Latest / Deep Values: Value Investing Deep Dives / Value Investing in AI Era
Transcript
- 0:00Picture this that you know, just imagine the scene on one side of
- 0:03the room. You have your traditional value
- 0:05investor. Right, just sitting there
- 0:06quietly with a massive stack of 10 KS.
- 0:09Exactly. Yeah, totally ignoring the daily
- 0:12market panic, just sizing up capital allocation in terms of
- 0:16like decades. Playing the very long game.
- 0:18Get the longest game, but then on the other side of the room
- 0:21you have the bleeding edge of the AI revolution.
- 0:23Which is just, well, chaos. Total chaos.
- 0:27It's this hyper speed, unpredictable algorithmic force
- 0:31that's reshaping the global economy by the millisecond.
- 0:34And when those two fundamentally opposite worlds collide.
- 0:37Right. All our assumptions just
- 0:39fracture. Everything we think we know
- 0:41about the future of capital, labor, human ingenuity, it all
- 0:44gets thrown out the window. It really forces us to ask if
- 0:47patient capital can even survive in an era of microwave
- 0:52technology. Yeah, and whether those old
- 0:54psychological framework from the 20th century actually hold up
- 0:58when a machine can process the entire Internet and the time it
- 1:00takes you to blink. It's a huge question.
- 1:03It is, and that is exactly our mission.
- 1:05For today's Deep Dive, we're exploring how to navigate the AI
- 1:09era using the ultimate value Investing playbook.
- 1:12And we have some incredible source material for.
- 1:14That we really do. We're looking at a fascinating
- 1:16interview with Chang Jin of Himalaya Capital.
- 1:19But real quick, before we jump into how he decodes all this
- 1:22noise, I need to do a little housekeeping.
- 1:23Go for it. So free listening to access more
- 1:26advanced researches, applications and our latest
- 1:29published personal intelligent multi agent deep value
- 1:32intelligence. Welcome to explore and join the
- 1:35Values dot AI community. Also to track total market
- 1:38temperature. Welcome to search and download
- 1:41iOS app Buffett indicator from Apple App Store.
- 1:44And one more thing. Disclaimer All analysis is based
- 1:47on publicly available sources and may be subject to revisions
- 1:50or differing interpretations. All contents from this channel
- 1:53to the research and entertainment purposes only and
- 1:56does not constitute any investment or financial advice.
- 1:59Conduct your own due diligence before making investment or
- 2:02financial decisions. Don't forget to like this video.
- 2:05Much appreciated. OK, with that out of the way,
- 2:09let's get into Himalaya Capital. Yeah, let's talk about Chang
- 2:12Jin, because to really get this, you have to understand why his
- 2:15perspective carries so much weight.
- 2:16Right, he's not just some random fund manager.
- 2:19Far from it, Himalaya Capital is actually the only outside firm
- 2:22that the late Charlie Munger's family trusted to manage their
- 2:25own money. Which is just, I mean, that's
- 2:27massive. Munger was famously skeptical.
- 2:29Famously skeptical, ruthlessly rational, and, you know,
- 2:32fiercely protective of his capital.
- 2:34Yeah, handing over the keys to an outside manager is an
- 2:37endorsement that practically defies his entire public
- 2:40persona. Exactly.
- 2:41And that level of trust, it doesn't happen through slick
- 2:44PowerPoint presentations, right? It happens through absolute
- 2:48philosophical alignment. So what does that alignment
- 2:50actually look like in practice? Well, Munger really ingrained 2
- 2:54foundational concepts into Cheng Jin's approach.
- 2:58The first one is this idea of multidisciplinary mental models.
- 3:02OK, let's unpack this multidisciplinary meaning, like
- 3:05bringing in other fields. Right.
- 3:07Munger argued that analyzing a market purely through financial
- 3:11metrics is essentially intellectual negligence.
- 3:15Wow, intellectual negligence. That's a strong phrase.
- 3:18It is, but he believed it. He said to understand the true
- 3:22underlying reality of an asset, you have to pull frameworks from
- 3:25thermodynamics, evolutionary biology, psychology, sociology.
- 3:28So you have to emphasize all these disciplines because the
- 3:31real world doesn't just neatly fit into an Excel spreadsheet
- 3:34exactly. It's like, it's like having
- 3:36different lenses on a camera, right?
- 3:37If you only have a macro lens you get a great close up, but
- 3:40you totally miss the wider landscape.
- 3:42That's a perfect analogy. You need all the lenses to see
- 3:45the whole picture, but a lot of fund managers claim to read
- 3:48widely and, you know, think laterally.
- 3:51Right, Everyone says they do that.
- 3:53So what took this from just a fun intellectual exercise to a
- 3:57philosophy rigid enough to protect multigenerational
- 4:00wealth? That brings us to Munger's
- 4:02second pillar, the psychology of human misjudgment.
- 4:06The famous cognitive biases, yeah.
- 4:09He identified a highly specific matrix of about 25 cognitive
- 4:14biases, things like anchoring, social proof, commitment bias.
- 4:17The stuff that's just evolutionarily hardwired into
- 4:20our brains. Right.
- 4:21And these biases actively fight against our ability to allocate
- 4:25capital rationally. So if you're managing money,
- 4:27other people's life savings, you can't afford to fall for them.
- 4:30Exactly. In a fiduciary setting, yielding
- 4:33to those biases isn't just a bad trade, it's a fundamental breach
- 4:36of trust. So absolute rationality becomes
- 4:39like a moral obligation. Yes.
- 4:41Not just a strategy, a moral obligation.
- 4:43That raises a flag for me though.
- 4:45I mean, if you treat rationality, is this cold
- 4:47absolute moral imperative? How do you even function in
- 4:50normal society? Because society is inherently
- 4:53irrational. Yeah, exactly.
- 4:55Was Munger just operating as this walking algorithm in Los
- 4:58Angeles, viewing every conversation as some calculated
- 5:02exchange of utility? You'd think so, but it's
- 5:05actually really counterintuitive.
- 5:07His ruthless logic led him to extreme pro social behavior.
- 5:11Wait, really? How does that work?
- 5:13He applied those exact same mental models to human
- 5:16relationships, and he concluded that the most efficient,
- 5:19rational way to navigate life is to build what he called a spider
- 5:23web of deserve trust. A spider web of deserve trust.
- 5:27Yeah, because he recognized that friction, you know, legal
- 5:32battles, broken contracts, toxic partnerships.
- 5:35That friction destroys wealth and time.
- 5:37So fighting with people is literally just a bad investment.
- 5:40Exactly. So the logical solution is to
- 5:42operate with such structural reciprocity that people
- 5:45genuinely want you to succeed. So it wasn't altruism just for
- 5:48the sake of feeling good, it was almost a calculated compounding
- 5:51of social capital. Precisely like he donated
- 5:54heavily to institutions like Stanford and he sat on local
- 5:57hospital and school boards in LA.
- 5:59But I'm guessing he didn't just write a check to get his name,
- 6:01they want a plaque. No, never.
- 6:03He analyzed what those institutions structurally
- 6:06lacked, and he prioritized giving them the specific value
- 6:09they actually needed. So he was investing in his
- 6:11community ecosystem with the exact same rigor he'd apply to
- 6:15like, holding a stock. Exactly.
- 6:17You build the web thread by thread and eventually it catches
- 6:20everything you need. OK, so let's take that whole
- 6:22framework, the multidisciplinary models, the defense against
- 6:26bias, the patient compounding, and let's drag it into the
- 6:29current moment. The AI era.
- 6:31Yes, when a firm like Himalaya looks at the AI revolution, how
- 6:36do they even categorize it? Because the dominant narrative
- 6:39right now is just oh, AI is Web 3 point O, or the next iteration
- 6:42of the Internet. Right.
- 6:43And Chang Jin argues that categorizing AI is just another
- 6:47Internet is a massive analytical error.
- 6:50Really. Why?
- 6:51Well, the Internet was undeniably A paradigm shift, but
- 6:54its primary function was spatial.
- 6:56Meaning it just moved things. Yeah, it moved offline commerce
- 6:59and communication into a digital space, but AI operates on an
- 7:04entirely different vector. The framework he uses compares
- 7:08AI directly to the Industrial Revolution.
- 7:10OK, here's where it gets really interesting.
- 7:12But but wait, I have to push back there.
- 7:14Sure. The Industrial Revolution took
- 7:16the better part of the century to fundamentally alter the
- 7:18global labor force. Right, it was slow.
- 7:21And we watched ChatGPT hit 100 million users in what, 8 weeks?
- 7:26Yeah, the scale and the speed just don't seem remotely
- 7:29comparable. The speed is definitely
- 7:31different, but the mechanism of the transformation is identical.
- 7:34How so? Think about before the steam
- 7:36engine, human output was strictly capped by physical
- 7:39muscle. Right, you can only plow a field
- 7:41as fast as your horse could. Walk exactly The steam engine
- 7:44exponentially amplified physical capacity.
- 7:47It decoupled labor from biological limits.
- 7:49OK, I'll follow that. So what the steam engine did for
- 7:52muscle, AI is currently doing for the human brain.
- 7:54It's an exponential amplification of cognitive
- 7:57labor. Oh wow.
- 7:58So it's a cognitive steam engine?
- 7:59Yes, and the reason it's moving at lightning speed unlike the
- 8:03actual steam engine, is that AI doesn't have to lay a physical
- 8:07track. Because the Internet already did
- 8:08that. Precisely.
- 8:10It's layering on top of a global infrastructure that the Internet
- 8:14already digitized. The digital tracks are laid, the
- 8:17cognitive locomotive is just running over them at lightspeed.
- 8:20OK, if that's true true and we now have this cognitive steam
- 8:23engine that can process millions of data points, read every 10K
- 8:27ever filed, and analyze supply chains instantly, doesn't the
- 8:31human value investor become completely obsolete?
- 8:34You'd think so, but no. But if value investing relies on
- 8:38finding mispriced assets, an AI can parse the data infinitely
- 8:42faster than a human. Why wouldn't it just inherently
- 8:45be better at this? Because value investing isn't a
- 8:47data retrieval exercise. It's an exercise in probability,
- 8:51ambiguity and human endurance. Human endurance.
- 8:54Connect the dots for me there. Well, large language models.
- 8:56They thrive on highly structured tasks with binary outcome.
- 8:59Like writing Python code. Right.
- 9:01Or translating languages or structuring databases.
- 9:05If you ask an AI to write a script, it either compiles or it
- 9:09crashes. It's pass or fail.
- 9:11But the stock market isn't binary.
- 9:12Exactly. Capital markets are entirely non
- 9:15binary. A decision that is
- 9:17mathematically correct today might look disastrous for five
- 9:20years before it finally pays off.
- 9:21What does that actually look like in practice though, where
- 9:24the human beats the machine on endurance?
- 9:27Look at Himalaya's position in the Chinese manufacturing giant
- 9:31BYD. Oh, the EV company.
- 9:33Yeah, they have held a massive position in that company for
- 9:37over 2 decades and in that time frame, BYD stock price has
- 9:41crashed by more than 50% on 6 separate occasions.
- 9:44Wait, wait. Six drawdowns of half your
- 9:46capital. Yes, six times.
- 9:48That is, I mean, that's nauseating.
- 9:50It is. Now imagine a machine learning
- 9:52algorithm programmed for risk adjusted returns and volatility
- 9:55reduction. It would trigger a stop loss and
- 9:58liquidate that position during the very first 50% drop.
- 10:01Or the second, At best, the algorithm looks at the tape and
- 10:03just sees catastrophic risk. But a human investor doesn't.
- 10:07The human value investor utilizing Munger's mental models
- 10:11looks past the ticker. They're out there analyzing the
- 10:14underlying battery chemistry, the vertical integration of the
- 10:17supply chain, the relentless grit of the management team.
- 10:21They see the inevitable global transition away from internal
- 10:24combustion engines. Exactly.
- 10:26It takes stubborn, contrarian human conviction to watch your
- 10:31portfolio have in value and say you know what the marking is
- 10:34Panicking, but the factory floor tells me we are right.
- 10:38And an AI just cannot replicate that kind of psychological
- 10:41endurance. Not at all.
- 10:42It also brings up the architecture of the AI itself.
- 10:45We keep hearing about the hallucination problem with these
- 10:47models. Right where they just make
- 10:49things up. Yeah.
- 10:50And in the context of fiduciary duty, managing someone's
- 10:53retirement hallucination isn't just a quirky bug, it's a fatal
- 10:57flaw. Absolutely fatal.
- 10:58I look at current LMS like they're these highly eager,
- 11:01incredibly well read interns. It's a great way to put it.
- 11:04When they don't know the answer to a complex, ambiguous
- 11:07question, they don't just admit defeat, they predict the most
- 11:10statistically probable string of words to please the boss, which
- 11:15effectively means they just confidently invent a lie.
- 11:18Yeah, and that's the structural limitation of the technology as
- 11:21it exists today. LLMS are just advanced
- 11:25prediction engines, right? They lack the capacity for
- 11:28lateral leap thinking. You know, that uniquely human
- 11:31ability to connect to wildly disparate concepts to form a
- 11:36genuinely novel insight. So when they hit a gap in their
- 11:39knowledge, they just paper over it.
- 11:41Statistically, bridging the gap with a plausible narrative.
- 11:44Yeah. And in value investing, doing
- 11:46that as a shortcut to bankruptcy.
- 11:47Because value investing demands absolute intellectual honesty.
- 11:51Exactly. You have to clearly define your
- 11:52circle of competence. You must know exactly what you
- 11:55do not know. You can't just auto complete
- 11:57your way through a margin of safety.
- 11:59Perfectly said, no, you can't. So while AI is an incredible
- 12:02tool for gathering the data, the human is absolutely essential to
- 12:06judge the truthfulness of that output.
- 12:08And to hold the conviction when the market inevitably starts
- 12:12acting irrationally. Exactly.
- 12:14So how is a firm rooted in this kind of patient philosophy
- 12:19actually deploying capital in the AI space right now?
- 12:22It's interesting because. Because looking at the market it
- 12:24feels entirely top down. Everyone just assumes AI is the
- 12:27future, so capital is blindly flooding into broad AIETFS and
- 12:33literally anything with a semiconductor narrative attached
- 12:36to it. Yeah, and Himalaya explicitly
- 12:38rejects that top down macroeconomic guessing game.
- 12:41So they don't buy the hype. They approach AI companies
- 12:44strictly from the bottom up. They aren't betting on the
- 12:47abstract concept of quote the AI revolution.
- 12:50What are they looking at then? They're interrogating the
- 12:52fundamental plumbing of individual companies, looking
- 12:55for impenetrable moats and a massive margin of safety.
- 12:58Which makes their recent massive position in Google so
- 13:00fascinating. Right, because everyone,
- 13:02everyone assumes. Everyone assumes it's because
- 13:04Google is a magnificent 7 tech giant and an obvious AI winner.
- 13:08Yeah, but the timing tells a completely different story.
- 13:12It is a textbook contrarian play.
- 13:15They did not buy Google at the peak of the AI hype.
- 13:18When did they buy it? They bought it when the market
- 13:21was gripped by this narrative that Google was facing an
- 13:25existential threat. Oh, right, when ChatGPT first
- 13:28launched. Yes, the consensus was that AI
- 13:31was going to completely cannibalize Google score search
- 13:35business overnight. The market basically priced in
- 13:37the death of traditional search. Exactly.
- 13:39The market over extrapolated the immediate threat and that
- 13:42violently compressed Google's valuation.
- 13:44But Himalaya did a bottom up analysis instead.
- 13:47Right, They looked at the actual infrastructure.
- 13:50They saw the stickiness of Google's enterprise data, their
- 13:52proprietary Tensor processing units, their massive Android
- 13:56ecosystem. And running those inference
- 13:58queries for AI is incredibly capital intensive, right?
- 14:01Hugely capital intensive, and that's a cost structure Google
- 14:04was already uniquely positioned to absorb.
- 14:07So the irrational market panic create this artificial discount.
- 14:10So they basically bought the underlying infrastructure while
- 14:13everyone else was busy panic selling the narrative.
- 14:16Exactly. They found the margin of safety
- 14:18because people were overly pessimistic.
- 14:20But the source material also highlights the inverse of that
- 14:24right the extreme danger of irrational optimism.
- 14:27Yes, specifically regarding the current hardware frenzy.
- 14:31Yeah, everyone is looking at this staggering demand for chips
- 14:34like with NVIDIA and assuming the explosive growth curves are
- 14:38just going to compound forever. And this is where understanding
- 14:41capital cycles becomes critical. There is a fundamental
- 14:45disconnect between how software scales and how hardware scales.
- 14:49Walk me through that. Well, in the software space, if
- 14:51you launch a new AI tool and a million people want it, they can
- 14:55access it instantaneously. Right, the capacity scales
- 14:58infinitely with almost 0 marginal cost.
- 15:01But the physical chips required to train those models?
- 15:04Building a semiconductor fabrication plant takes years of
- 15:06permitting, billions of dollars in capital expenditure, and
- 15:10incredibly complex global supply chains.
- 15:12You can't just like spin up a new server instance for a
- 15:15physical silicon. Precisely, and the danger is
- 15:18that companies look at today's insatiable software demand and
- 15:22they over extrapolate it into the future.
- 15:24So they green light all these massive projects to build
- 15:27hardware capacity. Yes, but by the time those
- 15:31factory actually come online, say three to five years from
- 15:34now, the software demand might have plateaued.
- 15:36Or the algorithms might have become way more computationally
- 15:39efficient. Exactly, And suddenly you have a
- 15:42glut of manufacturing capacity, prices collapse and the late
- 15:45stage investors get totally wiped out.
- 15:47It's a classic boom and bust cycle.
- 15:49We saw the exact same mechanism play out in the solar energy
- 15:52boom. Oh yeah, massive initial demand
- 15:55led to rampant factory construction.
- 15:57Which resulted in massive overcapacity and brutal
- 16:00bankruptcies across the sector. AI hardware is bound by those
- 16:04exact same laws of physics and capital.
- 16:06Man, that is a huge warning sign.
- 16:09I want to pivot the lens a bit here, though.
- 16:11Let's move from financial capital to human capital,
- 16:13because this was arguably the most compelling part of Chang
- 16:16Jin's interview for me. It's deeply connected to
- 16:18everything we've talked about. It really is.
- 16:21We just talked about how building hardware ecosystems
- 16:24takes decades of patient capital.
- 16:26Well, Chang Jin argues that we have to apply that exact same
- 16:30patient structural investing to our communities.
- 16:33Yes, echoing Munger's spider web of trust.
- 16:36Right. And he highlights a very
- 16:38specific paradox in Silicon Valley regarding the Chinese
- 16:42American community. It's a really stark demographic
- 16:45contradiction. Explain it for.
- 16:47US on the engineering side, the Chinese American community is
- 16:50foundational to the tech industry.
- 16:52They are the bedrock of the STEM workforce building these very AI
- 16:57models. And that's largely the result of
- 16:59historical immigration policies. Right policies that's selected
- 17:02for highly specialized technical skills combined with a very deep
- 17:06cultural emphasis on educational stability.
- 17:08So they are building the infrastructure, they're writing
- 17:10the code, filing the patents. Yes.
- 17:12Yet when you look at the executive suites, the corporate
- 17:16boardrooms, the venture capital partnerships and certainly
- 17:18political leadership in Congress, there is a massive
- 17:21systemic gap. The community basically hits a
- 17:23ceiling. Exactly.
- 17:25They are hyper visible in the engine room and almost invisible
- 17:28at the steering wheel. And rather than just writing an
- 17:31op-ed about the bamboo ceiling and complaining about it, Chang
- 17:34Jin actually applied his capital allocation framework to fix the
- 17:38social infrastructure. He did.
- 17:40He Co founded the UCA, the United Chinese Americans
- 17:43Community Foundation. And they operate kind of like a
- 17:45seed stage venture fund for civic engagement, right?
- 17:48Giving out these micro grants of 500 to 5, $5000 to grassroots
- 17:52nonprofits. Yep.
- 17:53But honestly, and I have to ask this, how does a $500 grant
- 17:57bridge a systemic corporate leadership gap?
- 18:01I mean, it sounds well-intentioned but
- 18:03structurally insignificant. It definitely seems
- 18:06insignificant until you look at the compounding returns of civic
- 18:09muscle memory. OK, what does that mean?
- 18:12You don't bridge a corporate leadership gap by just asking
- 18:14for promotions. You bridge it by shifting the
- 18:17internal psychology of the community from essentially guest
- 18:20worker to foundational builder. Changing the mindset.
- 18:24Right. For example, one of the grants
- 18:26the UCA funded was a play about the life of Dean Lung.
- 18:29Dean Lung? Yeah, he was the Chinese laborer
- 18:31from the 1800s. Right.
- 18:33Yes, a valet and a laborer who, against staggering odds, saved
- 18:38his life earnings and donated them to Columbia University.
- 18:41Wow. And his donation was the
- 18:43foundational capital that established Columbia's East
- 18:46Asian Studies program. That's incredible.
- 18:49But ask yourself, why does funding a play about him matter
- 18:53today? Because it changes the
- 18:54historical narrative. Exactly.
- 18:56When a community's dominant narrative is that they are just
- 18:59recent economic immigrants simply here to do a job, they
- 19:03act accordingly. They keep their heads down.
- 19:05But when a community understands they have a deep historical
- 19:09legacy of foundational philanthropy in this country, it
- 19:12completely changes their posture.
- 19:14It builds civic pride. A $5000 grant to fund to play
- 19:18generates the cultural confidence required for the next
- 19:21generation to, say, run for a local school board, Which?
- 19:24Builds the political capital to run for City Council.
- 19:26Which builds A networking soft power required to eventually
- 19:29maneuver into a corporate boardroom.
- 19:31It is patient long term system investing.
- 19:35You are literally building the spider web of trust on a
- 19:38generational timeline. That's exactly what it is.
- 19:41Which naturally brings us to the final piece of the puzzle here,
- 19:44the next generation. If AI is automating logic and we
- 19:49need our kids to eventually navigate these corporate
- 19:52boardrooms and civic institutions, how do we actually
- 19:56prepare them? It's the $1,000,000 question.
- 19:58Seriously, what is the curriculum for a completely
- 20:02unpredictable future? Well, the source outlines a
- 20:04critical 2 prong strategy. The first prong is non
- 20:08negotiable absolute AI literacy. Meaning they have to understand
- 20:11the tech. Deeply, the next generation has
- 20:13to understand the architecture of these tools, their
- 20:16statistical limitations, and how to wield them.
- 20:19The old cliche really holds true.
- 20:21AI won't take your job, but a human wielding AI absolutely
- 20:24will. OK, so they don't necessarily
- 20:25need to like hand code the neural network from scratch, but
- 20:28they need to know how to expertly prompt it and evaluate
- 20:31the output exactly. So what is the second prong?
- 20:33A radical doubling down on the traits that cannot be automated.
- 20:37AI cannot replicate human empathy, it doesn't have a moral
- 20:40compass, it doesn't possess faith, and it cannot generate
- 20:43truly lateral, unpredictable imagination.
- 20:46So the educational imperative is to cultivate those profoundly
- 20:50human characteristics. But practically speaking, I
- 20:53mean, how does a parent actually do that?
- 20:56The interview touched on the friction between traditional
- 20:58tiger parenting, which is, you know, incredibly structured and
- 21:01disciplined, and the more modern exploratory American happy
- 21:06education model. How do you balance those When
- 21:10the future economy is just this moving target?
- 21:13Ching Jin synthesizes the two because leaning too far into
- 21:16rigid discipline creates a worker who is excellent at
- 21:19following instructions, which is exactly what AI is going to
- 21:22automate. Right, the machine will always
- 21:23be better at following the rules.
- 21:24But leaning entirely into unstructured exploration often
- 21:28fails to build the grit necessary to overcome true
- 21:30adversity. So you need a hybrid.
- 21:32The framework requires the strict discipline to build
- 21:35foundational habits and resilience, paired
- 21:37simultaneously with the psychological freedom to explore
- 21:41obscure interests without this paralyzing fear of failure.
- 21:44Oh I love that you essentially need the discipline to master
- 21:49the rules so that you have the context to know exactly when and
- 21:53how to break them. Perfectly stated, and the
- 21:56overarching goal isn't just to make them competitive in the
- 21:58labor market. What's the bigger picture?
- 22:00As technology collapses borders and tightly interconnects the
- 22:04global economy, the ultimate goal is to raise youth with a
- 22:07global perspective, individuals who have the cultural agility to
- 22:11strive for a cooperative global society rather than retreating
- 22:15into 0 sum ideological conflicts.
- 22:18Man, let's pull the threads together here, because this has
- 22:20been a journey we started with Charlie Munger's mental models,
- 22:24the realization that true rationality requires
- 22:27acknowledging our cognitive biases and actively building
- 22:30structural trust in our communities.
- 22:32And then we looked at AI not as some magic Oracle, but as a
- 22:35cognitive steam engine. It can process the data, sure,
- 22:39but it desperately needs human intellectual honesty and patient
- 22:43conviction, like holding through those massive drawdowns in BYD
- 22:48to actually generate long term value.
- 22:50And we also saw how that framework prevents us from
- 22:52getting swept up in macro narratives.
- 22:55It allows investors to buy the artificial pessimism in a
- 22:58company like Google while avoiding the capacity traps of a
- 23:02hardware bubble. And most importantly, we saw how
- 23:05that exact same philosophy applies to human infrastructure,
- 23:09using targeted micro grants to rewrite cultural narratives and
- 23:13build the civic soft power that future generations are going to
- 23:16need. It's all connected.
- 23:18It really is you listening. Remember, the tools are evolving
- 23:20at a speed we can barely comprehend, but the core
- 23:23variables? Emotional control, contrarian
- 23:25patients, and genuine community building.
- 23:28Those are the only things the algorithms can't replicate.
- 23:30Which leaves us with a truly vital question about the future
- 23:33of our market. If we increasingly outsource our
- 23:35baseline financial analysis, our logic, and our risk modeling to
- 23:40AI, does the next generation slowly lose the cognitive muscle
- 23:44required for rational value investing?
- 23:46Oh wow, if everyone is just relying on the exact same hyper
- 23:49efficient LLMS to make all their decisions.
- 23:52Exactly. We could drift into a reality
- 23:55where the markets are entirely driven by algorithmic consensus
- 23:58and sudden massive hallucination bubbles.
- 24:01That is terrifying. In a world where the machines
- 24:03perfectly automate standard logic, the only true source of
- 24:07alpha left might just be the stubborn, highly flawed human
- 24:10being who simply refuses to plug into the consensus.
- 24:13The ultimate margin of safety might just be our own human
- 24:16stubbornness. That is a lot to chew on.
- 24:18Keep questioning the consensus, keep building your own web of
- 24:21trust, and we'll see you on the next deep dive.