Aakash Gupta
@aakashgupta • 265,106 subscribers
✍️ https://t.co/8fvSCtBv5Q 💼 https://t.co/STzr4nqxnm 🤝 https://t.co/SqC3jTyP03 🎙️ https://t.co/fmB6Zf5UZv
Shorts
Videos
This is how they make a 535-pound bluefin tuna that sells for $3.24 million. Each fish eats 10 kg of fresh sardines per day for roughly two years to get there. Bluefin biology dictates the ratio. A bluefin runs its red swimming muscle, brain, and viscera 10 to 21°C above the surrounding water using a countercurrent heat exchanger called the rete mirabile, which traps metabolic heat that would otherwise vent through the gills. The heat is what lets the fish cruise at 7 mph and burst at 40. It is also a furnace that needs feeding. The mechanical constraint is the second half. Bluefin have undersized swim bladders, so their bodies are denser than seawater. They sink unless they swim. They are also obligate ram ventilators, meaning they cannot pump water across their gills the way most fish do. They keep moving forward with their mouth open or they suffocate. A 1-meter tuna swims roughly 43 km a day just to breathe and stay neutrally buoyant. A 200 kg market-size fish does far more. Now the feed math. Measured feed conversion ratio for ranched bluefin runs 20 to 30 to 1. To put 1 kg of meat on the fish, the farmer pours in 20 to 30 kg of sardine, herring, mackerel, or anchovy. A 200 kg tuna eats roughly its own body weight in baitfish every 20 days. Across a two-year fattening cycle, one fish runs through several thousand kilograms of wild-caught forage species. The $3.24 million number was the Oma bluefin that cleared Toyosu's January 2026 New Year auction. Ordinary wholesale runs closer to $30,000 per high-grade fish. A 28:1 conversion ratio is the entire reason bluefin exists as a luxury market. Compare to species that domesticated cleanly. Salmon FCR sits around 1.2:1. Chicken sits around 1.7. Each is a herbivore-leaning omnivore that converts plant protein into muscle at near parity. Bluefin will not eat pellets in any commercial volume. Their gut, search pattern, and muscle physiology are built for live oily fish at high speed. Ichthus Unlimited has spent two decades formulating a soy-based diet that brings the ratio to 4:1 in laboratory conditions. Commercial uptake stays negligible. The other route out is closed-cycle hatchery production. Kindai University in Japan completed the full life cycle in 2002. Twenty-four years later, fewer than 5% of farmed bluefin come from hatchery eggs. The rest are wild juveniles caught in the Mediterranean or off Mexico, towed back to pens at walking pace, and fed wild baitfish for one to three years. Evolution built one species that runs hot, swims forever, and refuses to eat anything that is not already swimming. The 28:1 ratio is what it costs to keep that animal alive in a net pen. The seagulls in the video figured out the trade before the economists did.
Aakash Gupta3,163,637 görüntüleme • 28 gün önce
Tennis players live 9.7 years longer than sedentary people. Not 9.7 months. 9.7 years. Nearly a decade. The Copenhagen City Heart Study tracked 8,577 people for 25 years and ranked every sport by how much life it adds. Badminton: 6.2 years. Soccer: 4.7. Cycling: 3.7. Swimming: 3.4. Jogging: 3.2. Tennis almost triples jogging. A separate study of 80,000 adults found racket sports cut all-cause mortality by 47% and cardiovascular death by 56%. Swimming hit 41%. Aerobics hit 36%. The question is why racket sports destroy everything else. Three mechanisms stack on top of each other. First, the physical demands. A tennis rally requires explosive sprints, lateral cuts, and sustained aerobic output. You're training fast-twitch and slow-twitch muscle fibers simultaneously. Most cardio only trains one system. Second, the cognitive load. You're reading spin, predicting angles, adjusting position, and executing motor patterns in real-time. Your brain is solving spatial puzzles at 80+ mph. That hand-eye coordination and strategic processing builds neural connections that protect against cognitive decline. Third, and this is the one researchers keep coming back to: you literally cannot play alone. Every racket sport requires another person on the other side of the net. That forced social interaction triggers neurochemical benefits that solitary exercise cannot replicate. Strong social connection alone increases your chance of longevity by 50%. Jogging is you and your thoughts. Tennis is you, a strategic opponent, and a community. Dr. Daniel Amen is right. The data is overwhelming. If you want the single highest-ROI activity for a longer life, pick up a racket.
Aakash Gupta5,245,831 görüntüleme • 1 ay önce
Your 8pm workout is shrinking your brain. It's also building belly fat and more than doubling your odds of dying in the next decade. Andrew Huberman is naming the mechanism. Cortisol is supposed to peak when you wake up and crash by 11pm. Hard training after 7pm spikes it back up at the worst possible window. Four nights a week of evening workouts and the daily curve flattens. The morning peak softens. The evening trough inflates. The shape inverts. A 2017 meta-analysis of 80 studies and 23,000 participants found that flatter cortisol curves predict higher mortality, more inflammation, worse cognition, and greater visceral fat. The hazard ratio for early death was 2.40. Total daily cortisol output didn't predict any of it. Curve shape carried the variance. Whitehall II followed 4,000 British civil servants for six years and got the same answer. Flatter slope, hazard ratio 1.30 for all-cause death, 1.87 for cardiovascular death. Independent of every covariate they measured. Robert Sapolsky's primate work shows what happens next. Subordinate baboons getting harassed daily show lower morning peaks and higher evening troughs than dominant baboons. The flattening comes first. Hippocampal atrophy follows. Cushing's patients show the same loss. PTSD veterans show the same. Three different pathways into chronic glucocorticoid exposure, one consistent brain effect, partially reversible when the curve gets restored. There's another loop running underneath. Visceral belly fat manufactures cortisol locally. The enzyme is 11β-HSD1. It converts inactive cortisone to active cortisol inside the fat cell. Mice with adipose-specific overexpression of this enzyme develop visceral obesity, insulin resistance, and metabolic syndrome on completely normal blood cortisol. The systemic level looks fine. The local tissue is bathed in glucocorticoid signaling. That's why "meditate to lower cortisol" doesn't shrink the belly. The belly is making its own. What rebuilds the curve is cheap. Train hard before late afternoon. Keep the four hours before bed for walks and stretching. Get morning sunlight in the first hour after waking. Anchor sleep timing within a 30-minute window. Each one steepens the morning peak and deepens the evening trough. Total daily cortisol barely moves. The curve gets sharper. Make the morning loud. Make the night quiet.
Aakash Gupta2,143,843 görüntüleme • 1 ay önce
There's a reason Walmart's Vizio TVs are so cheap. Walmart paid $2.3 billion to put advertisers in your living room. Every new Vizio will not turn on without a Walmart account. Walmart paid $2.3 billion in December 2024 for the right to make that the rule. The TV was never the product. Vizio's hardware business runs at a wash. The Platform+ segment, which is essentially advertising and ACR data, accounted for all of Vizio's gross profit in the years before the deal. The TVs themselves were the customer acquisition cost. Walmart bought 19 million active SmartCast accounts. That base grew about 400% from 2018 to 2024. Every account is a household with a screen pointed at the couch every night. The mechanism is automatic content recognition. The TV reads frames from whatever you're watching and matches them against a database. Cable box, game console, DVD, the Walmart app on your phone over the same Wi-Fi. The source doesn't matter. The screen is the sensor. In March 2026, Walmart announced the account requirement at IAB NewFronts and turned on the flywheel. 19 million viewing households on one side. About 150 million weekly Walmart shoppers on the other. Run them against each other and you get closed-loop attribution: saw the ad, drove to the store, bought the product. CMOs have been trying to build that for 30 years. The comparison is Amazon. Amazon ran $68 billion of advertising in 2025, roughly 8% of its $830 billion in GMV. Walmart did $6.4 billion against $713 billion in sales. That's 1%. Same playbook, $50 billion of headroom. The margin gap is the punchline. Selling a TV nets low single digits. Selling an ad against that TV's viewing data clears 70%. $2.3 billion divided by 19 million households is $121 per living room. The TV is the cheapest thing Walmart has ever sold you.
Aakash Gupta1,383,744 görüntüleme • 27 gün önce
Hailey Bieber sold Rhode to e.l.f. for $1 billion in three years. Charlotte Tilbury took 7 years to clear $1 billion. Byredo took 16. The Rhode deal closed August 5, 2025. Rhode shipped 10 products at the time of sale. The 8-K disclosure was unusually granular for beauty M&A. $212 million in net sales for the 12 months ending March 31, 2025. Cost of goods around 19% of revenue. Zero physical retail. Direct-to-consumer only. That COGS line is the part Wall Street kept rereading. Estée Lauder runs closer to 30%. Charlotte Tilbury's pre-acquisition EBITDA margin sat at 3.8%. Rhode was printing margin because the SKU count was small enough to manufacture at scale on each item. The marketing line is where the real story sits. Tribe Dynamics tracked Rhode at $248 million in Earned Media Value in 2024 alone. Ranked #1 in skincare globally. 367% year-over-year EMV growth. Rhode ran a quarter-billion-dollar marketing campaign and paid for none of it. Reported marketing efficiency at the time of the deal was around 9x. Every dollar in brought back nine. Beauty incumbents run that ratio in the 3-4x range. The product strategy was the discipline most celebrity brands miss. Rhode launched June 2022 with three SKUs: peptide lip treatment, peptide glazing fluid, barrier restore cream. $20 to $40 pricing. Roughly one new product per quarter. The full catalog still fit on a single page at acquisition. KKW Beauty carried about 50 SKUs at peak. Kylie Cosmetics ran past 100. Both had 10x the operational complexity and a fraction of the multiple. The $35 phone case with a slot for the peptide lip treatment was the giveaway. Wearable advertising. A daily product reminder visible to every person Hailey's customer met that day. e.l.f. paid $800M at close plus a $200M earnout tied to growth e.l.f. now controls. Sephora distribution rolled out fall 2025. International through e.l.f.'s existing channels. TIME just named Rhode one of the 100 most influential companies of 2026. The 8-K already named it $21 million in net sales per SKU. Hailey was the only marketing department Rhode ever needed. e.l.f. paid $1 billion for the headcount.
Aakash Gupta1,564,302 görüntüleme • 1 ay önce
Disney spent $1 billion in 2019 building a Star Wars theme park where you were not allowed to meet Luke, Leia, Han, or Darth Vader. Galaxy's Edge at Disneyland was set on Batuu, a backwater planet in a narrow window between The Last Jedi and Rise of Skywalker. By that timeline Luke was dead. Han was dead. Vader had died 30 years earlier. Leia was alive but had no canonical reason to show up at an outer rim smuggler outpost. Imagineer Scott Trowbridge spelled out the design rule in 2022. Characters on Batuu would stay locked to their specific era. No visitors from other Star Wars timelines. The immersion was the entire point. In practice guests flew to Anaheim for Star Wars and walked through a $1 billion set to meet Vi Moradi and Dok-Ondar. The locals of Black Spire Outpost. The parallel failure was Galactic Starcruiser. $5,000 for two nights in the same sequel-era window. No Luke, no Vader, no Han, no Leia. Disney wrote down $250 million to close it 18 months after opening. On April 29, Galaxy's Edge at Disneyland abandons the rule. Darth Vader will roam Batuu hunting Luke. Leia and Han will appear at the Millennium Falcon. Kylo Ren is being pulled from the land and relocated to Tomorrowland. The ambient Batuu music gets replaced with the John Williams score. Disney spent seven years defending the design principle. Then Galactic Starcruiser closed with a $250 million write-down. Luke Skywalker showed up for one limited event last year and got swarmed by guests. The rule quietly got dropped. Avengers Campus figured this out on day one. You put Captain America in the Avengers land.
Aakash Gupta2,152,825 görüntüleme • 1 ay önce
Disney paid roughly $200 million just for Robert Downey Jr. and the Russo Brothers to come back. The first full Avengers: Doomsday trailer was shown at CinemaCon this afternoon. In one shot, Doctor Doom catches Stormbreaker with one hand. Stormbreaker was forged in a dying neutron star. The MCU wrote it as the only weapon capable of killing Thanos. Thor had to hold a star's mold open with his bare hands to make it. Doom holds it up. No Infinity Stones. Two fingers. That single frame is the entire MCU reset. Thanos needed six stones. Doom needs none. Every film in the next phase gets calibrated against a villain the trailer has already established as stronger than the one that took a decade of setup to make credible. The financial math behind that frame is wild. Kevin Feige's last three team-up bets collapsed. The Marvels barely cleared $200M globally against a $270M negative cost. Quantumania did $476M on a $200M budget, a loss after marketing. The entire Kang arc got scrapped after Jonathan Majors' conviction. Bob Iger returned as Disney CEO in November 2022 and cut the Marvel TV slate. Feige rebuilt development around three anchors. RDJ as Doom at a reported $100M+ solo. The Russos at $80M for both films, with escalators at $750M and $1B. And now Chris Evans returning for the first time since Endgame in 2019. Seven years off the board. Back in the room today. Disney showed the trailer only at CinemaCon. Exhibitors-only. Played twice in the room. Not posted online. The 13 seconds below is the bootleg someone recorded from the theater floor, and it's the only way anyone outside that room will see Doctor Doom's face before December. That's what $200M of star power pays for: a single frame fans will break an NDA to share, for a movie Marvel won't officially show them for months.
Aakash Gupta2,008,539 görüntüleme • 1 ay önce
Amazon's internal metrics said customers waited under 60 seconds for customer service. Jeff Bezos picked up the phone in a meeting and waited more than 10 minutes. The head of customer service had been defending the number. Bezos said "Ok, let's call." He dialed Amazon's 1-800 line on speaker. The room sat there for over ten minutes before a rep answered. The metric didn't survive the meeting. Bezos has a saying: when the data and the anecdotes disagree, the anecdotes are usually right. Metrics don't measure reality. They measure what you designed them to measure. Customer service dashboards commonly filter out abandoned calls, cap hold time at the IVR timeout, and start the clock after the menu tree completes. Every one of those choices pushes the average down. The customers hanging up at minute 9 are not in the denominator. The 60-second number was technically accurate and practically wrong. That call broke through a defended metric in a way no spreadsheet could have. The head of CS had dashboards and a team whose job was to report that number going down quarter over quarter. Bezos had 10 minutes of hold music and a room full of people watching. This is the executive test almost nobody runs. Call your own 1-800 line. Try to buy your own product in incognito. Every senior leader can do it in under 15 minutes. Almost none do, because the dashboards feel like the truth and the dashboards say things are fine. The measurement got redesigned. Wait times actually fell. When your data says you're winning and your customers say you're losing, the customers are right. The data was built by people whose job depends on it going down.
Aakash Gupta1,741,163 görüntüleme • 1 ay önce
Elon told Katie Miller it would be impossible to restart with $1,000. He's worth $800 billion, more than the next four wealthiest people on Earth combined. He's exactly right. The reason is the most underrated rule in capitalism. Reputation and knowledge are the asset. The cash is downstream. Sam Altman got fired from OpenAI on November 17, 2023. Within 48 hours, Microsoft offered him a new AI division with effectively unlimited resources. Same person, same skills, zero equity at the moment of firing. The world's largest software company opened its checkbook before the weekend ended. Mira Murati left OpenAI in late 2024. She raised $2 billion at a $12 billion valuation in July 2025 with no product. Andreessen Horowitz led the round. Four months later the company was in talks at $50 billion. Tinker shipped in October. Adam Neumann blew up WeWork at a $47 billion valuation. Two years after he was forced out, a16z wrote a $350 million check at a $1 billion pre-launch valuation for Flow. Largest single investment in the firm's history. Travis Kalanick got pushed out of Uber in 2017 mid-scandal cycle. Saudi PIF wrote a $400 million check into CloudKitchens at a $5 billion valuation two years later. Microsoft led an $850 million round at $15 billion in 2021. The market prices founders, not founder bank accounts. The bank account is the lagging indicator. The inversion settles it. Sam Bankman-Fried had $26 billion before FTX collapsed. After, he could not raise a dollar. Elizabeth Holmes had the magazine covers. After the verdict, nothing. The market funds belief in future compounding. Strip Elon of cash and the belief is intact. Strip SBF of belief and the cash did not matter. Give Elon $1 and you get back more than $1. That is the entire claim. The $1,000 thought experiment assumes the founder is the same person stripped of capital. The market does not think that way. The market thinks the founder IS the capital. The $1,000 is a rounding error on whatever round they want to raise next. Every founder gets asked this. Elon gave the honest answer.
Aakash Gupta925,083 görüntüleme • 28 gün önce
Joe Rogan just explained how one Lone Star tick bite ends your ability to eat red meat for life. Eat a burger for dinner. At 2am you wake up in anaphylaxis with no idea why. 450,000 Americans have alpha-gal syndrome. The tick injects a sugar called alpha-gal into your bloodstream. Your immune system tags it as a threat and starts making IgE antibodies. The same alpha-gal sugar sits in beef, pork, lamb, venison, and rabbit. Every meal from a mammal becomes a delayed reaction three to five hours later. That's why he's down to eggs. Fish is safe. Chicken is safe. Turkey is safe. Alpha-gal is found in nonprimate mammals only. Old World primates lost the gene that makes it about 28 million years ago. Birds and fish never made it in the first place. So the only safe protein comes from species across an evolutionary line drawn before humans existed. For some patients dairy goes too. Gelatin capsules and gummies go. The cancer drug cetuximab, manufactured in a mouse cell line, has triggered fatal first-dose anaphylaxis in alpha-gal patients. Pig heart valves have triggered reactions on the operating table. Between 2010 and 2022 the CDC logged over 110,000 lab-confirmed cases. The actual number sits closer to 450,000 because alpha-gal syndrome isn't nationally reportable and most physicians outside the Southeast don't recognize it. A 2025 analysis of 114 million patient records showed mammalian meat allergy diagnoses rose 5,500% from 2015-2020 to 2021-2025. Climate change pushed Lone Star ticks north of their Southeast range. White-tailed deer populations recovering from near-extinction and second-growth forests fueled it. Suffolk County, Long Island is now a hotspot. Martha's Vineyard hit establishment by 2014. Forestry workers in heavy-tick regions test 20-30% positive for alpha-gal IgE. In summer 2024 a 47-year-old man in New Jersey died after what his wife later remembered as 12 itchy ankle bites. He had assumed they were chiggers. The autopsy went unsolved for months until allergist Thomas Platts-Mills identified it as the first confirmed AGS fatality on record. There are almost certainly more. The trigger food gets eaten three hours before the cardiac event, so nobody connects the steak at dinner to the call to 911 at midnight. Eggs are safe because birds never made the sugar.
Aakash Gupta878,379 görüntüleme • 26 gün önce
At the end of Interstellar, Murph is nearly 90 and dying. Cooper is still physically in his 40s. The distance between them isn't years. It's a black hole spinning at 99.99% of maximum angular momentum. Kip Thorne, the Nobel laureate who consulted on the film, spent hours proving this was mathematically possible. A time dilation factor of 60,000x on a stable orbit. Nolan made that number non-negotiable. Thorne thought it was impossible, ran the Kerr metric equations, and found it was marginally achievable. Then Nolan broke two of his own filmmaking rules to shoot it. He filmed McConaughey's reaction in close-up first. Directors never start there. And McConaughey hadn't seen the video messages from his on-screen kids. Those tears were real. First take. Ellen Burstyn was 82 playing elderly Murph. McConaughey was 45. No de-aging tech. No prosthetics. The age gap between father and dying daughter was real because the physics demanded it. Jonathan Nolan's original script had no reunion. The ending was darker. Cooper never made it back. Christopher read his brother's draft, added this scene because he was a parent, and called the father-daughter relationship "the north star of the film." The most devastating goodbye in modern cinema exists because a physicist found a loophole in Einstein's equations and a director became a dad.
Aakash Gupta1,548,833 görüntüleme • 1 ay önce
Warren Buffett, in his first sit-down since stepping down as Berkshire CEO, gave the cleanest indictment of legalized gambling in a decade. He called it a tax cut for the wealthy. The math proves him exactly right. Americans wagered $165 billion at legal sportsbooks in 2025. They lost $16 billion of that. FanDuel pulled $6 billion of the losses. DraftKings pulled $5.3 billion. Every state with legal mobile sports betting collected a tax on the bettor side. New York alone took in over $1.2 billion in 2025 sports betting tax revenue. Layer the lottery on top. State lotteries generate over $90 billion a year. The bottom half of income earners account for roughly 70% of total spend. The average lottery player makes $38,000. A household earning $20,000 spends three times more on tickets than one earning $30,000. The implicit tax rate, meaning whatever the state keeps after prizes, runs 30 to 50% depending on the game. No other revenue source in America has that base and that rate. The structural design is the engine. A single straight sports bet carries a hold of 4 to 5%. A four-leg parlay carries a hold above 30%. FanDuel and DraftKings spent five years rebuilding their apps to make parlays the default product. FanDuel's blended hold rate hit 11.4% in 2025, up from roughly 7% in 2022. The product got worse for the customer and the customer wagered more anyway. Now look at the substitution. Nine US states have no state income tax. Seven of those nine run state lotteries. Seven of those nine have legalized sports betting. The states most committed to never taxing wealth are the same states running the largest extraction machines on people who cannot afford to lose. Read it as policy. Here is what Buffett is actually pointing at. The state needs revenue. It can raise income tax on the top decile, or it can run a lottery plus a sports betting tax. The second option raises the money from the people who can least afford it. The first option becomes politically optional. New York's $1.2 billion in 2025 sports betting tax is $1.2 billion the state did not have to ask of someone earning $5 million. DraftKings and FanDuel sell a privatized collection mechanism for a regressive tax that the state never has to defend at the ballot box again. Voters approve legalization once. Collection runs forever. The state takes a cut. The wealthy get a quieter top bracket. The bettor's cut shrinks every quarter as the parlay menu gets pushed harder. The function of a government, Buffett said, is not to play its people for suckers. Thirty-nine state governments now do.
Aakash Gupta933,851 görüntüleme • 1 ay önce
This is only the second chimpanzee civil war ever documented. The first was Gombe, 1974. Fifty years apart. Genetic data suggests chimp communities split roughly once every 500 years. Gombe was 9 males breaking off from a group of about 60. The splinter faction was hunted down over four years. Every male killed. Goodall had nightmares about it for decades. Ngogo is operating at a completely different scale. 200 chimps. The largest studied community on Earth. They functioned as one unit for 20 years, shared territory, ran border patrols together, fought neighboring groups side by side. Then in 2015 something snapped. Five key males died, possibly from disease. Those males were the social bridges between two clusters. Once the connective tissue disappeared, the cliques hardened. By 2017 they occupied separate territories and patrolled borders against each other. By 2018 the killing started. 28 dead so far, 19 of them infants. The Western faction, despite being smaller, initiated every attack. They ripped infants from mothers. The lead researcher calls himself a war correspondent. The part that rewrites the textbook: no ethnic divisions, no religion, no ideology, no resource scarcity. Just a social network that lost its bridge nodes. The researchers think the same mechanic explains human civil wars better than cultural theories do. Polarization isn't about beliefs. It's about who stopped talking to whom.
Aakash Gupta1,312,304 görüntüleme • 1 ay önce
This guy literally broke down how to use Claude Code like an expert: 1:40 - Code vs Cowork vs OpenClaw 6:51 - Setting up context status line 12:03 - Sub-agents 17:49 - Creating skills 23:58 - Ask user questions tool 33:33 - Tool-powered skills: Tavily 36:57 - CLI vs MCP vs API hierarchy 39:30 - Make slides skill w/ Puppeteer 43:32 - Auto-invoking skills with hooks 46:49 - Jupyter notebooks for data trust 55:09 - The operating system file structure
Aakash Gupta1,150,817 görüntüleme • 2 ay önce
Lionsgate spent $100 million on Ballad of Songbirds & Snakes in 2023 and got $349 million back. Lowest opening weekend in franchise history. $44 million domestic debut for a series that used to open north of $150 million. Their response was to double the cast budget on Sunrise on the Reaping. Ralph Fiennes. Glenn Close. Kieran Culkin. Jesse Plemons. Elle Fanning. Maya Hawke. Jennifer Lawrence and Josh Hutcherson returning for the first time since 2015. That's four Oscar nominees and two Emmy winners in a single YA adaptation. The cast page reads like an awards ceremony guest list stapled to a Succession reunion. And the math behind it is actually rational. The Hunger Games franchise has made $3.3 billion across five films. Catching Fire, the one installment that involved a Quarter Quell (the 75th), is the highest-grossing Lionsgate film ever at $865 million. Ballad of Songbirds & Snakes proved the IP still has a floor: $349 million worldwide on a cold-start prequel with zero legacy cast. The floor held. So Lionsgate is running a calculation. The Second Quarter Quell is the most requested story in the fandom because Haymitch's games were only described in two paragraphs of the original novel. Suzanne Collins turned two paragraphs into a 400-page book. Lionsgate turned the book into a cast sheet that makes you do a double take. Kieran Culkin as Caesar Flickerman is the kind of casting that sells tickets on name recognition alone, before a single frame of footage exists. The trailer drops tomorrow. Mockingjay statues with flowers showed up in Millennium Park, Chicago. Snake statues covered in ivy appeared in Times Square. The marketing campaign is treating this like an event film, which tells you exactly how Lionsgate sees the stakes. Their entire theatrical slate depends on this franchise producing $600M+ worldwide. Anything less and the prequel strategy dies. 48 tributes. The most stacked cast in franchise history. And a studio betting everything that the one story fans have been asking about for 14 years can bring the audience back.
Aakash Gupta821,847 görüntüleme • 1 ay önce
Rafael Nadal was diagnosed with Müller-Weiss syndrome at 19 years old. The navicular bone in his left foot was collapsing. There is no cure. The condition is degenerative. It only gets worse. The navicular is the keystone of the human foot. It catches the head of the talus and connects to the first three toes. It absorbs the majority of load when you change direction. In tennis, players change direction hundreds of times per match. On clay, where the surface forces you to slide into every shot, the stress on that bone multiplies. His sport demands exactly the one thing his body could no longer do without pain. He won 22 Grand Slams after the diagnosis. Fourteen of them at Roland Garros, the clay court tournament that punished his foot the hardest. His record there: 112 wins, 4 losses. A 97% win rate across 23 years at the single venue that required the most from the bone that was failing him. For context, the other Grand Slam dominance records: Djokovic at the Australian Open has a 91% win rate. Federer at Wimbledon had 88%. Nadal's 97% at Roland Garros isn't just the best in tennis. There may not be a comparable number in any individual sport at any single venue, ever. He once told reporters he doesn't remember what the feeling of playing without pain is. The condition is most common in women aged 40 to 60. He got it at 19 and kept winning for 19 more years. The Rafa documentary drops May 29. During the French Open. The tournament he won 14 times will be happening without him while 300 million subscribers watch what it actually cost him to own it. He turns 40 on June 3. Netflix timed this so the stadium that was his is full of players trying to fill a void that 97% says might be permanent.
Aakash Gupta767,254 görüntüleme • 1 ay önce
This guy literally broke down everything you need to master Claude: 6:07 - Why to Stop Using Chat 9:56 - Cowork vs Code vs Dispatch 18:44 - Skills and MCP Connectors 25:03 - The Skills Marketplace 29:06 - Strategy Canvas Demo 35:14 - Skill Iteration Cycle 40:46 - Why You Need Code 44:43 - Building a Second Brain 56:00 - Self-Improving Knowledge 1:10:00 - Dispatch and Remote Work 1:21:07 - Top Mistakes and Future
Aakash Gupta257,090 görüntüleme • 20 gün önce
She literally broke down how to run evals in Claude Code (built the whole thing live): 01:34 - What people get wrong with evals 04:35 - Why product taste is the alpha now 09:28 - Building a PM agent from one prompt 19:00 - Instrumentation without writing code 22:00 - Watching traces stream in live 28:00 - Getting Claude to write your first eval 33:58 - When vibe evals work and when they don't 48:50 - The self-improving loop (this part is wild) 01:03:00 - Same-day shipping is real 01:06:00 - The context graph unlock
Aakash Gupta150,362 görüntüleme • 12 gün önce
Ryan Lopopolo leads a team at OpenAI where the PM writes a PRD on Monday and ships a pull request by Friday. No human writes the code. Ryan Lopopolo broke it all down: 0:00 - "Code is a liability" 3:23 - Why your most expensive asset is now free 6:01 - "What's the point of roles anymore?" 8:04 - What replaces the PM/design/eng triangle 13:10 - 1M lines of code, zero written by humans 16:05 - Engineers can't touch the keyboard 18:13 - First month was 10x slower than solo 20:07 - Recursing 8 levels deep for one primitive 20:47 - PM writes PRD Monday, ships PR Friday 25:06 - The feature they had to trash 28:02 - How designers ship UI without a backend 31:40 - What's actually inside the harness 37:03 - Failing the build over curly quotes 40:02 - Inside Ryan's actual Codex setup 46:25 - The codebase that grades itself 50:49 - "A billion tokens a day or you're negligent" 52:19 - 350M tokens on a single PR 53:46 - GPT 5.2 changed everything overnight 57:00 - Every engineer is now a staff engineer 59:19 - The ego problem nobody talks about 1:00:39 - Monday morning roadmap for normal teams 1:08:19 - One skill to build this weekend 1:10:57 - Why one agent beats multi-agent
Aakash Gupta102,298 görüntüleme • 9 gün önce
Karpathy told Dwarkesh that a 1 billion parameter model, trained on clean data, could hit the intelligence of today's 1.8 trillion parameter frontier. That is a 1,800x compression claim. The math behind it is more defensible than it sounds. When researchers at frontier labs look at random samples from their training corpus, they see stock ticker symbols, broken HTML, forum spam, autogenerated gibberish. Not Wikipedia. Not the Wall Street Journal. The actual pretraining dataset is mostly noise, and the model is burning parameters to vaguely remember all of it. One estimate pegs Llama 3's information compression at 0.07 bits per token. Well-structured English carries around 1.5 bits per token of real information. The trillion-parameter model is holding a roughly 5% resolution image of the internet it trained on. So when a lab ships a 1.8 trillion parameter model, the overwhelming majority of those weights are handling rough memorization. They are compression overhead for a noisy training set, taking up capacity that could be doing reasoning instead. Karpathy's proposal is to separate the two. Build a cognitive core: a small model that contains only the algorithms for reasoning and problem-solving, stripped of encyclopedic memorization. Pair it with external memory the model queries when it needs a fact. A 1 billion parameter reasoner plus retrieval beats a 1.8 trillion parameter model trying to do both. The data already supports this direction. GPT-4o runs at roughly 200 billion parameters and outperforms the original 1.8 trillion GPT-4. Inference costs for GPT-3.5 level performance fell 280x between 2022 and 2024, driven almost entirely by smaller, cleaner, better-architected models. The trend line is pointing where Karpathy says it should. The real implication for anyone tracking the AI trade: data quality is the actual constraint. The companies winning the next phase will be the ones who figured out what to train on, and what to throw away.
Aakash Gupta507,340 görüntüleme • 1 ay önce