:48investor, pivoted from predictable Chinese liquor into
- 19:51becoming a technology hardware maximalist.
- 19:54It's a historic pivot. We've seen how he uses the
- 19:56concept of omission risk to justify paying high multiples
- 20:00for tomorrow's cash flows. We've watched him consolidate
- 20:03almost 40% of his wealth into Alphabets vertically integrated
- 20:07toll bridge. And aggressively buy up the
- 20:09physical bottlenecks of the AI revolution with TSMC and Micron.
- 20:13And all while utilizing mathematically decaying
- 20:16leveraged ETFs to juice his returns.
- 20:18Exactly. And reviewing these
- 20:20institutional 13 of filings isn't about blindly copying
- 20:23their trades. You know.
- 20:24The objective is to dismantle the mental models of brilliant
- 20:27capital allocators and determine what risk actually means to you
- 20:30in your own financial framework. Right, it's about understanding
- 20:33the why behind the what. Which leaves us with a final
- 20:36thought to chew on If every traditional value investor
- 20:39eventually redefines massive, high multiple technology
- 20:42companies as safe, structural toll bridges, does the
- 20:46definition of value investing lose its meaning entirely?
- 20:49That's a fascinating question. Are we just creating a new
- 20:52hidden bubble where everyone convinces themselves that
- 20:55absolute tech monopolies are the new boring consumer staples?
- 21:00I guess the market will deliver the answer to that question over
- 21:02the next decade. Maybe that classical architect
- 21:05building the neon skyscraper hasn't lost its is mine.
- 21:08Maybe he just realized that in a world where the ground is
- 21:11shifting beneath our feet, stone monuments are the most dangerous
- 21:14things you can build. Very true.
- 21:17Thank you for joining us on this deep dive.
- 21:19Keep exploring, keep questioning, and we'll catch you
- 21:21next time.