Mag7 Product leader. Concentrated investment portfolio and evidence-based research. Expertise within Robotics, Fulfillment, Supply Chain & Enterprise products
Joined February 2023
- Tweets3.4K
- Following296
- Followers1.3K
- Likes8.9K
Commitment fulfilled ✅
1000 $RIVN shares bought.
If it goes to $12 without any structural news coming from Rivian you have my commitment I will buy at least another 1000 shares😄
I truly believe this is a company that just needs time, they have the perfect plan just needs to execute.
I’m so excited about them I’ve been even keeping an eye on potential roles for me within Rivian and Mind Robotics
Good morning everyone,
Last week was a rough one, I’m personally down -13% for October.
My friend with 20 years experience in the industry put it: “last week was one of the most violent flows I’ve seen since 2025s tariff introductions”
I did some spot checking of my core holdings $NBIS $RIVN $COIN $CRSR and found nothing structurally changed.
But then you might say what about the rising interest rates/yields or the uncertainty in world markets?
And my answer to that is I am not in a state where I can control any of it.
The game I play is to understand deeply the core value offering of my investments and then actually allow them to play out in the long term. Time is the remedy and weapon of choice for me and anyone attempting to react to forces beyond their comprehension is doomed to fail.
Price is driven by sentiment, money in and money out — zoom out and relax. Your conviction can’t come crashing down because some hedge funds are cashing out for reasons you will never know.
Benjamin Graham put it nicely (paraphrasing): if someone knocked on your door today and offered to buy your house for -20% less than its value, would you sell it to him?
No, you know the true value better. But Mr. market will continue to knock every day regardless .
$RIVN calls bought ✅ and shares pending
These prices now have me approaching a “what should I buy?” mindset.
Another hefty loss so far today.
CPI rose 3.7% YoY vs the expected 3.6% driven primarily by gas and energy prices.
Almost everything has been hit hard. My particular interests to add either in shares or calls would be in:
$RIVN ($12 or thereabouts range)
$NBIS (low 180s for calls)
$RKLB (<$60)
$CMC (<$60)
Anyone else eyeing something?
These prices now have me approaching a “what should I buy?” mindset.
Another hefty loss so far today.
CPI rose 3.7% YoY vs the expected 3.6% driven primarily by gas and energy prices.
Almost everything has been hit hard. My particular interests to add either in shares or calls would be in:
$RIVN ($12 or thereabouts range)
$NBIS (low 180s for calls)
$RKLB (<$60)
$CMC (<$60)
Anyone else eyeing something?
These prices now have me approaching a “what should I buy?” mindset.
Another hefty loss so far today.
CPI rose 3.7% YoY vs the expected 3.6% driven primarily by gas and energy prices.
Almost everything has been hit hard. My particular interests to add either in shares or calls would be in:
$RIVN ($12 or thereabouts range)
$NBIS (low 180s for calls)
$RKLB (<$60)
$CMC (<$60)
Anyone else eyeing something?
Send it lower, I’m ready 😎
$RIVN
If it goes to $12 without any structural news coming from Rivian you have my commitment I will buy at least another 1000 shares😄
I truly believe this is a company that just needs time, they have the perfect plan just needs to execute.
I’m so excited about them I’ve been even keeping an eye on potential roles for me within Rivian and Mind Robotics
$COIN
Spot margin, the product that lets eligible traders borrow against collateral to take larger spot positions with up to 10x leverage on major assets, has the clearest near-term path to Coinbase generating transaction revenue.
It multiplies trading volume on the existing Advanced platform while also generating interest on loans
I like the business decisions being made here by the team
YTD now up to +64% with 3 year returns at +277%
Largely no action, I am not a frequent trader or anywhere close to it. My best gains come from doing nothing while time works in my favor.
That being said, in spirit of transparency, here is what I did do:
1. Bought more $RIVN calls for Dec'26 at $17.5 strike
2. Sold 1/3 of $ABSI for 100% gains, now remaining 2/3 is all house money.
3. Cashed in $COIN calls for 128% gain, no change to shares -- still 15% of portfolio
No change to rest of portfolio
As promised, the Q3 Year-To-Date Results with brief commentary.
Year-to-Date: +56%
S&P 500: +13%
--
HOLD:
$NBIS | 46% of portfolio | no action. Fantastic operational performance. On track for energized power of 1GW by year-end, contracted power guidance raised and go-to-market strategy of winning enterprise contracts continue. My long-term thesis also continues to evolve, with AvRide improving its footprint as a robotics operator.
$COIN | 15% of portfolio | no action. Agentic transactions continue to ramp up, product development across companies utilizing x402 continues and all while aided by a tailwind of rising Bitcoin price.
$RIVN | 10% of portfolio | bought some calls, largely no action. Another company in my portfolio with an amazing operational showing. Raised delivery guidance, increased production and continues to meet its long-term partnership agreements with VW, Uber and Amazon.
$HIMS | 5% of portfolio | no action. I think this is now mostly a waiting game for the peptides to formally get added to the 503A list. A 2023 JAMA Health study found that the FDA has historically followed advisory committee recommendations 88% of the time. I think it would be surprising to see the agency go against that kind of majority. That’s a pretty bullish signal in my view.
$ABSI | 3% of portfolio | closed my calls, playing with house money. No one talks about this company on X, but I've been posting my thoughts on it since it was in $3s, I'm up about 180%. AI-assisted drug design is so early.
SELLS:
$CRSR | 6% of portfolio | Sold my Feb'27 $10 strike calls, now its house money. Corsair came in exactly as I expected with improved operating margins, new product launches and scaling direct to consumer. Now I wait for their AI workstations to bring new revenue.
$TOST | Fully sold | I bought at the bottom $25...and sold at the bottom of $25. I rotated it into higher asymmetry like Coinbase and have "lost" gains since its now $36. I still stand by my thesis. Very prudent management and pragmatic approach to growth.
BUYS:
$RKLB | 4% of portfolio | started buying at $77s, then added more in 60s. I hate how high the p/s ratio is (>30) and I cannot find a grounded assessment that justifies it and hence being cautious.
---
Cash is at 4% but this doesn't include about 15% of portfolio that I withdrew and locked away earlier in the year.
$RDDT was at $233 at the time of this post, it's now down -37% to $147.
my statement from last year resulted from 3 facts:
1) Reddit held a 25x price-to-sales ratio at the time -- which is absurd given their competitors ranged from 2-4x
2) The median return of the typical IPO after 3 years is -25% and I couldn’t be confident Reddit doesn’t follow the trend
3) Content licensing by Reddit to LLMs was giving up their only moat and thus risks website visits/engagement long term
Some state you don't need domain knowledge in the field you're investing, but my reading in this Druckenmiller interview is that you in fact do.
You yourself may not need to have it, because that expertise can come from trusted team members rather than personal mastery alone.
Druckenmiller is asked: "Do you have to be an expert ...someone who understands the whole pipeline...to get that right?"
But he rejects personal expertise as mandatory:
"Thank God the answer is an emphatic no."
Instead he stresses reliance on others: "But I've got to have is an expert at Duquesne who is and trust his judgment."
He is not going in blind -- he has a profound understanding of the company either through his own research or his team.
For us without a dedicated team, this simply means we must seek out the experts in the field and study their works and hypotheses and always know before investing big, that there may be an angle we are missing, simply because we are not native to this domain.
But by combining the experts specialized view with our perspectives and thoughts, we may form a complete investment picture.
Happy Sunday everyone😎
$COIN
Fantastic interview as always by The Decoder host @reckless with his guest @eastdakota
Another affirmation by reputable CEO of what I've been saying: agents and micro transactions will capture the digital economy specifically through enabling technologies like x402.
In his words: "We're working with companies like Coinbase and Stripe in order to say how do we make sure that we can allow people who are creating content to say that in exchange for that thousands and thousands and thousands of times more traffic that's going to come to you... you're not going to maybe get kind of the advertising revenue that comes from it, but maybe you can get a fraction of a penny every time somebody actually accesses that information."
Multiple thousands of times more traffic, primarily by agents, who are funded by either end-users or customers to access the creators content.
Further, Matthew also argues that by shifting to a payment-for-access model, the internet can transition from an attention economy built on rage-baiting to a value-based system where content creators are directly compensated for contributing new, true knowledge that fills the "holes in the cheese" of AI models
He even adds how his own local newspaper company is set to now earn more money off AI licensing deals this year instead of digital advertising, specifically because local media is more valuable today rather than volume and scale because the local company has access to the niche data others don't.
$EL up about 29% since my chat with my wife and $NKE down about 31%.
Listen to your wife. Yes this is financial advice.
The inflection point for $RIVN being crossed right before our eyes.
Beating guidance is expected behavior — they must showcase there is pent up demand and secondly prove they can match it.
Things only get easier from here
Rivian set records for both deliveries and production in Q3.
19,248 delivered and 19,751 produced, 24% and 13% above the previous highs from 2023.
And the second R2 shift hasn't even started yet
$RIVN
$NBIS was mid-pack (I counted #13 on the chart) on the ClusterMAX chart for time until a loaded model answers “I’m ready.”
Lambda was first at about 36 seconds. Inferize, which Nebius just acquired into Token Factory, is built for that same kind of wait of reducing the cold start, the gap between GPUs being assigned and the model taking a request, so spare GPUs do not sit idle
A complementary acquisition -- now we need to see how much it cost them.
Let's go!
nebius.com/newsroom/nebius-a…
Short interest for $CRSR decreased from 22% to now 18%.
An orderly retreat by Wall Street as funds re assess Corsairs path to sustained profitability to be more likely
$NKE has a ton more pain ahead while $ADDYY Adidas is considerably outperforming in terms of cultural relevancy and product innovation.
Adidas is killing it with collabs like with Malbon, a golfing brand that sells posh golfing gear and outfit which is especially popular among trendy fashionistas.
They also are excelling at capturing sub cultures. Jennie, a K-pop star who signed a deal with Adidas was one. The other was Simone Rocha, a popular designer who collabed with Adidas and the design shoes sold out within minutes.
In addition their Shanghai line like their Chinese new year jacket was an amazing success. It was limited to only the Chinese market, continued to be sold out, Americans wanted it and started buying from resellers and Adidas capitalized on that finding and brought it to the US market
On the other hand competition is fierce from others: Puma came out with Speedcat line, a new variation of their flat shoes which is also being promoted by Kpop singers like Rose
Nike needs to immediately give customers new products that actually fits the current trends. They need to step up their collaboration partnerships and sign with key stars around the world. They keep selling the has been of the world that newer generations don’t care for like Air Force. They’ve completely missed establishing any cultural relevancy with their products let alone capturing any subcultures.
How I invest into companies where I do not have domain knowledge. Typically I customize the checklist to the stock. This is the version I used on Absci.
case study: my $ABSI bet I made in March where I'm up +240% since buying in $3's and the stock now trades around $9.
Foundational rules:
1. No more than 2-3% risk portfolio.
Remember, you don't really know this space deeply. You may end up betting on the right theme (e.g AI platforms unlocking faster drug development) but end up betting on the wrong horse. I do not know antibody design or prolactin-receptor biology but I know AI/ML enough that the use case for faster drug development is real.
2. Executive team must have deep expertise in the field their company is in.
An AI-native drug company needs people who have shipped drugs and people who have shipped models, sitting in the same room. For Absci I checked that filter in March and I still check it.
Sean McClain founded the company in 2011 and has been CEO since. He built it from a basement lab into a public company with an integrated wet-lab plus generative-design loop.
3. There must be broader political or societal support.
I only buy outside my expertise when the country already wants the problem solved and where Washington typically is in favor. In my case, Make America Healthy Again (MAHA) was the bet where the theme they supported was chronic-disease relief and support and broader efforts to reduce bureacracy and speed up approval timelines. Both exactly what ABSI supported and needed.
4. Bet on the technology, not individual milestones or results.
In the early 2000s PayPal was losing over $10 million a month to credit-card fraud and human reviewers couldn’t keep up with the scale. So Thiel built a machine-learning model that sifted through millions of transactions, flagged the suspicious ones for human review, and turned PayPal profitable for the first time.
This was my bet for ABSI. That their AI platform continues to improve, be smarter, catch the variables that human researchers miss and speed up drug design.
I know full well that some milestone or key reading might be missed by ABSI, but the bet is that by that time they would have amassed an incredible amount of proprietary research data.
There’s an incredible amount of educational content here. Some notes as it relates to $NBIS and @LambdaAPI
The coverage now runs to 323 cloud providers in SemiAnalysis market map (up from 209), with 77 of them put through the same ClusterMAX 3.0 hands-on suite on the providers own hardware and setup.
Nebius: They are one of two Platinum names this cycle, with CoreWeave. Platinum is the top ClusterMAX grade where the managed cluster is treated as something a lab can rent without owning the scheduler (the software that decides which machines run which job, and when)
Lambda kept Silver, unchanged from ClusterMAX 2.0. The nuance though is the tracked providers went from 209 to 323, and only 19 providers earned any medal after they raised the bar, so they held Silver after more competition got added and a higher bar was raised
Now on to the specifics:
Lambda won first place where the measurement is specifically at getting a model loaded and answering “I’m ready” which they clocked about 36 seconds on Kubernetes. That wait is wasted GPU time and Nebius as well as AWS scored over a minute.
Testers also injected fake GPU errors that Lambda cleared within 15 minutes while other peers never even pushed an alert or saw it and left it in the pool (Google) for an operator to intervene
Finally a specific for Nebius was that on a GB300 rack (72 GPUs wired as one unit), testers faked a GPU failure. Nebius saw it immediately and put the node back in the pool by itself. The loop took 8 hours 40 minutes but most other GB300 racks they tested did not auto-return at all.
Theyre already building the closed-loop infrastructure automation needed to operate these increasingly complex GPU clusters without requiring humans to babysit every failure. Thats a meaningful capability as the scale and complexity of these clusters continues to increase.
ClusterMAX 3.0: The Industry Standard GPU Cloud Rating System Returns
In gory detail: reliability, performance, support, pricing
—and, of course, security—
in our most thorough analysis of GPU cloud providers globally
semianalysis.substack.com/p/…
