Sampath

#Fibonacci #Timing & #Price #TechnicalAnalysis #Trade Ideas & Fibonacci Charts are for educational purposes only. Trade at Your Own Risk.

[Sampath @spakkal]

@narrgis007 @spakkal SPX has cluster symmetry support around 7330 on the 60 minchart. watch for triggers to go long. @sssvenky @QueenOfFibs @henrygambell

[Sampath @spakkal]

@narrgis007 @spakkal SPX has cluster symmetry support around 7330 on the 60 minchart. watch for triggers to go long. @sssvenky @QueenOfFibs @henrygambell

[Sampath @spakkal]

@narrgis007 @spakkal SPX has cluster symmetry support around 7330 on the 60 minchart. watch for triggers to go long. @sssvenky @QueenOfFibs @henrygambell

[Narrgis @narrgis007]

🚀 4th July sale 🇺🇸
$101 OFF!
The markets don’t wait & neither should you.

3 month deal ! for $70 per month for 3 months
Get the edge to stay on the right side & catch the big moves in stock market.
Use code: SUMMERSALE26

🥳🔥🎇🥳
https://t.co/fSsVrmxgIg

$QQQ $SPY $MU $AMD $AAPL

[Shay Boloor @StockSavvyShay]

THE $NVDA VERA RUBIN SUPPLY CHAIN BREAKDOWN

Nvidia’s Vera Rubin is ramping into full-scale production and reshaping the AI rack bill of materials in three major ways: more HBM per rack, a shift from pluggable to co-packaged optics and the move to 800VDC power.

Here are the companies positioned to capture each transition:

Memory Storage (content-per-rack story)
• $MU, $SNDK, $SKHY and Samsung sit at the memory and storage layer as next-gen AI systems require more bandwidth, more capacity and faster data access.

Advanced Packaging (bottleneck that gates everything)
• Foundry / packaging: $TSM, $AMKR, $INTC
• Packaging equipment: $AMAT, $LRCX, $KLAC, $ASML
• Test / validation: $TER, $AEHR
• Substrates / materials: $TTMI

Optical Communication (CPO transition)
• Optical modules: $COHR, $LITE, $AAOI
• CPO / switching chips: $AVGO, $NVDA, $MRVL
• High-speed connectivity: $CRDO, $ALAB
• Fiber / networking: $GLW, $NOK, $CSCO, $ANET
• Silicon photonics / packaging: $TSM, $GFS, $TSEM

800VDC Power Supply (architecture shift)
• Wide-bandgap semis: $STM, $ON, $NVTS, $POWI, $WOLF
• Power management / hardened devices: $MPWR, $ADI, $TXN, $AOSL, $VICR, $MCHP
• Motherboard / connector / power module: $FLEX, $APH, $TEL
• Infrastructure / power delivery: $VRT, $ETN, $GEV

Compute System (integrators)
• $DELL, $HPE, $SMCI provide the system-level integration layer that turns all of these components into deployable AI infrastructure.

[Alexandru Fartade @KullAxel]

There are open source alternatives to Microsoft, Office and GPU drive.

No one is lobbying to take this open source alternatives down, while for sure Anthropic is lobbying to take open source AI down.

You can decide to be closed source, but deciding to lobby for banning your open source alternatives is not good

[Sampath @spakkal]

@spakkal #SPX has Fibonacci timing for a low on the daily chart. watch for triggers to go long. @QueenOfFibs @sssvenky @narrgis007 @johnfcarter

[Sampath @spakkal]

@spakkal $MSFT & $GOOGL has a 9 for a buy using Tom Demark Sequential Indicator plus oversold on the 120 min Chart. watch for triggers to go long. @sssvenky @narrgis007
https://t.co/6vfnBUZOvs
https://t.co/lv8G69pZe7

[Varun Kalra @_varunkalra]

Lets be honest about what “distillation” means here, because the word is doing sci-fi work. Nobody cracked open the model and copied the weights — those never leave the building. Distillation via API means one thing: you send prompts, you collect answers, you train on the input-output pairs. That’s it. Which raises the awkward question of what, exactly, was taken. Not the architecture. Not the training pipeline. Not the RLHF process, the data mixture, the infrastructure — none of the actual expensive discovery. What was captured is behavior: text the model emitted, through a paid public interface, to a paying customer. If sampling a model’s outputs and learning from them is theft of the system itself, then you’ve just described what every lab did to the entire internet — except the internet never agreed to terms of service, and API customers did. So the strongest version of the claim isn’t “they stole our model.” It’s “they violated our ToS by learning from text we sold them.” That’s a contract dispute, not the crime of the century — and it’s a narrower form of the very act the labs call fair use when they do it.

There’s also a ceiling problem the extraction narrative ignores. You cannot distill your way past the teacher. Output pairs capture surface behavior, not the capability frontier — a student model trained on a teacher’s answers inherits its mistakes, misses its reasoning process, and tops out below it. That’s why the accused labs still publish their own pre-training runs, their own RL pipelines, their own research. If distillation alone could replicate a frontier model, the “full cost of discovery” argument would be self-refuting — nobody would ever need to discover anything. The fact that frontier labs keep their lead is proof the moat is the process, not the outputs. You can’t photograph a sauce into existence.

[James Altucher @jaltucher]

MARKETS ARE HAVING A 'KIMI MOMENT'

The markets are having a "Kimi moment", similar to when DeepSeek was released in early 2025. Markets are selling off thinking that big AI companies won't need to build out as much infrastructure . But this isn't true.

Kimi K3 is the latest open source AI model out of China (interestingly, it first came out the box saying, "Hi, I'm Claude", suggesting, as many people think, that it trains on Claude data it steals). Kimi is good. Very good.

And so people are worried: will there be the same demand for data center buildouts if open source models are just grabbing the trophies? They had the same worry in January 2025 when DeepSeek came out.

The answer is: demand will be greater than ever. Any selloff due to fears of no buildout will be reversed.

A) JEVON's PARADOX. When a technological commodity becomes more efficient, demand for that commodity actually goes up.

An example is DeepSeek, the first open source AI model out of China. The same selloff happened.

Guess what? Capex went up every quarter since then.

When a resource becomes cheaper, people think of more applications for it, and demand continues to rise.

B) THE MODEL IS DIFFERENT FROM THE INFERENCE - just because someone trained a model more efficiently, doesn't mean demand for inference goes down. If demand goes up (as suggested in "A"), then the need for GPUs, CPUs, interconnects, etc go up.

C) THE SELLERS OF COMPUTE DON'T CARE. Oracle, Microsoft, Coreweave, Facebook, etc - anyone selling compute and inference, they don't care what models they are hosting - they just want to povide inference for the people. If anything, their margins will go up if they use cheaper models.

D) AGENTS - Last week was the first week that human traffic on the internet was exceeded by agent traffic (traffic being done by AI-created bots). Agents require 50x the amount of "tokens" (the units that AI is prices in) than humans do. This does not change with new models being released. If anything, the better the model, the more agents will be made with it.

E) FRONTIER MODELS TAKE ON THE COMPETITION

Anthropic proved last February that many of the Chinese open source models are stealing Claude's inference to train their own models.

Does this mean Anthropic and OpenAI give up? Of course not. They will just make better models. Even since February, Anthropic has released Fable 5 and OpenAI has released ChatGPT 5.6, which are still the best two models. These trillion dollar companies won't rollover. They will build faster than ever.

F) ITS ALREADY DONE

Land purchases, power purchase agreements, GPU sales, Memory sales, Interconnect sales, etc etc are already contracted out through 2029. Let's talk then but chances are the rise of new models will continue to create new demand for inference and the buildouts will continue. But, for sure, the next three years trillions of dollars will be spent.

G) THE LINUX EXAMPLE. Another example of Jevon's Paradox. Linus was an open source free operating system in the 90s. A cheaper operating system allowed MORE software, data centers, operating systems, etc to be created, resulting in a huge data center buildout in the 2000s.

This Kimi Moment will mirror the DeepSeek moment of early 2025. Everyone will reiterate capex. Money will be exchanged. Chips will be bought. Stocks will flourish.

[Satya Nadella @satyanadella]

https://t.co/xv6csf1SbV

[Deirdre Bosa @dee_bosa]

“The model alone is no longer the product"

@AravSrinivas says the real product is now the harness around it: orchestration, tools, enterprise context, and cost performance.

The post-frontier AI race is about systems, not just the smartest model

[Space Recorder @1spacerecorder]

India won the World Cup because of Bumrah's bowling, Sanju Samson's batting, Surya's captaincy, and Hardik Pandya's all round performance.

Apart from them, the rest are the Jhantu bits and pieces players, they will perform once in ten matches, but not consistently.

[Shay Boloor @StockSavvyShay]

$GOOGL profit is expected to nearly double by 2028.

When compute is the scarcest input in the economy then the company with the lowest internal cost per token wins out.

Google is the only hyperscaler running its frontier model end to end on its own silicon with Gemini trained entirely on TPUs.

[Matt Brezina 🌳 🌊 @brezina]

Why won’t Tesla just make a dang minivan! Sliding passenger doors, adult-sized rear seats, TRUE fold-flat rear seats. They are truly the best suburban family vehicle.

I want a FSD van. Ideally gasoline hybrid electric. Who is building this?

[Satya Nadella @satyanadella]

The future of the firm is a learning loop in which human capital and token capital compound.

With our new Frontier Co., our ambition is to help every enterprise build its own AI capability, and to help create a frontier ecosystem where every organization can turn its knowledge, workflows, and judgment into its own AI systems that continuously improve. https://t.co/mvYhkRFyqa

[Harsh Goenka @hvgoenka]

AI is brilliant at patterns. Humans are still better at judgment. Ford rehiring engineers after AI fell short in quality control is a reminder that the future isn’t Human vs AI. It’s Human + AI. https://t.co/k4bXAiEzTN

[Oh Come On! @BluthCapital]

$MU CEO, in effect:

"For > decade $AAPL has been buying our chips for $5, gluing it inside a metal box, & selling it to consumers for $99 upgrades & laughing at our attempts to get $7.
Now we're charging them $50 & they turned around & raised prices on their customers $250."

[Fiscal.ai @fiscal_ai]

Microsoft is now trading near its lowest multiple since 2017.

EV/EBIT: 18.3x

$MSFT

[Barchart @Barchart]

Microsoft $MSFT is now trading below its 200-week moving average 🚨 Excluding the Dot Com Bubble and the Global Financial Crisis, this has historically been a great time to buy 👀

[Charlie Bilello @charliebilello]

Insanity: with a market cap of $2.96 trillion, SpaceX just passed Microsoft to become the 4th largest company in the world.

Microsoft Sales: $318 billion
Microsoft Net Income: $125 billion

SpaceX Sales: $19 billion
SpaceX Net Income: -$9 billion