563: The 5.5 Million-Percent Man, OpenAI vs Microsoft, Netflix & HBO, Cliff Sosin & Carvana, ChatGPT Energy Use, CRISPR Pigs, and Andor

"The theory explain everything — except why no one else can replicate it!"

In looking for someone to hire, you look for three qualities: integrity, intelligence, and energy.

But the most important is integrity, because if they don't have that, the other two qualities, intelligence and energy, are going to kill you.

—Warren E. Buffett

🛀💭📺 Netflix should create its own HBO-like division, give a clear mission, its own name and identity, lots of autonomy, and make it a separate premium subscription upsell for Netflix subs.

Why have something separate? By segmenting it out, they wouldn’t have to target the lowest common denominator/median Netflix watcher, and incentives become much sharper (never underestimate the power of incentives! 🥕).

During its golden age, HBO could make The Wire and Rome because they knew they weren’t targeting the middle of the bell curve of TV watchers — in fact, their customers were a self-selected group of viewers who cared deeply about quality and were paying them specifically to get something different and better! 💵🤝🎬

CBS could never have made The Sopranos or Deadwood. Netflix is now more like CBS than HBO.

I think a lot of people would gladly pay an extra $10-20/month for truly great shows.

😵‍💫💭 Ever feel like you’re being gaslit by a robot?

I asked AI to correct the spelling for something I wrote, and it flagged the abbreviation “f.ex.” as non-standard English:

“f.ex.” is not standard in English. It’s a shorthand that’s more common in Scandinavian or European contexts (e.g., Norwegian, Danish, Swedish, Icelandic), where it stands for "for example" (similar to “e.g.” in English). 

But in native English writing, especially in professional or widely-read contexts, it’s rarely used and may confuse readers who aren’t familiar with it.

In English, the standard is: e.g. (from Latin exempli gratia, meaning “for example”)

At first, I thought this had to be a hallucination. It wouldn’t have been the first. I double-checked from a few other sources.

😳

Well, the damned machine was right!

I’ve been using “f.ex.” to mean “for example” for years and never realized it wasn’t common. I wonder where I got it..? 🤔

My best guess: as I was learning English in my teens, I had a lot of Scandinavian friends on music forums. Maybe I picked it up there and never gave it a second thought.

Well, I’m sure you knew what I meant when I used f.ex. (even if it raised an eyebrow 🤨)

Variety is the spice of life! I gotta keep you on your toes with a few curveballs ⚾️

💚 🥃 🙏☺️ If you’re a free sub, I hope you’ll decide to become a paid supporter.

You’ll get access to paid editions, the private Discord, and get invited to the next supporter-only Zoom Q&As with me.

If you only get one good idea per year, it’ll more than pay for itself. If it makes you curious about new things, that’s priceless.

The next best thing is to share something you enjoyed here with someone who might like it too. Drop a screenshot and a link in your group chat! 💬

---

🏦 💰 Liberty Capital 💳 💴

🐐📈💰 The 5.5 Million-Percent Man (and that’s just Berkshire…)🏆🥇

As I’m sure you’ve seen, Warren Buffett announced at the Berkshire Hathaway annual meeting that he would be handing over the CEO role to Greg Abel at the end of the year.

As usual, Byrne Hobart (💚💚💚💚💚 🥃 ) has the best analogy:

In The Lord of the Rings, the end of the Third Age begins with an elderly and highly-respected hobbit throwing an unusually elaborate version of his usual annual party and then, at the very end and to the surprise of everyone except a handful of close relatives, announcing that he's soon to disappear forever.

In some ways, this surprised everyone, including Berkshire’s next CEO, Greg Abel. But in others, it was widely expected. Buffett is 94 years old, and we haven’t yet defeated the diseases of aging.

We’ll see what the final scorecard looks like for the Buffett era at Berkshire at the end of 2025, but the 1964-2024 run shows a cumulative 5,502,284% gain, or a 19.9% CAGR over 60 years! 😲

Morgan Housel pointed out:

Berkshire could fall by 99% and he'd still be ahead of the SP500 since Buffett took it over.

And that’s not even taking into account the Limited Partnership years. His seven partnerships compounded at ˜31–32% CAGR between 1956 to 1969!

There are many ways of doing the math: Net returns? Or do gross returns more accurately reflect his investing prowess? How to deal with the 1965-1969 overlap between BRK and the LPs? The partnerships began on May 5th, 1956, but Buffett’s letters only report full‑year results from 1957, etc.

But the bottom line is: 10s to 100s of millions of percent! 🤯

Incredible! Truly the greatest to ever do it, and the investing world will likely never again see a 60+ year run like this.

But as amazing as this is, I want to highlight why it’s even *more* impressive than it appears.

📊 Academics have tried to “actually” Buffett’s success

Tyler Cowen recently posted a paper by academics that says:

Berkshire Hathaway has realized a Sharpe ratio of 0.76, higher than any other stock or mutual fund with a history of more than 30 years, and Berkshire has a significant alpha to traditional risk factors.

However, we find that the alpha becomes insignificant when controlling for exposures to Betting-Against-Beta and Quality-Minus-Junk factors. Further, we estimate that Buffett’s leverage is about 1.6-to-1 on average. Buffett’s returns appear to be neither luck nor magic, but, rather, reward for the use of leverage combined with a focus on cheap, safe, quality stocks.

Decomposing Berkshires’ portfolio into ownership in publicly traded stocks versus wholly-owned private companies, we find that the former performs the best, suggesting that Buffett’s returns are more due to stock selection than to his effect on management. These results have broad implications for market efficiency and the implementability of academic factors.

If it's just a bit of leverage and safe stocks, why isn't it common?

Academics have had to keep adding sigmas to the asterisk next to Buffett’s name over the years..

It reminds me of the (surely apocryphal) story about Eugene F. Fama, best known for developing the Efficient Market Hypothesis (EMH), buying Berkshire stock "just in case."

Why is Buffett such a unique creature?

Why don’t we see a bunch of people from Buffett’s cohort with 50 or 60-year track records of compounding capital at 20%/year? Why have all the “mini-Berkshires” underperformed the mothership over longer periods — except Fairfax, thanks to its recent comeback, but it’s still a much shorter record — despite having *much* smaller capital bases?

What the academics write makes sense in theory, but is it that simple in practice? The rewards are so great that if their theories did work, we’d see a lot more Buffetts.

The theory explain everything — except why no one else can replicate it!

🧭 Compared to other billionaires, Buffett’s path is singular

His peers in finance largely got rich by charging big fees (2/20) on large pools of AUM while Buffett has paid himself very little at Berkshire, and most of his wealth comes from equity appreciation.

Btw, when looking at the above, remember that in 2006, Buffett owned 474,998 A shares of Berkshire, or 42.5% of A shares outstanding at the time. He’s donated close to 60% of them since. If he hadn’t been so generous, he’d be worth about $365 billion today.

And his centi-billionaire peers? Almost all of them built rocketship businesses that grew super-fast organically: Bezos, Zuck, Jensen, Ellison, Gates, Bloomberg… The closest to Buffett on that list is probably Bernard Arnault, but his path is still very different and still mostly built on a specific industry’s growth.

Buffett built all this from thin air, mostly by reading, thinking, making phone calls, and buying the same public stocks that anyone else could see.

He didn’t invent a product. He didn’t disrupt an industry. He didn’t have one big hit that created 99% of his wealth.

He just kept going, patiently compounding capital. That’s unique.

🎓 Buffett isn’t just a top investor, he’s also a top teacher

And while doing this, he was teaching everybody else what he was doing and creating his own competition. He effectively opened-sourced his version of value investing and spawned multiple generations of investors who were all looking for the same kind of stocks he was looking for.

Most other successful investors are pretty secretive and are deadly afraid of giving away their edge.

He was rewarded for this by becoming the buyer of choice for many business owners. They sought him out. That halo effect meant better deals, lower prices — even improved business outcomes, simply from being part of Berkshire.

But this is an *earned* reputation built over decades of success, and I still wonder if whatever is gained there fully makes up for training millions of value investors.

Would a secretive Buffett have compounded even faster? That’s an interesting alternate history 🤔 Maybe, especially in the '70s to the '90s when stock-picking was more dominant than M&A at Berkshire.

Don’t get me wrong, Buffett can be ruthless and play hardball with the best of them. He has that killer instinct (show me a billionaire who’s a meek lamb…). But all things considered, he genuinely has a strong impulse to teach and share what he knows.

It’s a choice, he didn’t have to do that and spend hours in Q&A with small shareholders and such.

For that, I’ll be eternally grateful. I wouldn’t be who I am today if I hadn’t discovered him a couple of decades ago. 🙏

📈📉📈 Cliff Sosin on being down 99% in Carvana… and still winning big 🚘💣💥

I didn’t know Sosin at all before listening, but I enjoyed the interview:

• 🎧 Cliff Sosin: Investing in Carvana

The last part about his investment in Carvana is particularly interesting. It’s incredibly rare to be heavily concentrated in something that goes down 99% and then ultimately have that be a successful investment.

I have to show the chart because words can’t quite paint that picture:

I don’t know if his claim about “the emperor of intellectual pursuits” is right, but as an investor, I sure would like to believe that it is!

In academia, people write papers. They're wrong, they're right, eventually they die with a bunch of ideas, most of them wrong.

In politics, people have views and they're definitely wrong, unlike a lot of them.

And in business people, they have this narrow world and they have to be really good about executing in their narrow world, but they don't necessarily need a deep understanding. You can run a deli without necessarily having a deep understanding of why meat prices are what they are.

But in investing, it is wildly accountable.

You're making predictions about which businesses are going to win and which are going to lose, which to understand. Businesses are complex social structures embedded in our society, which is a complex social structure. So what do you need to understand how a business's success or failure over time?

It's like you kind of need to know everything. So unlike all these other pursuits, this one's highly accountable. I think it's the emperor of intellectual pursuits. I think that there's no arena to train people better about understanding the world. And I think if you're just curious about the world, there's nothing more interesting than studying business.

For more on Carvana, this podcast interview with Neil Mehta of Greenoaks Capital is also good (“How many companies can you name that went from $70 billion market cap to one and weren't like a fraud? Zero.”)

🤖🔐 OpenAI to remain controlled by non-profit, tries to slash revenue share to Microsoft ✂️💸

OpenAI’s attempt to free its profit engine from its non-profit structure looks like it has hit a regulatory wall 🤔

In a blog post, the board admitted the nonprofit will “continue to be overseen and controlled by that nonprofit” while its capped‑profit LLC will be changed to a Public Benefit Corporation (PBC).

They say they made this decision “after hearing from civic leaders and engaging in constructive dialogue with the offices of the Attorney General of Delaware and the Attorney General of California,” but it’s probably just a way of saying the AGs could have blocked the conversion.

OpenAI will stick with nonprofit control and give everyone else plain‑vanilla stock. No more weird capped‑profit waterfall:

• Structure: the LLC becomes a PBC so employees and future investors can get conventional equity while the nonprofit keeps a majority stake and board veto.

• Governance: the nonprofit board, reconstituted after the 2023 Altman‑gets‑fired‑then‑rehired drama, keeps the steering wheel, likely avoiding a long and costly legal battle over charitable‑asset rules. But how much of the PBC will be owned by the nonprofit? ¯\_(ツ)_/¯

• Capital raise: management still says it needs “hundreds of billions (maybe trillions) of dollars” (!) for compute and fabs, but that dough will have to come with dual‑mission strings.

This last part is important because fundraising is the lifeblood here.

Musk likely wanted to screw with OpenAI’s ability to access capital, which would be a big win for xAI/Grok as it tries to compete. But it looks like this new structure, while not optimal for investors, doesn’t prevent them from investing in OpenAI.

An IPO should be doable with the shares of the for-profit entity, even if the non-profit has ultimate control (and Sam Altman, thanks to the new board, should have pretty good control of the non-profit).

They’re trying to reduce Microsoft’s share:

OpenAI and Microsoft have been wrangling for months over amendments to their existing agreement, including changing the 20% revenue-share figure and how long Microsoft will retain the right to use OpenAI’s intellectual property. [...]

One part of OpenAI’s plan involves reducing the percentage of revenue it shares with Microsoft. In financial projections OpenAI shared with investors, that percentage would drop by at least half by the end of this decade. [...]

In an existing deal, OpenAI agreed to share 20% of its revenue with Microsoft through 2030—a year in which OpenAI has projected it would generate $174 billion in revenue. In recent weeks, though, OpenAI told some potential and current investors that by 2030, it would only share 10% of revenues with commercial partners including Microsoft

Microsoft hasn’t signed off on the plan, and negotiations are ongoing.

And of course, Musk will keep suing:

“OpenAI’s announcement is a transparent dodge that fails to address the core issues: charitable assets have been and still will be transferred for the benefit of private persons, including Altman, his investors and Microsoft,” Marc Toberoff, Musk’s lead counsel in pending litigation against OpenAI, said in a statement late Monday.

---

🧪🔬 Liberty Labs 🧬 🔭

🔌🤖⚡️ How much energy does a ChatGPT query use?

We keep hearing about how much energy AI consumes. Gigawatt data centers, etc. That may make you assume that your individual queries are using a lot of energy.

What does the informed-napkin math say?

We find that typical ChatGPT queries using GPT-4o likely consume roughly 0.3 watt-hours, which is ten times less than the older estimate. This difference comes from more efficient models and hardware compared to early 2023, and an overly pessimistic estimate of token counts in the original estimate.

For context, 0.3 watt-hours is less than the amount of electricity that an LED lightbulb or a laptop consumes in a few minutes. And even for a heavy chat user, the energy cost of ChatGPT will be a small fraction of the overall electricity consumption of a developed-country resident. The average US household uses 10,500 kilowatt-hours of electricity per year, or over 28,000 watt-hours per day. [...]

The training runs for current generation models that are comparable to GPT-4o consumed around 20-25 megawatts of power each, lasting around three months. This is enough to power around 20,000 American homes. Since ChatGPT has 300 million users, the energy spent on GPT-4o’s training run is not very significant on a per-user basis.

You can see the detailed math and assumptions in the appendix of this post.

Of course, this is changing fast.

New models are released, GPT-4o itself keeps getting updated and optimized. New hardware is coming out, and as Nvidia Blackwells eventually make their way to inference, the energy used per token will fall dramatically.

But there's an opposing trend: The number of tokens generated is rising fast. More people are using LLMs, and they’re using them more often. Reasoning models generate a lot more tokens per query (o3, DeepSeek, etc).

So yes, total AI energy demand is going up fast.

But your individual ChatGPT queries, especially with vanilla GPT‑4o, are low-impact. A Deep Research query that runs for 30 minutes certainly uses a fair amount of electricity, but that will go down over time.

🔔 The Real Bottom Line ＝ Value?

Knowing what something costs (in money or energy) is not very useful on its own.

What do you get in return?

The utility of tools like Google Search and AI Chatbots is incredibly high. The ROI on whatever small cost they have is some of the highest of any tools available to us right now.

If we’re looking for places to cut, these should be among the last things on the list.

🇺🇸🐷 CRISPR Pigs Greenlit for U.S. Market by FDA 🐖✂️🧬

There are about a billion pigs on Earth. Most are in China, and about 80 million are in the United States:

Most pigs in the US are confined to factory farms where they can be afflicted by a nasty respiratory virus that kills piglets. The illness is called porcine reproductive and respiratory syndrome, or PRRS.

A 2023 report estimated that pigs account for 34% of all meat consumed.

A few years ago, a British company called Genus set out to design pigs immune to this germ using CRISPR gene editing. Not only did they succeed, but its pigs are now poised to enter the food chain following approval of the animals this week by the U.S. Food and Drug Administration. [...]

Genus edited pig embryos to remove the receptor that the PRRS virus uses to enter cells. No receptor means no infection. [...] the pigs appear entirely immune to more than 99% of the known versions of the PRRS virus, although there is one rare subtype that may break through the protection.

Will this be the first big commercial moment for CRISPR?

There’s a chance the Genus pigs could turn out to be the most financially valuable genetically modified animal ever created—the first CRISPR hit product to reach the food system. [...]

Culbertson says gene-edited pork could appear in the US market sometime next year. He says the company does not think pork chops or other meat will need to carry any label identifying it as bioengineered. "We aren't aware of any labelling requirement," Culbertson says. (Source)

Humanity has been genetically modifying its crops and domesticated animals for millennia. Tools like CRISPR just make us much better and more precise at it. But cross-breeding/selective-breeding has also been very effective at massively changing species from their starting points (look up wild corn to see what I mean 🌽).

---

🎨 🎭 Liberty Studio 👩‍🎨 🎥

📺 ‘Andor’ Season 1: Turns out Star Wars isn’t completely dead! 💀

🟢 NO Spoilers Below

After 9 episodes, I’ve now seen enough of Andor season one to comfortably recommend it. And this is from someone who gave up on Star Wars years ago. I couldn’t even make it through The Rise of Skywalker, it was so bad...

Andor is a prequel to Rogue One, or more precisely, it tells the backstory of one of the main characters. But there’s important context to understand before you pre-judge the quality of Andor by association:

Originally, Rogue One was directed by Gareth Edwards. However, during post-production, the film encountered substantial issues that led to extensive reshoots. Tony Gilroy was brought in to fix things.

He rewrote key parts of the script and directed significant reshoots, particularly in the third act. Gilroy described the project as being in "terrible, terrible trouble" when he joined, emphasizing the need for major surgery.

Personally, I remember Rogue One being kind of a mess, with some very strong parts, some beautiful visuals, and some powerful moments… and now I’m that I’ve seen most of Andor, I suspect that Gilroy was responsible for most of the parts of Rogue One that I liked (his style is recognizable).

Andor is Gilroy creating a kind of HBO-style show within the Star Wars universe, grittier, more adult, and with environments that feel lived-in and real. You rarely get the feeling they’re in front of a green screen (or even “The Volume,” their 360° LED soundstage).

But because they took the time to do it right — building intricate sets, designing great props and costumes, filming on location in Scotland, etc — the show was very long and expensive to shoot. What was originally planned to be a 4-5 season run has been shortened to just two seasons, with the second recently released.

As a fan of old school sci-fi (Asimov, Heinlein, Banks, etc), I’m particularly impressed by how some of the mini-arcs feel like they could be great short stories. The ‘Narkina 5’ concept that begins in episode 8 is a great example of that, as is the heist subplot.

Aesthetically, it’s some of the best visuals in the SW universe. The designs! The use of different color palettes! Mon Mothma’s apartment on Coruscant looks great. Even Karn’s mother’s place looks better designed than the sci-fi apartments from 'The Fifth Element’.

I haven’t finished S1, but I’m looking forward to it, and I hope they kept this level of quality in S2. It’s not the best show ever, but it’s a surprisingly fun ride in Star Wars with fresh stories and no Skywalkers in sight (yet?), something that seemed impossible.