Latest / Deep Values: Value Investing Deep Dives / Michael Burry's 13F Final AI Accounting Short
Transcript
- 0:00Imagine you are sitting at a high spakes poker table with one
- 0:04of the most famous contrarian investors of our time.
- 0:07Right, the chips are stacked high.
- 0:08Exactly. The tension in the room is
- 0:10thick, and right before he cashes out, leaves the casino
- 0:14and just goes completely off the grid, he pauses.
- 0:18Oh, I love this analogy. Right.
- 0:20He looks right at you and he flips his cards over.
- 0:22He shows you his final hand. I mean, it is the ultimate
- 0:27parting gift for anyone who's actually paying attention, a
- 0:29clear look into the mind of someone who has, you know,
- 0:32historically seen what the rest of the market completely missed.
- 0:36So today we are looking at the Q32025 Form 13F for Michael
- 0:41Bury's Sign Asset Management. And for those of you tracking
- 0:44the industry, you know this is a massive milestone.
- 0:46Yeah, it really is. It's the final regulatory
- 0:48snapshot before Scion formally terminated its SEC registration
- 0:52back in November 2025. Exactly.
- 0:54He's moved his insight strictly into the shadows now to his
- 0:58independent Cassandra Unchained commentary, meaning this is it.
- 1:01This is his final public footprint.
- 1:02We have a stack of forensic financial analysis detailing
- 1:05this portfolio. And our mission today is to
- 1:08decode his ultimate 2 tiered investment architecture.
- 1:13So OK, let's unpack. This, well, we really have to
- 1:16approach this document with the right lens.
- 1:18I mean, this isn't just a list of stock tickers to blindly
- 1:20copy. It is a window into a highly
- 1:23specific forensic accounting philosophy.
- 1:26It's honestly a master class in modern value investing.
- 1:29The architecture here is what really stands out because we are
- 1:32looking at a deliberate extreme bifurcation of CAP.
- 1:37I look at this structure and I just see a financial barbell.
- 1:40Yeah, that's a good way to put it.
- 1:41On one side of the barbell you have this massive $1.38 billion
- 1:46notional derivative book that is heavily shorting the artificial
- 1:50intelligence bill. Which sounds crazy at first.
- 1:53Glance, it's like holding a high stakes defined risk insurance
- 1:56policy on your entire neighborhood burning down.
- 1:57And then you look at the other side.
- 1:59You have a heavily concentrated direct equity book of just $68.1
- 2:03million parked in 4 completely unloved, you know, boring
- 2:07businesses. It's like sticking your life
- 2:09savings into a hyper conservative cash flowing
- 2:12mattress. Exactly.
- 2:13And if we connect this to the bigger picture, what you are
- 2:16really looking at is a modern adaptation of Benjamin Graham
- 2:19and David Dodd. The classic value investing
- 2:22fathers. Right, but adapted for today's
- 2:24hyper fast complex markets. I mean, this strategy doesn't
- 2:28just screen for cheap stocks using basic price to earnings
- 2:31ratios. It hunts for a quantifiable
- 2:34margin of safety that's driven by extreme, often completely
- 2:38unjustified, market pessimism. So he looks for the panic, but
- 2:42he doesn't just buy because people are scared.
- 2:44No, not at. All he verifies that the panic
- 2:46is irrational by doing the exhausting line item forensic
- 2:50reading of the 10K footnotes. He's looking for hidden
- 2:53vulnerabilities or hidden strengths long before standard
- 2:58gap earnings actually reflect them.
- 3:00Because the standard accounting rules often mask the physical
- 3:03reality of a business. To really understand how this
- 3:05forensic analysis works in practice, we need to 1st examine
- 3:08where he parked that 68.1 million the the mattress side of
- 3:12the barbell. Let's talk about catching
- 3:14falling knives with durable cash flows, starting with the two
- 3:18largest direct equity holdings, which are Molina Healthcare and
- 3:22Lululemon. Molina represents 35% of this
- 3:25equity book and Lululemon is 26%.
- 3:28So like more than half is cash is in just two companies that
- 3:32Wall Street essentially abandoned.
- 3:35Molina Healthcare is a perfect starting point here.
- 3:37It's a managed care organization mostly dealing with government
- 3:40sponsored programs like Medicaid, and the market just
- 3:43absolutely dumped this stock. Yeah, big time.
- 3:46The share price dropped 34% and the shares in this portfolio
- 3:50were acquired around $191.00 a share.
- 3:54That implies A wildly depressed .2 times price to sales
- 3:57multiple. I mean you are paying pennies
- 4:00for every dollar of revenue they generate.
- 4:02But the market panic wasn't totally imaginary though.
- 4:04I mean, the data shows Molina's medical care ratio spiked up to
- 4:07over 89.1%, right? And for anyone unfamiliar, that
- 4:11ratio is basically how much they spend on actual medical care
- 4:14versus what they take in from government premiums.
- 4:16So a higher number means squeezed profits.
- 4:19And this spike happened because of post COVID Medicaid
- 4:22redeterminations. State started kicking healthier
- 4:26people off the Medicaid rolls, which leaves Molina with a
- 4:29sicker, vastly more expensive patient pool.
- 4:32So the margins just collapsed on paper?
- 4:34They collapsed, yes, but here is the forensic insight.
- 4:37This is the underlying mechanism the market algorithms completely
- 4:41ignore. OK, state Medicaid rate setting
- 4:43doesn't happen in real time. There is a structural 18 to 24
- 4:47month lag. Oh wow, so they're just waiting
- 4:50on the government? Precisely.
- 4:51State actuarial boards need time to process all this elevated,
- 4:55acclaimed data, run their own models, and then revise the
- 4:58capitation rates upward to match the new reality of the medical
- 5:02costs. Yeah, so it's a bureaucratic
- 5:03rubber band effect. You get stretched thin while the
- 5:05state does its paperwork, but eventually the rates have to
- 5:08snap back up to cover the actual cost of care.
- 5:11Exactly. And once those rates adjust,
- 5:13Molina's normalized operating margins will return to their
- 5:16historical 3.5 to 4.5% range. When you run that margin against
- 5:20their massive revenue base, the math dictates an unlocking of 18
- 5:24to 21 dollars in earnings per share.
- 5:26That's. Huge.
- 5:27Right. The market is pricing the stock
- 5:29as if the temporary margin squeeze is permanent.
- 5:31And I guess the margin of safety here is that they aren't going
- 5:34bankrupt while they wait out this 18 month lag.
- 5:37I mean, they are sitting on 8.3 to $8.9 billion in cash against
- 5:42just 4 billion in debt. Yes, they're a financial
- 5:45fortress. You basically just buy the dip
- 5:47and wait for the government paperwork to catch.
- 5:49The structural safety in the balance sheet is key, and we see
- 5:53a similar theme with the second major holding, Lululemon.
- 5:56This is a fascinating play because of the sheer conviction
- 5:59involved. There really is.
- 6:00I mean, he doubled down. He increased the position by
- 6:03100% up to 100,000 shares. And this is while the stock was
- 6:07basically in freefall. Dropping 50% from its highs.
- 6:10Right, due to slowing North American sales and some heavily
- 6:13publicized inventory issues with their color palettes.
- 6:16But I have to play devil's advocate here though.
- 6:19Go for it. Consumer trends are incredibly
- 6:21fickle. People can just buy cheaper
- 6:24athletic wear from emerging competitors on Instagram or
- 6:27TikTok now. Why step in front of this
- 6:29specific falling knife when the apparel industry is notoriously
- 6:33unforgiving? It's a fair question, but the
- 6:37answer lies in the undeniable structural math of their
- 6:40specific business model. When you lift the hood on
- 6:43Lululemon, you find 0 long term debt.
- 6:45Zero. Literally none, and they are
- 6:48sitting on $1.5 billion in pure unencumbered cash.
- 6:53OK. That is a massive safety net for
- 6:55an apparel brand. Beyond the safety net, it's the
- 6:57capital efficiency that separates them from standard
- 7:00mall retailers. They maintain an adjusted return
- 7:02on invested capital between 35 and 42%.
- 7:05That's insane. That is astronomically high.
- 7:07Every dollar they put back into the business generates a massive
- 7:10return. And while everyone was panicking
- 7:12over the North American sales slowing to 4% growth, the market
- 7:15entirely overlooked their growth in mainland China, which was
- 7:18compounding at over 30%. And the pricing power is what
- 7:21really validates that return on invested capital, right?
- 7:25Because traditional sportswear brands sell through massive
- 7:29department stores, they are entirely at the mercy of that
- 7:32retailer deciding to mark down prices by 40% just to clear the
- 7:36racks at the end of the season. It kills the brand prestige.
- 7:40But Lululemon controls over 90% of its sales directly to the
- 7:44consumer. They own the entire transaction
- 7:47because they operate that direct to consumer model.
- 7:49They never have to heavily discount to save a middle man's
- 7:52margins. They keep their premium pricing
- 7:54power intact, which is what drives their retail store
- 7:57productivity to an absurdly high $1600 per square foot.
- 8:02Wow, yeah. The brand is a cash generating
- 8:05machine that the market temporarily priced like a dying
- 8:08legacy retailer. So Molina and Lululemon are both
- 8:11classic deep value plays on temporary operational hiccups.
- 8:15The market overreacts to short term data.
- 8:17The forensic math proves the underlying business is
- 8:19incredibly safe. You simply wait for the
- 8:21normalization. But the next two equity plays on
- 8:23this barbell are structurally different.
- 8:26We're moving away from operational turnarounds and
- 8:29looking at sheer financial engineering capital structure
- 8:32arbitrage. Let's dig into SLM Corp, better
- 8:36known as Sallie Mae and Brooker Corporation.
- 8:38Sallie Mae is a master class in finding deep value in an
- 8:41actively despised sector. The position is substantial,
- 8:44over 480,000 shares, and they dominate the Private Student
- 8:49Lending market with over 50% market share.
- 8:52But mentioning student loans to any investor right now sets off
- 8:54alarm bells. I mean, the political headlines
- 8:57are entirely focused on loan forgiveness, massive default
- 9:01rates, and regulatory crackdowns.
- 9:02That is the macro public narrative for sure.
- 9:05But the forensic reality of Sallie Mae's specific private
- 9:08portfolio tells an entirely different story.
- 9:10They operate with an incredibly conservative, underrating
- 9:13profile. Over 90 to 93% of their new
- 9:16originations require a cosigner. Right, so a parent or guardian
- 9:19with actual assets is legally on the hook if the student
- 9:22defaults. Furthermore, the average FICO
- 9:24credit score at approval is over 750.
- 9:27Their credit quality is pristine despite the scary headlines.
- 9:31Their loss reserves are aggressively padded.
- 9:34But the real magic here is how management is allocating their
- 9:37profits. It is what we call capital
- 9:40cannibalism. Wait, explain that.
- 9:42Between 2020 and 2025, Sallie Mae retired an absurd 52% of its
- 9:47total Common Shares outstanding. They bought back more than half
- 9:51of their own company in five years.
- 9:52Yes, if you owned a slice of this pie in 2020, your ownership
- 9:56state has effectively doubled without you doing a single
- 9:58thing. That's wild.
- 10:00The mechanics of those buybacks are what make it a masterstroke.
- 10:03They are buying those shares back at single digit price to
- 10:05earnings multiples. Meanwhile, the underlying
- 10:08business generates a return on tangible common equity of over
- 10:1120%. So for the listener, return on
- 10:13tangible common equity essentially means looking at the
- 10:15pure hard physical cash and assets the company holds.
- 10:19Yeah, subtracting the fluff like goodwill and seeing how much
- 10:21profit they generate from it. Right.
- 10:23Generating over 20% on hard assets while trading at a single
- 10:27digit PE is incredibly rare. It's a compounding machine
- 10:30hiding in plain sight. The valuation is artificially
- 10:34depressed by macro political noise about federal student
- 10:36loans that fundamentally doesn't impact their pristine private
- 10:40loan book. Which brings us to the 4th
- 10:42equity position, Bruecker Corporation.
- 10:45And this is perhaps the most elegant move in the entire
- 10:48portfolio. It's brilliant.
- 10:50Earlier in the year, Bury held 250,000 shares of Bruecker
- 10:54common stock, but in this final 13 F filing, that common stock
- 10:58was entirely liquidated. You swap that capital into
- 11:0148,000 shares of Brooker's 6.375% series 8% stock.
- 11:07Think of it like moving from a highly volatile coach seat into
- 11:10first class on the exact same airplane during extreme
- 11:12turbulence. You cap your upside.
- 11:14You aren't going to get rich if the stock rockets to the moon,
- 11:17but you get a guaranteed 6.375% drink service and you get to get
- 11:21off the plane first if it crashes.
- 11:22That is a perfect analogy. It is a textbook capital
- 11:26structure arbitrage. Let's look at the company
- 11:28itself. Bruecker has a massive economic
- 11:31Moat. They operate a near monopoly
- 11:33duopoly in highly specialized scientific instruments.
- 11:37We are talking about multimillion dollar nuclear
- 11:40magnetic resonance and mass spectrometry equipment.
- 11:42The heavy machinery that university research labs and
- 11:46global biopharma companies absolutely require to function.
- 11:49Exactly. The problem for anyone holding
- 11:51the common stock is that the biopharma industry is currently
- 11:54slashing its research and development capital
- 11:56expenditures. Interest rates are up, funding
- 11:59is tighter, so labs delay buying new $1,000,000 machines.
- 12:03That creates massive volatility in broker's standard earnings.
- 12:07So by swapping into the preferred stock, you completely
- 12:09insulate yourself from that earnings volatility.
- 12:12Preferred shareholders sit higher up the capital ladder.
- 12:14Right, they have seen your dividend claims.
- 12:16Yeah, as long as broker doesn't go completely bankrupt and with
- 12:20gross margin sitting around 50%, they are highly solvent.
- 12:24You collect your guaranteed yield.
- 12:26You strip out the downside risk of the common equity while
- 12:29remaining parked in an elite life sciences franchise.
- 12:32It is the ultimate defensive posture.
- 12:34You secure rock solid cash flows and dividends across 4 uniquely
- 12:38insulated businesses. But these businesses do not
- 12:41exist in a vacuum. A market crash will still drag
- 12:44down their valuations temporarily.
- 12:46So to protect against the broader market euphoria, the
- 12:49portfolio pivots to the dark side.
- 12:51Here's where it gets really interesting.
- 12:54We have to talk about the $1.38 billion elephant in the room.
- 12:57Oh yeah, the massive asymmetric derivative bets against the AI
- 13:01boom. The filing shows 5 million put
- 13:03contracts on Palantir and 1,000,000 put contracts on
- 13:06NVIDIA. Before anyone assumes he bet a
- 13:08billion dollars against tech, it is vital to contextualize that
- 13:12$1.38 billion figure. That is the notional value.
- 13:15It represents the total value of the underlying shares controlled
- 13:18by the option contracts. The SEC requires it to be
- 13:21reported that way, but the actual capital at risk, the cash
- 13:25paid out of pocket for the option premiums, was only about
- 13:289 to $10 million per position. So the downside is strictly
- 13:33capped. He's risking 10 million to
- 13:36potentially make an absolute fortune if the tech market
- 13:39corrects. But the obvious question is how
- 13:42can you possibly short NVIDIA when artificial intelligence is
- 13:45structurally changing the global economy?
- 13:48Every major tech company on earth is buying their chips as
- 13:51fast as they can produce them. Isn't betting against that a
- 13:54widowmaker trade? What's fascinating here is that
- 13:57the thesis isn't actually about the technology failing.
- 14:00He isn't shorting the capability of AI or suggesting the chips
- 14:03don't work. He is shorting the accounting
- 14:05wait. Breakdown the mechanics of that.
- 14:06How do you short accounting? It comes down to how the biggest
- 14:09buyers of these chips, the hyperscalers like Microsoft,
- 14:12Meta, and Amazon, are recording these massive purchases on their
- 14:16corporate balance sheets. When they buy billions of
- 14:19dollars worth of AI data center servers, they don't expense the
- 14:22entire cost immediately under standard rules.
- 14:25They capitalize it under property, plant and equipment,
- 14:28and they depreciate it over what's called it's useful life.
- 14:32That makes intuitive sense. You buy a massive asset, you use
- 14:35it for several years, so you spread the cost out over time on
- 14:38your profit and loss statement. The forensic red flag is that
- 14:42over the last few years, to artificially boost their gap
- 14:44operating income, these hyperstalers have been quietly
- 14:47extending the estimated useful life of these servers.
- 14:50Really. Yeah.
- 14:51They moved the depreciation schedule from three years back
- 14:54in 2020 all the way out to nearly six years by 2025.
- 14:58By stretching the schedule out, you drastically lower your
- 15:01annual expenses, which magically makes your current profits look
- 15:05far higher to Wall Street. But wait, NVIDIA releases a new,
- 15:09exponentially faster chip architecture every single year?
- 15:14Yeah, they went from Hopper to Blackwell to Ruben in rapid
- 15:17succession. And the new chips are so vastly
- 15:20more energy efficient that it actually becomes a financial
- 15:23liability to run the older power hungry servers.
- 15:27Exactly. They're physically making their
- 15:29own hardware obsolete annually. That is the massive structural
- 15:32contradiction this short position targets.
- 15:34You have corporate spreadsheets claiming these massive server
- 15:37farms will be economically viable for six years, while the
- 15:40physical reality of the silicon market renders them obsolete in
- 15:4412 to 18 months. So the spreadsheets are
- 15:46completely disconnected from the physics of the data center.
- 15:49So if a company is sitting on billions of dollars of outdated
- 15:51servers that they are pretending have four years of life left,
- 15:55what happens? Well, the forensic calculation
- 15:57suggests this accounting mismatch is currently hiding
- 16:00roughly $176 billion in depreciation across the broader
- 16:05tech industry. It is overstating corporate tech
- 16:07earnings by roughly 20 to 27%. When physical reality finally
- 16:12forces these companies to write down those obsolete dosed
- 16:15assets, the earnings collapse will be violently sudden.
- 16:18It is a direct parallel to the Cisco fiber optic bubble.
- 16:21In the year 2000, people were laying millions of miles of dark
- 16:25fiber, assuming the accounting would hold up until the
- 16:28overcapacity cause the entire sector's valuations to compress
- 16:32overnight. Exactly the same playbook.
- 16:34He isn't betting that AI goes away, he's capturing the
- 16:37inevitable valuation compression when the capital expenditure
- 16:39cycle slows down down and reality catches up with a
- 16:42deferred accounting. And to perfectly balance the
- 16:44strategy, the derivative book wasn't exclusively shorts.
- 16:48There were also strategic long call options on Pfizer and
- 16:51Halliburton. Which again relies on that
- 16:54concept of convexity. You use cheap call options to
- 16:56get defined risk upside in totally unloved sectors without
- 17:00tying up your precious equity capital.
- 17:02With Pfizer, you get exposure to a post COVID pipeline turn
- 17:05around and a massive dividend yield.
- 17:08With Halliburton, you capture the upside of a global deep
- 17:11water drilling cycle without having to actually own
- 17:13incredibly volatile energy stocks in your main portfolio.
- 17:17It provides highly leveraged upside with strictly limited
- 17:20risk. It is paired perfectly with the
- 17:22extreme defense of the cash equities.
- 17:24We have the full picture now. Extreme defense in the cash
- 17:27equity mattress on one side, extreme calculated offense in
- 17:32the derivative book on the other.
- 17:33So what does this all mean for you, the listener?
- 17:35How does this philosophy stack up against, say, the godfather
- 17:38of value investing, Warren Buffett?
- 17:40It is a profound divergent in execution.
- 17:43Both stem from the original principles of Graham and Dodd,
- 17:46but they are built for entirely different temperaments and
- 17:48market realities. Buffett is looking for permanent
- 17:51ownership. He wants wide Moat compounders
- 17:54that he can buy and hold for decades, and he generally avoids
- 17:57shorting altogether. If the market gets too frothy,
- 18:00he will simply let his cash accumulate and wait.
- 18:02Whereas this portfolio architecture is actively hunting
- 18:05for what you might call roadkill value.
- 18:07Tactical holds. The strategy involves buying
- 18:10cyclically or operationally impaired companies at steep
- 18:13tangible discounts, holding them for maybe 3 to 36 months until
- 18:17the market wakes up and the intrinsic value normalizes, and
- 18:20then aggressively exiting. And crucially, it actively
- 18:23deploys convex derivatives to exploit specific identifiable
- 18:28accounting anomalies and market bubbles while they happen.
- 18:31It really is a master class in reading the footnotes most
- 18:34people ignore. Before we wrap up.
- 18:36If you want to access more Advanced Research, follow these
- 18:38gurus latest movements and leverage the latest Deep Values
- 18:41agents. You're welcome to download the
- 18:43Deep Values app from the App Store.
- 18:45And if you want to track stock market temperature yourself,
- 18:47welcome to download the iOS app Buffett Indicator from the Apple
- 18:51App Store or Bubble Watch from Google Play.
- 18:53Awesome tools. Just a quick reminder, you
- 18:55should know that all our analysis today is based entirely
- 18:58on publicly available sources and may be subject to revisions
- 19:02or differing interpretations. All content from this channel is
- 19:06for research and entertainment purposes only and does not
- 19:09constitute any investment or financial advice or service.
- 19:13Always conduct your own due diligence before making
- 19:15investment or financial decisions.
- 19:17Absolutely. And as we look at this final
- 19:20public filing, it leaves us with an incredibly provocative
- 19:23thought. Scion Asset Management is now
- 19:25closed to outside capital. The SEC registration is
- 19:29terminated. The 13 F filings are a thing of
- 19:31the past. As this deeply forensic mind
- 19:34moves strictly into the shadows of independent research, we have
- 19:37to wonder what other massive accounting illusions in the
- 19:40broader market are hiding in plain sight right now.
- 19:43You're at the poker table. The cards have been shown, but
- 19:46the game in the broader market is still going.
- 19:48What happens when the physical reality of the world finally
- 19:50catches up with the spreadsheets?
- 19:51Keep your eyes open until next time.