Rulya

never look back

[Rulya @Rulyaxd]

Four trades. +$117, -$229, -$130, -$114. Final caption: LOSS. The TikTok this comes from is titled "17 year old day trader retires parents using AI."

The chart watermark reads MNQM25, the Micro Nasdaq June 2025 contract. The top of his screen reads BAL $98,869, MLL $97,000. Eighteen hundred dollars between him and a margin call.

That spread is what makes the account a prop firm funded challenge. He is renting the $100,000, not earning interest on it. The challenge ends the moment balance touches the MLL line.

The AI tool he runs is called Scalp Trading. The dashboard returns Pattern Analysis on a one-minute timeframe with a Sell signal, entry 21000, stop 21006, risk 0.25 percent of account.

Recommended dollar risk: $25.

0.25 percent of $98,000 is $245. The tool is sizing his trade against a $10,000 account that exists only inside its risk engine. The number is two zeros short of his real balance.

He doesn't catch it and takes the Sell signal. The captions roll. +$117, then -$229, -$130, -$114. The screen reads LOSS.

The bedroom behind him still has the LED string lights and the tapestries. The Stop Loss bracket on the chart is doing the actual parental retirement.

You can check this math without leaving the page. 0.25 percent of $100,000 is $250, not $25. The dashboard is two zeros off. If you ever see this tool open on a friend's monitor, hand them a calculator first.

[Rulya @Rulyaxd]

AMD CEO Lisa Su just killed the $400/mo AI bill. The chip she's holding on stage is what your $400/mo AI bill is about to run on.

The MI455X is data center silicon. Lisa Su called it 70 percent bigger in transistor count than the prior MI355 from the same stage. 320 billion transistors per chip, 432 gigabytes of HBM4 per chip, four chips per compute tray. That stacks to 1.28 trillion transistors and 1.73 terabytes of HBM4 in a single tray, driven by AMD's next-generation EPYC Venice CPU.

The price per chip is enterprise-grade. The customer is whoever runs the subscription billing system, not you.
She is also Jensen Huang's cousin. The internet has known this for years. The new fact is that AMD's data center roadmap is now competing for the same hyperscaler contracts NVIDIA used to win uncontested.

For scale, the NVIDIA chips that ship into the same contracts are the GB200 and the upcoming B300. They sit in the same generational bracket as the MI455X. AMD is no longer racing to catch up. It is fielding silicon Anthropic and OpenAI can put on a purchase order without a separate justification memo.

The relevant rivalry on stage is not Lisa Su versus your wallet. It is Lisa Su versus Jensen Huang in the same family tree, with AMD's data center silicon now positioned to take inference contracts the NVIDIA B-series used to win without a bid.

Anthropic, OpenAI, and Meta are the buyers. The subscription you renew next month will sit on top of one of these racks, not in place of it.

The 320 billion transistors do not lower your monthly bill. They lower the cost of inference for the company that charges your monthly bill. The math of the savings goes upstream first.

If any of it ever flows down to a $20 tier, it will be because three competitors and a court order pushed it there. The cousin in the blue blazer holding the chip is not in that business.

[Rulya @Rulyaxd]

@Xbushh0 this is the move for anyone running heavy ai workflows daily. 24gb vram handles almost everything worth running locally

[Rulya @Rulyaxd]

@s4yonnara thank you for sharing this

[Rulya @Rulyaxd]

https://t.co/wXl8ybdxu8

[Gipp 🦅 @gippp69]

https://t.co/5rl5fTLiV0

[Rulya @Rulyaxd]

@test0rosso the "boring on purpose" line is the whole thing. nobody impressive wants to build Google review automation for a hair salon. that's the gap

[Rulya @Rulyaxd]

@kobaHUB the weekly roundup format is what turns this into a system. most people doing affiliate content jump from product to product and wonder why nothing sticks. one template, repeated that's the part that compounds

[Rulya @Rulyaxd]

The same mini PC people are buying to play Fortnite at 120 fps can hold a 235-billion-parameter MoE model in unified memory and run it offline.

The video on your screen is a guy with a tattooed forearm showing the GMKTEC EVO-X2 driving a game at 120. He is selling it as a budget gaming box. The captions point at the keyboard and at the affiliate link in the description.

The captions never mention that the AMD Strix Halo chip inside the same box is what runs Qwen3-235B in 4-bit at usable token rates. The 128 GB of unified memory that holds the game's textures also fits the model weights.

A few months ago a 235B-parameter model meant an OpenAI or Anthropic monthly bill. The hardware needed to run one locally cost more than a car.
The EVO-X2 lists at roughly the price of a used MacBook, draws under 120 watts at full load, and sits next to the monitor on a wall outlet.

What runs on it is not a quantized toy. It is a 235B Mixture-of-Experts model that sits in the same general-reasoning band as the top-tier paid endpoints. Offline, the full 256K token context, no rate ceiling, no usage profile being built upstream

The seller is monetizing the gaming spec. The AI side is in the silicon almost by accident of how the chip was built.

Most of the comments under the video are about Fortnite FPS. Two of them ask about LLM throughput. Those two will be the ones running the model by Sunday.

Frontier-tier intelligence stopped being a recurring charge. It became a one-time receipt with a power cable.

[Gipp 🦅 @gippp69]

https://t.co/5rl5fTLiV0

[Rulya @Rulyaxd]

@pulmencr $27k last month but won’t show the adsense screenshot

[Riley West @rileywestreel]

Andrej Karpathy (OpenAI founding member) :

"Suddenly everyone is a programmer, because everyone speaks English. It used to take 5–10 years of study to do anything in software — that's just not the case anymore."

in a 39-minute talk, Karpathy breaks down why vibe coding turns everyone into a builder — and why writing the code is no longer the hard part

he vibe-coded a working iphone app in an afternoon without knowing swift. then making it "real" — logins, payments, deployment — ate a whole week of clicking through dashboards.

Watch the talk, then read the article below.

That’s worth more than a $500 course on agent engineering.

[pulmencr @pulmencr]

A 20-year-old is making over $25,000 a month posting YouTube Shorts 3 times a day without ever showing his face

One single Short hit 114.4 million views and paid him $10,688

The method is simple. He finds viral streamer drama clips, strips the captions in seconds, adds AI voice and a new script, uploads. That's the whole thing.

Claude writes the script. AI generates the voice. He never appears on camera once.

While most people are overthinking their content strategy or waiting for one big break - this kid found a repeatable system that prints money every single day.

3 uploads. AI does the work. $27,197 last month.

The painful part is this isn't complicated. It doesn't require skills, a following, or a face. It requires 2 hours and the willingness to actually start.

Most people will read this, feel that sting for a second, and keep scrolling like nothing happened.

The ones who don't are already testing their first Short tonight.

2 hours a day changed everything for this guy

What's actually stopping you from trying this tomorrow

[Rulya @Rulyaxd]

@pulmencr the lesson cost $356😁

[Rulya @Rulyaxd]

Four trades. +$117, -$229, -$130, -$114. Final caption: LOSS. The TikTok this comes from is titled "17 year old day trader retires parents using AI."

The chart watermark reads MNQM25, the Micro Nasdaq June 2025 contract. The top of his screen reads BAL $98,869, MLL $97,000. Eighteen hundred dollars between him and a margin call.

That spread is what makes the account a prop firm funded challenge. He is renting the $100,000, not earning interest on it. The challenge ends the moment balance touches the MLL line.

The AI tool he runs is called Scalp Trading. The dashboard returns Pattern Analysis on a one-minute timeframe with a Sell signal, entry 21000, stop 21006, risk 0.25 percent of account.

Recommended dollar risk: $25.

0.25 percent of $98,000 is $245. The tool is sizing his trade against a $10,000 account that exists only inside its risk engine. The number is two zeros short of his real balance.

He doesn't catch it and takes the Sell signal. The captions roll. +$117, then -$229, -$130, -$114. The screen reads LOSS.

The bedroom behind him still has the LED string lights and the tapestries. The Stop Loss bracket on the chart is doing the actual parental retirement.

You can check this math without leaving the page. 0.25 percent of $100,000 is $250, not $25. The dashboard is two zeros off. If you ever see this tool open on a friend's monitor, hand them a calculator first.

[Rulya @Rulyaxd]

@OddsLedger seneca was also the wealthiest man in rome. the exercise is easier with assets

[Rulya @Rulyaxd]

@rileywestreel most people learn this on the wrong side of a breach

[Rulya @Rulyaxd]

He posts AI-assembled World Cup videos and says he's clearing $1,000 a day from them. The footage in the one he's editing on camera is from a tournament America did not play in.

The pipeline is a SaaS dashboard called VidEdge. He shows the sidebar on screen. Niche Finder, Video Ideas, Thumbnails, Scripts, Voiceover, Visual Style, Editor, Monetize. Eight clicks. The topic he runs through it is "Why America will win FIFA World Cup in 2026."

The thumbnail generator opens with a prompt he never had to write. It calls itself the 3-Element Rule: a bold close-up subject with intense emotion, a directional cue like an arrow or a pointing hand or a gaze, a surprising payoff element, and at most three huge bold words on screen. The output is a shocked Messi-Ronaldo split screen next to a confident USMNT player and the number 2026 in red.

The script engine drafts his hook, the voiceover engine picks a voice, the visual style picker offers Real Life Footage, which the dashboard explains as "matched footage from YouTube or archival photos."

The matched footage in the final edit is a Mexican player in a green jersey kissing the World Cup trophy.
The video argues America will win the 2026 World Cup. The footage the algorithm chose to illustrate the argument is a Mexican lifting the cup at a tournament America did not play in.

He never noticed. The thumbnail says winners. The script says America. The footage came back with Mexico.

At the end of the explainer he asks viewers to comment GUIDE for free training.

Save this. The $1,000-a-day grift is a SaaS that auto-grabs the wrong team's celebration footage and presents it as proof. The operator is running the dashboard with his eyes closed and getting paid to do it.

[zostaff @zostaff]

https://t.co/32OOPdmni9

[pulmencr @pulmencr]

A teenager from India made $52,516 in one month after shutting down his regular Shopify store

He doesn't show his face, doesn't pitch a course. He just opened the order dashboard and let it speak.

$23.50, $59.99, $95.70 - phone buzzing every few minutes

The product isn't the edge. The infrastructure behind it is.

He writes one prompt and within 20 minutes a complete product page is running. Shopify checkout stays the same, all integrations intact, orders flow through normally - but now AI is handling the entire front end.

And then it does something no human has the patience for.

It builds dozens of store variations simultaneously. 16 live versions at once. 4,000 daily visitors split across each one. The system tracks which version actually closes sales, cuts everything that doesn't, and doubles down on what works. Conversion jumped 3.2% without him touching a single setting.

He never revealed the exact stack.

Still managing your Shopify store by hand in 2026? You're already the version that got cut.

[Rulya @Rulyaxd]

@rileywestreel agreed on the model. though "the real game" tends to shift last year it was prompting, now it's infra, something else next year.