Hi Multis
Anthropic's boss is afraid of frontier AI models, but so far, they have not killed the Overview Of The Week, so we are good up to now. :-) Let's go!
Articles In The Past Week
After a few weeks with many articles, this week was calm. This is just the second article.
The previous one was about Axon. You can find the analysis and the answer to whether the stock is a buy now here.
A New Potential Multibaggers Pick!
So, why only one previous article?
Let me start with the minor reason. I was at the Portfolio Day in Brussels on Wednesday, a conference for investment professionals. You can see a picture here from Brussels, together with Pieter Slegers (Compounding Quality), Kevin Schoovaerts (Atomical Capital), Sam Hollanders (Chess Capital) and Jurgen Vluijmans (JVL Communication), some of the people I go to Omaha with every year.
From left to right: Sam, Jurgen, Kevin, me and Pieter.
But that’s not the main reason that I only published one article. It’s much more exciting! I have been working on another Potential Multibaggers pick! I still have quite a bit of writing to do, but it will be published in the coming week. If I had to guess, I'd say Thursday.
You can start guessing what the pick is! :-) Hint: it’s on the Stocks On My Radar list.
Memes Of The Week
I made the first meme myself. I have often been made curious when I hear an analyst talk about his/her model. But when I see it, it’s just junk.
For Duolingo, social media has always been a great driver for the brand. For a while, it was less funny and less unhinged, but it seems management turned that back. When Apple released its iPhone Duo, this was the reaction Duolingo’s X account posted.
Interesting Podcasts Or Books
This week, I (re)listened to multiple podcast episodes about the new Potential Multibaggers pick. There was also one that made me add a stock on the Stocks On My Radar list (see below).
But I also listened to Leandro of Best Anchor Stocks, my second newsletter that Leandro fully runs. If you don’t know Leandro, there’s a reason I started working with him almost 5 years ago. He’s very smart, does outstanding deep research and he can explain what’s important and what’s not. He did so in the most recent Chit Chat Stocks episode. He talks about two subjects: BBB Foods, a stock Leandro wrote about and invested in in January of this year, and which is up 50% since then.
The second part of the episode is about the robotics industry. If you haven’t read Leandro’s deep dive about robotics yet, I highly recommend it.
You can listen to the interview with Leandro here.
The markets in the past week
What did the indexes do this week? Most of us will guess the market dropped but by how much?
As you can see, the Russell 2000 was by far the biggest loser, with 2.41%. The S&P 500 was down 0.66%, and the Nasdaq fell 0.66%.
The Russell 2000 dropped much more than the Nasdaq and the S&P 500 because of rising oil costs. The Russell 2000 has more industrial companies, which means greater exposure to oil, which has been rising quite a bit recently. Those companies also face greater fear of an economic slowdown than companies fueled by AI.
The Greed & Fear Index dropped further into FEAR territory.
Quick Facts
1. Anthropic, OpenAI and Elon ask AI Slowdown
It was all over the news recently. The 27-year-old British researcher Jacob Coxon resigned from Anthropic after just 4 months and wrote a post about it. It starts like this.
In the thread that follows, he accused both Anthropic and OpenAI, his former employer, of irresponsible behavior. With self-improving AI, he argues the genie is out of the box because developers don’t know how to control it. As you can see, the post got more than 170 million views. He now has 290,000 followers, but before the post just a bit over 2,000. This post already caused a stir.
Four days later, Anthropic’s founder and CEO Dario Amodei published an essay titled We Must Pace the Frontier. In the essay, Amodei argues frontier labs should slow the pace of innovation because safety work is falling behind.
To start with, Amodei proposes permanent outside evaluators with deep access to AI labs. Next, he wants frontier companies in democratic countries to agree on safety standards and limits in AI progress, checked by regulation. The end goal is international collaboration, including China. I can tell you: this won’t happen. It’s impossible. On top of that, the timing makes it impossible to ignore the bigger picture.
Now, first, before the rest, many people in AI have been warning about the dangers of the technology. Amodei has been one of them for years, and the same can be said about Elon Musk and pioneer Demis Hassabis, the founder of DeepMind, which was bought by Google.
But there’s another context as well. Open-weight models are improving fast and getting dramatically cheaper. As Guillermo Rauch posted recently, open models went from 28.4% to 62% of token volume through Vercel’s AI Gateway in just two months.
I mentioned earlier in my Duolingo article that the company could bring down its AI costs by 99% compared to the beginning, also by switching to open-weight models. Frontier model tokens are very expensive.
But of course, that raises a question. When the leaders of the most capital-intensive closed AI labs ask to slow down together, who benefits most? I had to think, as so often, of the quote from Charlie Munger.
OpenAI, X and Anthropic have by far the most money, together with Google and Meta. With Anthropic’s pending IPO, it could add another $100B of cash to its balance sheet. Regulation and controls favor those with money, not the new players.
On top of that, we have seen this move before. In March 2023, just four months after the launch of ChatGPT-4, Elon Musk signed the open letter calling for a six-month moratorium on training AI systems more powerful than GPT-4. The letter talked about an “out-of-control race” and argued that development should stop until the risks were better understood.
Four months later, Musk launched his own AI company, xAI and another four months later, Grok was released, Musk’s own frontier model. In other words, it’s not crazy to think that it was developed in the same period Musk was publicly calling for the industry to pause. That could make him catch up, obviously.
There’s also an economic angle. It might slow down the capex race between the frontier models. Is this what Amodei wants his investors to believe? Regulation would definitely be hard for smaller competitors and for open models. Don’t forget that many of the open-weight models come from China (Kimi, DeepSeek, Alibaba’s Qwen, ...), which makes it easier to convince lawmakers to deem them ‘unsafe’ until proven otherwise. That clearly benefits the big frontier models.
And then there’s a legal argument which makes a lot of sense. The Hugging Face hack probably scared these companies. What if the bots had hacked a bank or a healthcare provider? What would the implications have been? For the internet, Section 230 protects internet platforms from much liability for third-party content. AI models don’t have such protection.
So, if these companies can show they use outside evaluators, document risks, and follow recognized safety procedures, it could help if something like a bank hack and theft happened. The companies can then claim they exercised good care and, therefore, are not negligent.
Now, all this doesn’t mean that AI has no potential dangers, but the accidents that have happened so far (see the Hugging Face hack I wrote about last week) are still contained. But also: I don’t believe that frontier models will just stop developing new models and that’s also not what Amodei says Anthropic will do. So I don’t think this has a chance. And I don’t think safety is the main reason. Control is. And if you think this is new... This is from more than 7 years ago, when Amodei still worked at OpenAI.
PS: Just before I published this, I saw Mark Zuckerberg came out with a statement that AI development needs to move faster. So far for the American AI consensus.
The new pick will be released on Thursday. If you want it from the moment it’s published, this is the time to join as a paid subscriber.
Below: why I think investors should skip the Anthropic IPO, the pick that just launched in the US (the first reactions are great), the pick that’s taking on NASA (and why I like the move), the new name I just added to my radar after one podcast, and much more. Don’t miss this. Don’t miss the pick!