I aim to make the best video games that push creative and technical boundaries. I'm also deeply into stock investing, cutting-edge tech, and ice hockey.
Rising CapEx can be toxic for some software companies. Token, memory, CPU and GPU card costs are rising — firms that must buy physical hardware or rely on tokens are vulnerable in this market. #TechStocks#Investing#Stocks
The use of #TechnicalAnalysis is likely to increase further as #AI agents trained in the full set of technical analysis rules are deployed for #DayTrading.
$PLTR has dropped 10% since its Q1 results:
It seems stupid. And it is.
Here is why:
1) Palantir can't meet demand.
Jefferies analyst Thill complains that Palantir doesn't hire more people.
Why do that when you can have people selling for you?
Palantir wants to keep elite talents focused on products + key clients.
Meanwhile, the Partner ecosystem sells and deploys to smaller clients.
ie Accenture formed a +2,000 team to deploy Palantir.
2) Results are so insane that investors don't understand them.
85% Revenue growth with margins expanding to 53% GAAP Net Income Margin.
Yes, including SBC.
Yes, this is black magic.
AIP has a perfect structural market fit.
3) Recurring Revenues on long-term relationships mean ENDURANCE.
As it is very hard to accelerate growth, it is very hard for growth to just stop.
You don't easily stop a rocket.
4) "US Commercial Missed."
Bloomberg reported that US Commercial slowed, but apparently wrote the article without listening to the call.
"One US Commercial" client moved to US Government. Without the change, growth would have been 143%, up from 137% YoY in Q4."
Funny and sad at the same time?
5) Palantir reaps all the benefits
While other AI players spend hundreds of billions of dollars building data centers and training models, Palantir benefits from these expenses without incurring them.
Better models make Palantir AIP more valuable.
AIP is the AI infrastructure level that enables an organization to deploy hordes of agents effectively, compliantly, and safely.
"The number of tasks that you can trust to a model without the right harness exponentially declines.
"More tokens mean more slop. And the more commodity cognition you consume, the more you need a system that can prevent the economic harm so you can harness the economic value." - @ssankar
AI labs are in a race to the bottom. Palantir's pricing power increases the more it can unlock critical use cases at scale.
You can build a simple workflow directly with Code.
You don't want to have 1000 agents without guardrails.
6) Analysts keep underestimating the earnings power:
Revenue YoY growth estimates:
+75% in '26
+44% in '27
+42% in '28
FCF YoY growth estimates:
+115% in '26
+46% in '27
+40% in '28
Analysts believe:
- growth is peaking
- margins can't expand further
Given endurance and a focus on expanding the product over sales, I believe both revenue and margin estimates will be smashed.
7) Is Valuation actually cheap?
Palantir trades at ~60x FCF NTM, while more than doubling FCF YoY.
Even assuming that Palantir delivers FCF in line with (low) estimates, I consider that appealing.
eg $CRWD trades at 77x FCF with 22% Revenue Growth at -3% GAAP Profit Margin.
At least 80x EV/FCF deserved?
$PLTR is the professional house painting company.
Anthropic is like painting a house with a toothbrush. Possible but a bad idea.
Open AI is like trying to paint a house with a rake while naked. Ugly and pointless.
$STRC is financializing $BTC using $ETH rails.
Paul Barron says Saylor sounds more excited talking about DeFi than Bitcoin 😭
He’s describing a DeFi credit machine, tokenized securities, and yield coins… all running on Ethereum infrastructure.
So is the real money $ETH, not $STRC?
The AI infrastructure trade is increasingly shifting from training toward inference as models move from development into real-world deployment across the global economy.
$NBIS CEO Arkady Volozh said “most of capacity was spent on training new and new models. But eventually, there will be more and more usage of these models, more and more inference.” That is a major reason why infrastructure tied to AI cloud platforms, networking, memory, and inference optimization continues becoming more strategically important.
The important takeaway is that the next phase of AI demand may not just come from frontier labs training larger models, but from millions of enterprises and applications continuously running those models at scale every single day.
UiPath $PATH keeps getting written off, but I don’t buy it. ARR is growing, the cash cushion is solid and operating leverage is showing. Cloud AI won’t automatically replace RPA—enterprise workflows are hard. Watch Q1 on May 28 and judge by the numbers.
Tom Lee explains his $62,500 ETH price target
“If we clear this Middle East problem and the US economy holds up through higher oil, I think we’re looking at a bull market that could run through 2028. A major move in equities is the setup… and here’s something to keep in mind. Since the war started, the best performing asset in the world — outperforming energy stocks — was Ethereum. It outperformed the S&P 500 by almost 20 percentage points, and you can see it has massively outperformed gold and silver. And if you take a look at Ethereum’s chart over the last 10 years, I think it’s going through a massive consolidation.”
In its first consolidation of 2016, Ethereum went on to rise by 227x. In the second consolidation of 2018 and 2019, Ethereum rose by 54x. Tom points out that Ethereum is in the midst of its third consolidation:
“I think there is a massive move coming in Ethereum, driven by a couple of things: tokenization and agentic AI… I think this means you can get something like a 25x for Ethereum.”
Tom gives a quick overview of the tokenization thesis:
“I think we’re going through an important moment in the financial system that’s not too different from 1971. Tokenization is making almost every asset synthetic, and it follows a roadmap that happened when the US went off the gold standard in 1971. This led to a huge unleashing of innovation and products from money market funds to currency futures to CDOs to indexed futures all because the US was trying to preserve the sovereignty of the dollar when we went off the gold standard. I think that’s happening today because we’re digitizing everything.”
He points to the following quote from JPMorgan CEO Jamie Dimon (formerly one of crypto’s biggest skeptics): “Crypto is better than the current financial system.”
Tom continues:
“I think everyone who’s building in crypto is going to develop these future products — stablecoins, tokenized equities, monetized reputation. It’s also part of the future agentic system.”
On a separate slide he points out all of the things Agentic AI will need that work better on crypto rails. Two are identity and payments. “Agents almost certainly won’t want to use PayPal or Visa or MasterCard to do micropayments,” Tom argues.
Lastly, Tom turns to price:
“Blockchains should gain relevance against against crypto’s store of value, which is Bitcoin. In our minds, the way to think about the future of Ethereum is its price ratio to Bitcoin The 8-year average was 0.0479. The high was 0.087… We think fair value for Bitcoin is $250,000, so if Ethereum goes back to the 8 year average, that’s $12,000 ETH. If Ethereum goes back to its 2021 high, that’s $22,000 ETH. But of course I think it’s better positioned today than it was in 2021. So that gets us to what we think is the ‘payment rails’ number — that Ethereum is going to be roughly a quarter of the value of Bitcoin. And that gets you to $62,500. And that’s kind of following the previous historical price cycles.”
Source: @ParisBlockWeek (May 2026)
Read Etherealize's "Productive Money" report on the path to $250,000 ETH below 👇
$NBIS CEO Arkady Volozh basically just confirmed that the AI bottleneck remains capacity, saying companies “cannot grow as fast as they could have grown” because there still is not enough infrastructure being built fast enough.
That is exactly why names tied directly to AI infrastructure expansion like $CRWV, $IREN, and $CIFR continue to matter. The market is realizing this cycle is no longer just about owning GPUs, but about owning the datacenter capacity, power, cooling, and inference infrastructure needed to actually deploy these models at scale.
More importantly, Nebius is positioning beyond simply renting GPUs by the hour and instead optimizing the full stack for inference with faster, cheaper, low-latency AI workloads. That inference layer may become one of the most important monetization engines of the next phase of the AI cycle.
$PLTR Q1 deal activity was 🔥🔥🔥
Here is the list of the deals you need to know before Monday's Earnings:
• Maven became an official Program of Record for the US Military;
• Maven Smart System (AI targeting) expansion in the Czech Republic;
• Up to $ 1bn with the US Department of Homeland Security;
• $ 1bn for 5y with the Office of Procurement Operations;
• Airbus expansion rumored to be ~$1bn over 10 years;
• Bain global expansion;
• 5y expansion with Stellantis;
• Nvidia expansion to deliver a sovereign AI OS reference architecture;
• HD Hyundai expansion worth “hunders of millions” over several years ;
• Polymarket and TWG partnership to police sports betting;
• GE Aerospace expansion to increase the J85 engine’s readiness;
• LG CNS partnership to drive AI across the LG Group;
• Centrus Energy partnership;
• Moder partnership to co-build an AI-powered mortgage operations platform;
• Ondas and World View partnership to advance the capabilities of the Multi-Domain Intelligence Platforms;
• OneMedNet partnership to launch the Real-World-Data Platform;
• Valar Atomics disclosed as client;
• Rackspace partnership to enhance UK Government Managed Operations;
• Sibyl, a French consultancy, partnership;
• Instinctools, a German software engineering company, partnership.
Q2 continues strong with:
• Ukraine Reconstruction;
• $300m with Dep. of Agriculture.
—————
Top notch clients.
Deeper and deeper relationships.
Any chances growth can slow?
Not anytime soon.
Yours,
@arny_trezzi
$PATH is so far ahead of investors that Wallstreet hasn't even grasped what they've built with Maestro
Investors will catch up and get hit with UiPath FOMO in due time
Investing is a lot like getting joke, then waiting months for the slow kid to also get the joke
$PATH now trades at an EV (enterprise value) of $6.43B
They're doing $1.8B+ in ARR (annual recurring revenue) growing double digits and have ~84% gross margins
40%+ net margin potential company trading at a forward EV/sales of 3.6x
No debt. GAAP profitable. Genius founder CEO.
Update to Donut Lab 8.12.25
Behind Donut Lab's all-solid-state batteries is an international technology network that combines German manufacturing technology, Swedish innovation, and Finnish materials research. The goal of the entity has been to create a battery that is completely fire-safe, lasts practically for the entire lifespan of the vehicle, and charges in a few minutes.
The source of the technology is the Swedish company Holyvolt, which developed nanopaste printing technology in its Munich laboratory. This method is based on traditional screen printing, where the internal structures of the battery are printed layer by layer with extreme precision. The technology transfer worth approximately ten million euros realized in 2024 created the basis for starting to apply this German expertise on an industrial scale in the Nordic countries.
In terms of materials, the technology relies heavily on Finnish scientific knowledge. Behind the exceptional performance of the battery is the long-term work on silicon nanomaterials by Professor Vesa-Pekka Lehto's research group at the University of Eastern Finland. Silicon is an excellent material for energy storage, but it has traditionally broken during charging cycles. The Finnish invention of stabilizing the silicon anode has enabled the battery to withstand up to a hundred thousand charging cycles without significant wear.
Industrial manufacturing is the responsibility of Nordic Nano Group, operating in Imatra. The company's factory in Imatra utilizes roll-to-roll nanoprinting, where the battery cells are printed as a continuous strip. The project has been boosted by significant public support, such as a Just Transition Fund (JTF) development grant of approximately three point six million euros. The expertise of the workforce has been ensured through local cooperation with Saimaa Vocational College Sampo, through which the first line operators have been trained by the beginning of 2026.
The final product and its integration are handled by Donut Lab, which acts as the technology department for Verge Motorcycles. Donut Lab has combined Holyvolt's printing technology and Finnish materials science into an entity that can be installed in electric vehicles. The first applications will be seen in Verge TS motorcycles, where the battery is innovatively placed as part of the rear wheel hub motor structure. The whole is thus an industrial chain formed by several specialized operators, where each has their own critical role from basic research to mass production.
... until 2026.