ZD

Added more $META.

Added $META after the close yesterday.

Also sold my Nasdaq100 ETF, which was about 5% of portfolio, now in cash. Sold at a modest 8% gain, but I had bought it with money from selling $TSLA which is down about 10% in the meantime, so combined that's not too bad in terms of capital preservation and some appreciation. Also, 8% gain while very modest is the average yearly return for s&p500 on a historical basis, so realising in in a month or two is not too bad. Reason to turn that into cash is that I do not feel super comfortable with stock running like crazy in the past couple weeks while Iran situation still not resolved and its impact on the world economy still not fully realised. This is of course balanced by the ongoing AI boom, following the inflection for AI agents in the end of 2025. Still about 94% invested, 6% in cash. Biggest challenge now - I cannot keep cash and have always stayed 100% invested so far. I will try hard to change that now but not sure I'll succeed for long...

Added some $SE and keeping some cash for earnings season.

Trimmed 15% of my $AMD position at 75% gain.

Thanks for sharing. I did exactly the same, all documented on my account. I am thinking about it this way - there were many races before and Elon was winning a few of these races. There is only one race now - AI. And in this race Elon is now competing with the most formidable competitors. He is also loosing this race right now because he was late and because he doesn't have the resources that the others have - even OpenAI and Anthropic are raising way more money than XAI, not to mention the vast resources of Google, Meta, Microsoft, Amazon, Nvidia. Elon's only chance is to focus on XAI, do everything to catch up and stay competitive. This is the right thing to do for his companies as a whole and, in my view, for the world. However, doing everything it takes will have its cost and Tesla shareholder value is not something Elon will give priority to or refuse to sacrifice if necessary. And rightly so. In addition, for lack of focus or something else, he did some missteps with Tesla along the way.

After holding for well over half a decade – and accumulating through dollar-cost averaging over the years – I have sold the vast majority of my 6k $TSLA shares in recent weeks. I purchased my first few shares after Model 3 was presented, as its possible impact was (almost) obvious. I'm still bullish on the company long term, but in the short to medium term I’ve lost confidence in meaningful upside. Even if Robotaxis were to scale tomorrow, much of that success is already priced in, given the sky-high P/E ratio. The potential downside now seems much larger, as we are running on fumes. Producing and selling cars appears to have become an afterthought, as Elon has bet everything on autonomy – which may well prove to be the right call in the long run. But as of now, I see too many indicators that the Robotaxi software simply isn’t ready. I don’t want to see piles of Cybercabs sitting on Tesla lots while the share price evaporates into thin air – and management stays traditionally mum. Optimus has yet to prove it's for real, and competitive. I'm tired of empty promises and dangling carrots. Risk-adjusted, TSLA is way too hot for me now. We’ve been through variations of this before, but it didn't fazed me as much because the trajectory was clear and I believed Elon had shareholders’ backs. That trust, however, has taken a major beating over the past few years. Elon’s shenanigans – selling Tesla stock on the open market to buy Twitter, funneling money to xAI while granting Tesla only a tiny stake at a sky-high valuation – are just two examples. Then there’s the apparent desire to merge @SpaceX with Tesla, for reasons that are certainly good for Elon, but doubtful for Tesla shareholders. Elon’s utter lack of enthusiasm on the recent earnings call sealed the deal for me. Trust is hard to gain and easy to lose. And even though I can’t fully articulate it, I’ve lost confidence in Elon’s willingness to make Tesla shareholders whole. Honorable mention to the ever-bullish $TSLA accounts on X, whose opinions are based on pure fantasy, utterly unmoored from market realities. These guys have blood on their hands, misleading retail investors, and they know it. Any pushback gets the standard response, “ok sell your shares then!” Alright. I hope this is temporary and things work out in the end anyway. I still hold a few hundred Tesla shares and will continue to do so for nostalgic reasons – maybe things will work out after all. I’m still rooting for the team, and I might buy in again later. I still love the cars, even though the model range, too, could use some attention. I’ll also continue to hold a few hundred shares in a family portfolio I manage. This might turn out to be a stupid call of epic proportions right before a wide Robotaxi launch – akin to selling Apple before it went parabolic – but at my age I want to sleep without too much worry. Over recent months, Tesla has given me more worry than pleasure, so it’s time to reallocate. Holding Tesla was fun when it felt like we were a band of visionary pirate underdogs, with Elon as our lead cheerleader. But something has changed. @elonmusk has new toys, and maybe we’re now part of the establishment – and anyway, there’s no need to be married to a stock. For what it’s worth, it’s been an interesting ride. And anyway, isn’t it all about the friends we made along the way? 🐟🐠🐡🦈🐠🐟

Yes, I interpret it as Elon talking about "that software" being v15 - that he thinks doesn't make sense to scale unsupervised when they have a much safer version in v15 upcoming. This is a bit of a nonsense, because there will always be a new and better version coming, but I think this is his way of saying not to expect large scale unsupervised deployment this year, which matches his statement that there will be no material revenue from unsupervised this year. I've been wrong with Tesla on many occasions before (too optimistic) so I may be wrong here again (too pessimistic).

@cybr4747 do you interpret this differently?

At this point, no regrets for selling out of $TSLA. FSD v14.3 was supposed to be "the last big piece of the puzzle". Sounds like there's a new puzzle now.

Elon basically confirmed it at minute 33 - no large-scale unsupervised FSD before v15, v15 landing hopefully end of this year or early next year. Knowing Elon's timelines, earliest - early 2027. Then, as a major architectural re-make, it needs a few month of validation and ironing out rough edges. Bottom line - no large-scale unsupervised FSD before mid-2027. This is too late. Nvidia and Waymo aren't standing still either.

@cybr4747 this was reiterated on the call. can you show me in the transcript where they said they specifically needed v15 for scaling unsupervised?

This has been my investment thesis this year, following the inflection for agents in December 25, the winners will be the companies with the most compute - they will be able to run the best models, both for their customers and internally. The bottleneck is no longer the model performance (which is already great and improving), it is how much compute you can give to the model to perform to the best of its abilities. Hence, the big majority of my portfolio is in the compute rich $META $GOOGL $AMZN $MSFT $ORCL and the chipmakers $NVDA $AMD $MU. The selloff triggered by the big capex announced in end January was really a gift - this capex will have great return for the hyperscalers because not only they will put to use all the compute they can get but will also have pricing power as demand for compute will exceed supply for the foreseeable future. Same for the chipmakers - there is no end to this "cycle" in sight. That's my thesis anyway, can be wrong, we'll see.

Added $MSFT $ORCL

Will $MSFT pull a Meta and throw its hat in the ring? I hope so - better for my portfolio and better for the world. I am 100% sure it is better for all of us to have choice between many frontier models, rather than concentrating all the power in the hands the likes of Dario and Sam. But the question is can Microsoft execute?

$META is my second largest position, just behind $NVDA and just ahead of $GOOGL. Because of its undemanding valuation, my investment thesis didn't require Meta to re-enter the frontier model race, but it now seems the are on their way to do it. TBH, I didn't expect Meta to release a model so early that beats Sonnet 4.6 and Grok 4.20, as well as all the leading Chinese models. Meta obviously still behind the 3 leaders (also taking into account the Mythos-generation models) but it looks like they are on the right trajectory. With the insane amounts of AI compute Meta is bringing online this and next year, Meta has a real chance of catching up.

Sounds like great progress but does that mean Unsupervised is expected with v15 (at the earliest)? If that is the case, I expect rough months ahead for $TSLA. I sold out so that may be confirmation bias on my side.

I have shifted my portfolio heavily to the companies that have the most AI compute - this is the reason: AI compute will be limited for the foreseeable future. Not all companies will have access to enough compute to run at max capability the biggest and most powerful AI models. Those companies that have access to enough compute to run the at max capability the biggest and more powerful AI models will have a huge advantage. The public companies with access to the most AI compute are the four hyperscalers (maybe plus Oracle) and Nvidia.

Maybe it doesn't mean anything, or maybe it will only be unsupervised in Robotaxi, but not for private customers. We'll see.

My investment thesis for $TSLA before selling out was that 14.3 will achieve unsupervised after a few point releases. Does this 👇mean Tesla is not preparing to make 14.3 unsupervised for customers? "Improve driver monitoring system sensitivity with better eye gaze tracking, eye wear handling, and higher accuracy in variable lighting conditions"

$TSLA competition is much tougher now than it was a few years back. Then, Tesla was competing against legacy auto which was doomed because of incompetence and inertia. Now, you can argue it is competing against the much more capable China EV makers which are much better at battery tech and, importantly, at software. But even that is not the worst. At a > trillion $ valuation, Tesla is not competing in the auto market. It is competing in AI. Because only leadership in AI can make this valuation attractive. And in the AI race, Tesla is competing against Nvidia, Google, Microsoft, Amazon, Meta and the the so called frontier labs. These companies are led by CEOs that are much closer match to Elon, compared with the retard legacy auto executives. What's even worst is that these companies have vastly greater resources than Tesla to invest in the AI race. And the AI race is increasingly turning into a resources game.

@The_AI_Investor And I am not sure "analysts" have an idea.

"Capex growth to peak" is different from "capex to peak" Can be very misleading choice of words.

Also have some S&P500 and Nasdaq100 ETFs at a total of about 5% of portfolio.

My portfolio now in order of size: $NVDA $META $AMZN $MSFT $GOOGL $MU $XYZ $SE $AMD $COIN $ORCL $MELI $RBLX $DUOL I'm going all in on AI compute - those who sell it and those who have the most of it.