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
Position Journal retweeted
Every auto maker is screwed if that happens, including Rivian. In the last twenty years China subsidized their auto industry north of $200B for the build out and they now have a supply chain that is optimized for their auto manufacturing.
They will continue to do so unless everything is mandated to be manufactured here, in other words, they can’t ship parts from China just to assemble in US and slap on a “built in US”, they need to buy every part and component in the US, including labor.
US cannot compete given we sell EVs within 50-60K range while they sell within 25-30K. We compete domestically only because of national restrictions
Position Journal retweeted
For those consistently concerned about why $NBIS runs to 110-120s and fizzles out back <100:
The price action today, reflects the market expectations of tomorrow.
ARR is not actual revenue, not yet. Contracted power is not energized power, it doesn’t bring in money right now
Expectations are heavily mood and sentiment based and hence you see major outflows and inflows on the regular. If the company had the fundamentals to back it, you’d see less volatility
You might be asking what to do?
If you BELIEVE $NBIS will execute to their promises like achieving the 1GW energized power by year end or servicing existing contracts, then just hold.
If you do NOT believe execution will happen, then sell. This is an overvalued company based on current revenue and net income.
For what it’s worth I’ve been holding since 2021, since Yandex days, so my answer to the question on execution is I’m a strong believer
Position Journal retweeted
Editing my latest Stock Spotlight episode with @PositionJournal on $RIVN.
We cover the R2 launch, manufacturing scale, Volkswagen partnership, software/services opportunity, autonomy, dilution and what Rivian must prove for the stock to work.
Coming this Thursday. 🎙️
Position Journal retweeted
$CSCO
Watched an interview with Cisco CEO, link in reply.
Here's some high level takeaways (I'm still digesting what I heard and will read more):
-Cisco builds the networking hardware that connects GPU clusters for hyperscalers and AI companies, and their proprietary silicon differentiates them from competitors using merchant chips.
I'm thinking less reliance on vendors, standardization across their supply chain and in-house knowledge and customization that would all reduce costs for Cisco and make them cost competitive
They're their own masters.
Robbins: "We're one of 3 companies in the world that can build the networking silicon that's needed to connect these GPUs"
Cisco reported >$2B in AI infrastructure orders from webscale (hyperscalers) in FY2025, with FY2026 guidance raised to >$5B in orders and >$3B in recognized revenue from hyperscalers alone.
As it relates to security,
-With AI agents, security has to happen at the network layer (protects the infrastructure transporting data like blocking bad IPs) because the application layer (protects the software and data itself) security is too slow.
What this means is an additional line of defense one level upstream of the software. This gets interesting as it relates to data sovereignty (eg GDPR)
Robbins: "We’re the only networking company that has a big security business. None of our security competitors have a networking business."
No need to pay two vendors, you can just get Cisco and they take responsibility for two critical functions for a customer: security and infra build out
They're also trying to figure out ways on how to build the future infra as countries enforce data sovereignty and a desire for regional controls (e.g Australia banning <16 year olds using social media).
Cisco is redesigning architectures (historically global/cloud-centric) for fully sovereign deployments
On top of all this, they're actively researching the capabilities and supply chain needs to support data centers in space.
It seems like they're effectively a part of the 'picks and shovel' play for AI, so I'll be keeping close.
What are peoples thoughts on Cisco? Tag some Cisco bulls
Position Journal retweeted
Making real money lies somewhere between "Everything is priced in" and "It's not an efficient market".
Position Journal retweeted
True to my word, I exited $HOOD today completely.
From initial buys around $8 back in 2022 to big averages up in 2024 and selling near the top in 2025, this was one of my best investments.
And I was wrong about it too.
When sentiment was in the gutter for Robinhood at $8, my thesis was two things:
1) Once stock market comes back up, the already captured millennial and GenZ user base will come back
2) Worst case, their UI is SO GOOD that they will be bought out.
Vlad, "the little boy from Bulgaria" as was the meme back then, has turned it into an absolute powerhouse of a financial platform.
Why my exit?
Because as I mentioned in my $META post:
When large organizations pivot and shift strategic objectives (ie attacking other segments like crypto, betting, banking) then CapEx increases massively, markets start panicking, shots at the CEO start occurring again, and money outflows rapidly.
But somewhere in year 3, once Vlad executed to his goals, they'll all come back.
I'll be back to it I'm sure. I'll just let Mr. Market completely misprice this amazing company and then buy back in.
Cutting down my investments just to six companies (soon to be five with $HOOD exit) has done wonders for me. I finally am forgetting to check the stock market daily, like a sane person would instead of being glued to it.
I’ll probably even go down to 4 this year
Position Journal retweeted
🧵3/3
$RXRX + $ABSI or The TechBio Bet
Recursion and Absci are a combined bet to replace the trial-and-error guesswork that makes traditional drug development slow and expensive.
Recursion runs an AI-native end-to-end platform integrating biology, chemistry, and clinical development into a unified intelligence system powered by proprietary multimodal data. In other words, they run a giant AI operating system that maps biology at industrial scale, feeding millions of experiments into machine learning models to find drug candidates that humans may not spot on their own.
Absci uses generative AI to design drugs from scratch, targeting biological mechanisms that traditional pharma has never been able to reach.
Together, RXRX and ABSI are two different paths to the same destination: AI makes drug development faster and cheaper.
Risk: The risk is that the AI platform story has not been clinically validated yet. ABSI carries similar fragility at a smaller market cap, making it even more sensitive to macro risk-off moves and binary clinical outcomes.
Why it is high reward: Roche, Sanofi, and Merck have already paid Recursion over $100 million in partnership milestones, and two of its cancer drug programs are due to report clinical trial results in 2026. Absci claims its platform cuts development time by 14 months and costs by 75%, with real human trial data on a hair regrowth drug expected by mid-2026.
RXRX: Ended 2025 with $754M cash equivalents. Runway extends into early 2028
ABSI: Similar runway to H1 2028
Position Journal retweeted
Give it a read via following
open.substack.com/pub/sixsig…
Position Journal retweeted
📝Here are 4 HIDDEN lessons that you missed from the Druckenmiller interview that can make you a better investor.
I went a layer deeper so you don't have to.
1/4🧵
LESSON 1: Investors DO require deep domain knowledge for stock picks, but 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 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 and profound goes much deeper than what you see in the 10K/10Q. Build your circle before betting big.
Position Journal retweeted
@mattshumer_ I strongly disagree with your post. I am curious what you think of these counter points since many people seem frightened by your take.
1⃣ The COVID analogy isn't relevant in this context. COVID had verifiable body counts from day one with first confirmed death in Wuhan in January 2020. By end of March CDC logged >2,000 U.S. deaths (from 122K cases). Source: CDC MMWR report
AI lacks these clear measures. Productivity gain reports stay unquantified and anecdotal, like the evidence in your article.
2⃣ AI code generates way more bugs than humans. According to CodeRabbit's 2025 study AI code changes produced 1.7x more issues than humans:
"Across 470 PRs, AI-authored changes produced 10.83 issues per PR, compared to 6.45 for human-only PRs. Even more striking: high-issue outliers were much more common in AI PRs, creating heavy review workloads." Source: CodeRabbit State of AI vs. Human Code Generation Report.
In essence, the actual coding was commoditized but the end-result is not driving meaningful results.
3⃣ My final point: According to the St Louis Fed study, employees saw only 5.4% increase in productivity gains (or two hours on a 40 hour schedule) due to generative AI, while on a company-level these gains have not yet materialized. Even the Amazon layoffs Jassy quoted the layoffs were about "culture" and reducing layers and not AI driven.
We need to treat AI as a major technological advancement which shakes up the status quo. This I agree in essence with your article. But the doom & gloom around it, at current state, lacks evidence.
Position Journal retweeted
You’re missing the bigger picture that Apple not participating in the brutal AI CapEx race will give them the last mover advantage. They’ll sidestep all the mistakes that the predecessors committed and potentially save billions
Position Journal retweeted
Wow, that is such a thorough research. I was at Pinterest between 2019 and 2023 so my cost basis is 50.05 unfortunately.
I am also targeting to sell at around 42 and 60, which kind of aligns with your analysis 😁 Hope you will have more $PINS posts. Thanks.
Position Journal retweeted
Dana White talks AI, says Meta just hired 10 people aged 22–28, with an average salary of $65M.
xitter.cf/JustTheFights4/status/…
Position Journal retweeted
Following is my "Blue Sky" thesis for Nebius ( $NBIS ) to reach $1,000 per share by December 2027.
At a current price of ~$100 (Dec 2025), reaching $1,000 implies a 10x return and a market capitalization of approximately $250 billion. This valuation would move Nebius from a "tier 2" infrastructure provider to a valuation peer of top-tier enterprise software monopolies. I share details of valuation and the fundamental changes needed for this below.
This thesis relies on three analytical Pillars:
1. Hyper-Scaling Revenue,
2. Margin Expansion via Software, and
3. Geopolitical "Alpha."
The Core Thesis:
From "GPU Landlord" to "AI Sovereign Cloud"Nebius will reach $1,000/share not by simply renting chips, but by becoming the primary alternative AI ecosystem to the US Hyperscalers (AWS/Azure/GCP).
By 2027, Nebius must evolve from a hardware utility into a high-margin AI Operating System, commanding a SaaS-like valuation multiple on massive infrastructure revenues.
Pillar I: Revenue Hyper-Scaling (The "Shadow Hyperscaler" Effect)
To justify a $250B valuation, Nebius needs to generate roughly $12B–$15B in annualized revenue by late 2027. This requires a Compound Annual Growth Rate (CAGR) significantly higher than the market average.
The Microsoft/Meta Floor: The $19B Microsoft and $3B Meta deals provide a guaranteed revenue "floor." However, to hit the aggressive $1,000 target, Nebius must prove these are not one-off rentals but the start of a permanent migration.
The "Shadow" Capacity Crisis: As LLMs scale to trillions of parameters by 2027, the "Big 3" clouds will hoard their own capacity for internal models (e.g., Gemini, OpenAI, Llama).
This creates a massive vacuum for independent capacity. Nebius captures the overflow demand from Fortune 500 enterprises who cannot access sufficient capacity from AWS/Azure.
Analytical Metric:
Nebius must increase its connected power capacity from ~100 MW (2025) to 1.5 GW+ by 2027, effectively doubling its deployment speed every year.
Pillar II: The Multiple Re-Rating (The Software Pivot)
Infrastructure companies (like REITs) typically trade at 5x–10x revenue. Software companies trade at 20x–30x.
To reach $1,000, Nebius must act like software.
Nebius Token Factory & AI Studio:
The thesis hinges on the success of these platforms. Instead of just renting raw GPUs (low margin), Nebius sells "Inference-as-a-Service" (high margin).
Scenario : By 2027, 40% of Nebius's revenue comes from the Token Factory—managing the serving of models rather than just the training.
Effect: This shifts gross margins from ~30% (hardware rental) to ~65% (software platform), justifying a valuation multiple expansion from 15x to 25x.
The "SaaS-ification" of Compute:
Investors stop viewing Nebius as a "box mover" and start viewing it as the "Snowflake of Compute"—an essential data platform layer.
Pillar III: Geopolitical "Sovereign AI" Moat
Nebius has a unique advantage: it is not a US Big Tech company.
Data Sovereignty: European and Middle Eastern governments/enterprises are increasingly wary of the US Cloud Act and data dependencies on Microsoft/Google. The "Swiss Bank" of AI: Nebius (HQ in Amsterdam) positions itself as the neutral, sovereign alternative for AI infrastructure.
This allows them to win exclusive government contracts in the EU and potentially regions like the Middle East that the US Hyperscalers cannot easily service due to geopolitical friction.
The Math: The Bridge to $1,000
2027 Revenue : $12.5 Billion
Rev Growth Source: Massive Software/Inference Adoption
Gross Margin: 60%+ (Platform/Software Led)
Valuation Multiple: 20x Sales (SaaS Premium)
Implied Market Cap: ~$250 Billion
Critical Assumption - Dilution Control:
This math assumes Nebius funds its massive CapEx (buying GPUs) primarily through debt (project financing backed by the Microsoft contract) rather than issuing new shares.
If they dilute shareholders significantly to raise cash, the $1,000 target becomes mathematically impossible.
Conclusion:
For Nebius to reach $1,000/share, it must execute a "perfect pivot": utilising the cash flow from its Microsoft mega-deal to fund the construction of a high-margin proprietary software platform. It must stop being a "landlord" for Nvidia chips and become the "operating system" for the global AI economy.
$NBIS (CMP $83) - Deep Dive Investment Thesis with Bull, Bear & Base case price scenarios! 🧵
Nebius is building the picks & shovels for the AI revolution. After a brutal 30% drawdown, the risk/reward has never been more compelling. Here’s why this could be a generational opportunity:
1/ THE SETUP
•Current: $83
•Analyst avg target: $157 (88% upside)
•Bull target: $211 (153% upside)
•5 Buy ratings, 0 Sells
•Recent 30% drop created entry point
2/ THE OPPORTUNITY
AI infrastructure is the most supply-constrained market in tech. Hyperscalers spending $405B in 2025, projected $500B+ in 2026. GPU compute is gold, and NBIS is positioned as THE third-party provider.
3/ REVENUE TRAJECTORY
•Q3 2025: $146M (+355% YoY)
•2025E: $500-550M
•2026 target: $7-9B ARR
•That’s 1,300-1,600% growth potential over 12 months
4/ CONTRACT BACKLOG
•$3B Meta deal (deploying Dec ’25)
•$17.4B Microsoft Azure agreement
•2.5 GW contracted power, 1 GW connecting by end 2026
•Customer diversification = reduced risk
5/ THE BULL CASE (PT: $250-300)
✅ Hits $6B+ revenue in 2026
✅ GAAP EPS: $6+
✅ Captures 3-5% of AI infra market
✅ Profitability ahead of schedule
✅ Stock re-rates to 40-50x forward earnings
6/ THE BASE CASE (PT: $140-170)
→ $4.5-5B revenue in 2026
→ GAAP EPS: $4-5
→ Minor delays but solid execution
→ Trades at 30-35x forward earnings
→ Meets analyst consensus
7/ THE BEAR CASE (PT: $50-70)
❌ Only $3-4B revenue (execution issues)
❌ Power constraints delay capacity
❌ AI spending slowdown
❌ Dilutive capital raises (25-35%)
❌ FCF burn accelerates to -$8-9B
8/ KEY RISKS
•Burning $658M FCF currently
•CapEx raised from $2B to $5B
•P/E: 97x (stretched valuation)
•Power constraints & supply chain issues
•Profitability timeline uncertain
9/ WHY NOW?
1.Stock down 30% despite $3B Meta deal
2.Testing technical support at $82-83
3.Oversold conditions (WR signal)
4.No analyst downgrades post-earnings
5.2026 is THE inflection year
10/ THE VERDICT
This is a binary bet on AI infrastructure scaling. If NBIS executes, $200+ is realistic by late 2026. If they stumble, $50s possible.
Risk/Reward: 3:1 upside/downside
Time horizon: 12-18 months
Position sizing: 2-5% max (although my size is closer to 40% of my portfolio)
11/ MY TAKE
At $83, the market is pricing in significant execution risk. But with $20B+ in contracts and analysts maintaining Buy ratings, the setup favors patient capital.
Not financial advice. DYOR. 📊
Position Journal retweeted
I am revising my $NBIS bear case valuation metrics but keeping my bear case price target of $150
The main reason for this revision is that after completing my deep dive on CEO Arkady Volozh revenue guidance over 44 quarters while at Yandex there has never been a single year from 2011-2022 where he did not meet or exceed guidance (even during COVID in 2020 or the Russia/Ukraine conflict in 2022 where guidance was withdrawn but ultimately still met or exceeded). Nebius guidance was just confirmed Q3 setting the stage for 13 years straight. No other major tech CEO has ever achieved this—not Huang, Jobs, Su, Pichai, Nadella, Cook, Zuckerberg or Musk come even close.
Nebius has guided for 7B-9B ARR by end of 2026 but during the Q3 call also indicated that OVER HALF of this has already been contracted for. Considering this head start on the ARR and Arkady’s flawless history of meeting or exceeding guidance even during extreme duress, I just don’t think I can reasonably justify assuming even in a bear case scenario that Nebius will come in 2 billion under the low end of the guidance (my prior bear case disaster scenario).
So instead my bear case will assume the lowest end of the offered guidance at 7B ARR.
Now let’s assume a bear market ARR multiple akin to the lows seen in the bear market of 2022. At that time no public company growing at 100% ARR YoY was trading at a multiple less than 7X. Per Grok during the worst trough of the 2022 bear market —“No public SaaS or cloud companies trading at 100%+ YoY revenue growth (measured via LTM or forward guidance) dipped below 7x EV/forward revenue multiples during the trough (roughly Q3-Q4 2022).”
So we have 7X 7B ARR or 49B valuation.
Of course I will continue to view the ~5B cash and all subsidiaries (6B+) purely as capex for the buildout.
I will continue to assume FULL dilution (~22M more shares based on the convertible bond sale) plus full dilution for the 25M ATM optional shares which gets me to ~297M shares outstanding.
I WILL ADD 10% MORE DILUTION: my prior bear case did not assume additional dilution. Nebius management has indicated a willingness to sell their subsidiaries for non dilutive capex but let’s say they don’t sell them next year. I will add an additional 10% share dilution of nearly 30M more shares (note that this is hypothetical and there is no such plan to do so yet) to take the shares outstanding up to 326.7M.
49B market cap with 326.7M shares outstanding comes to $150 per share
