Discovering the most undervalued companies
Wall Street
Joined July 2025
- Tweets145
- Following2.2K
- Followers241
- Likes2.8K
Following the recent sell-off, memory card producers look dirt cheap again. Concerns about debt financing and the profitability of AI companies are justified, but in my view, the pendulum has swung too far against memory producers again.
The market caps of SK Hynix and Samsung are currently below their expected cumulative FCF over the next three years. They should have largely earned back their current market caps by the time additional production capacity—their own expansion plus CXMT’s—really kicks in from 2029 onwards. There is effectively no terminal value priced into today’s valuations. Micron is slightly more expensive, but for good reason. Visibility on pricing and orders remains strong for 2027 and 2028, and demand keeps growing.
In my view, this offers an attractive risk/reward: limited downside and plenty of optionality from further AI acceleration, from agents and data centers to physical AI. At the same time, these positions could partly serve as a hedge against a potentially highly disruptive, self-reinforcing acceleration in AI.
It reminds me of late July, when these stocks fell to similar valuations following a sharp sell-off driven by market panic and deleveraging at hedge funds (Situational Awareness, Leopold Aschenbrenner) and in the Korean stock market. Once again, there are rumors of a hedge fund (VARA) running into trouble, alongside a bloodbath in Korean equities. I think these stocks will rebound sharply this week, possibly as early as tomorrow.
What makes it even more interesting is that you can buy both companies at a significant discount to the multiples shown on the image:
- SK Hynix through a holding company, SK Square, trading at a 46% discount.
- For Samsung you can buy the preferred shares trading at a 21% discount to the common shares.
I bought both SK Square and Samsung (preferred shares) last night on the Korean market. I’m considering buying Micron too.
I wanted to let my followers know that I bought Peugeot Invest yesterday at €49.50.
The NAV per share was and still is around €138.
The discount to NAV was around 64.5% at the time and still is around 62% as of today.
This steep discount is absurdly large. Stellantis was a drag in the past, but it’s only 17% of NAV anymore. Even if you value Stellantis at zero, the discount is still larger than 50%!
Investment funds and other investments performed well in H1. The portfolio has become much more diversified.
I haven’t got the time to write a deep dive as I’m leaving on holidays. I’ll probably write a deep dive in a week or two.
This hedge fund just replaced its entire team by AI agents
An eight-person trading desk at a hedge fund costing roughly $5 million a year has been replaced by four AI agents costing just $30,000 to $40,000.
The new “team” behind Brian Kelly’s trading firm Bracket22 is almost too good not to share. He gave each of his AI agents names and functions:
- Houston is the mission controller, bringing all the information together.
- Desmond focuses on quantitative strategies.
- Steffi identifies technical patterns in price charts.
- Doocy is the professional contrarian. His job is to attack every investment thesis developed by the other agents and expose its weaknesses.
The agents monitor crypto, equities and commodities around the clock. Kelly estimates that they have made him at least ten times more productive, although he still makes every final buy and sell decision himself — or, as he puts it, he remains “the meat in the seat.”
Bracket22 only manages Kelly’s own money, the system has been running for just six months and no actual performance figures or benchmark have been disclosed.
The most interesting claim is not that the agents find more winning trades. According to Kelly, their main contribution is helping him avoid bad decisions.
That sounds plausible. In investing, AI’s greatest value may not lie in discovering some secret piece of information. It may lie in enforcing a disciplined process, challenging assumptions and fighting confirmation bias — 24 hours a day.
This is either the future of investment research or a very polished AI experiment.
Probably a bit of both.
Meta rose 11.4% yesterday, following the success of its AI-assistant Muse.
It has been one of my core portfolio positions since 2022. I knew the stock was cheap, but even I was surprised by just how cheap it had become: at its recent low, Meta traded at only 13.8x forward earnings.
That was for a company with one of the strongest competitive moats in the world, while still growing revenue by close to 30% year over year.
I considered writing about Meta on my Substack, but it never quite seemed to fit the “deep value” label.
Perhaps 13.8x was deep value after all.
This stock almost doubled in two days. Now it's back.
In May, a buy recommendation sent this small cap soaring. The share price doubled in just a few hours!
Four months later, the excitement has cooled down, but the bull thesis remains.
The stock now trades below book value, while the underlying business offers exposure to several structural growth themes — including physical AI, sensing and specialty semiconductor demand.
I sold into the spike. Now I've bought again.
I think the current setup offers a very interesting risk/reward.
Full deep dive — and the name of the stock — later this week. Make sure to subscribe.
One of my favourite writers and thinkers, Nassim Nicholas Taleb, just liked one of my posts.
And yes, it’s actually the real one. Crazy stuff.
I want to show my colleagues, all big fans of him, but I can’t. They don’t know I have this account 😂
The greatest value investor who ever lived, Warren Buffett, has stepped down as chairman of Berkshire Hathaway.
Over more than 60 years, he transformed Berkshire into one of the largest companies in the world.
Words fall short of expressing my deep admiration for Warren Buffett—as an investor, a businessman, a philosopher, and a human being.
"Father time always wins. He has, however, been generous with me."
~Warren Buffett
Read my latest article:
Vivendi: A Double Discount on Universal Music Group 1000deepvaluestocks.substack…
#Vivendi $VIV.PA #UMG #Universal $UMG.AS #Bolloré $BOL.PA
Deep Value Stocks retweeted
“One of the reasons why I am in favor of less government is because when you have more government, industrialists take it over, and the two together form a coalition against the ordinary worker and the ordinary consumer.”
— Milton Friedman
@1000ValueStocks is dedicated to proving him wrong.
Value investing has certainly gotten more difficult as information became abundant. But I believe value investing still works and sometimes the discrepancy between market price and fundamental value gets so wide that the potential reward is far greater than the risk.
Value investing worked because information was scarce. Today, if a stock is cheap, there's a reason, and the reason is terrible:
Alix Pasquet (@alixpasquet), managing partner of hedge fund Prime Macaya Capital Management, explains:
"Value investing taught you to look at things bottom up. You look at the business, and that's it. And that wasn't because Graham or Warren were bad at understanding an industry. It's because back then, they didn't have access to the information we have now."
"One of the reasons value investing worked really well is you could buy something cheap that was very good, and no one knew it was good, because there wasn't enough information on it."
"But today there's so much information that if something is cheap, usually there's a really good reason why it's cheap, and that reason is terrible."
"You're not gonna have a really good athlete trade cheaply for a short period of time. People are gonna know."
"I look at certain long-only, quality-focused investors and say, I bet you 100 bucks to a hat pin that Visa or MasterCard are in that portfolio. And yes, it is. The cat's out of the bag."
"But guess what happened this year? Quality took a dive, a really big one. It's gonna take those guys a couple years, if not five, to dig themselves out of that. It's gonna be painful."
Deep Value Stocks retweeted
S&P 1500 Software revenue per employee has gone parabolic.
If you are looking for evidence that AI is starting to impact the real economy, this is exhibit A.
ChatGPT for finance just dropped.
And it’s a gamechanger. Insane value.
For the price of $100 you can’t even get a call with a decent financial analyst or a consulting firm. With ChatGPT finance, $100 can create at least tens of quality models or research reports that would require a full time employee a month of work.
Do not sleep on this. Also, stay away from all consulting & data firms.
openai.com/index/introducing…
