Cherenedene

My top issue with X: The Grok button disappeared On Jan 28, the Grok button appeared on all posts on X The next day, it vanished from posts in "Notifications" This issue hasnt been fixed in over six months, and drastically cuts my Grok usage Can the Grok button be restored in notifications? @elonmusk @abustin @phroo @ibab @512x512 @TobyPhln @TheGregYang @chenchen @chrisparkX @nikitabier

@NuttyCLD So it just takes 10 days longer? Shouldn't be a problem then right?

The Middle East crisis is hitting memory and logic asymmetrically. Korea sources 97.5% of bromine from Israel and 64.7% of helium from Qatar, putting Samsung and SK Hynix on a faster materials depletion timeline than $TSM TSMC. This is structurally different from 2021's chip shortage or 2022's neon crisis. In 2021, demand overwhelmed intact supply. In 2022, it was a single gas disruption, and prior diversification efforts allowed resolution within six months. Now helium, bromine, gallium, and LNG are all cut simultaneously, and the Strait of Hormuz closure adds 10+ days to any alternative routing. TSMC has confirmed manageable inventory levels. Taiwan does face energy exposure with 33.5% of its LNG sourced from Qatar, but on materials concentration, Korea is far more acute. The timing compounds the risk. Samsung is ramping 2nm while trying to close the foundry yield gap with TSMC. A helium shortage lasting beyond three months could structurally lock in TSMC's lead in advanced nodes. This stops being a cost problem and becomes a competitive positioning problem. If Korean memory production takes a real hit, $MU Micron is increasingly cited as a potential beneficiary given its more diversified sourcing. 🔗 cnbc.com/2026/03/19/the… 🔗 scmp.com/tech/article/3… 🔗 trendforce.com/news/2026/03/0…

Why did the odds of a SpaceX IPO by June 30 drop by 20% today?

@elonmusk @DannyLimanseta The problem is that you are assuming the intention of the creator and that you could determine what is interesting or not It could be purpose built or have an inscrutable purpose You presume too much

@DannyLimanseta For the same reason we run simulations: to observe interesting outcomes. That’s why the most interesting outcome, especially if ironically entertaining, is the most likely: all the boring simulations were terminated.

I had dinner once with a top physicist and a top computer scientist and asked what they thought the probability was that we were in a simulation. They answered simultaneously at 0% and 100% respectively. It was like a double-slit experiment, but with humans.

@techdevnotes I don't necessarily agree Citations are possibly more important than the text itself. They have to be easy to access Could they be improved? Sure. But I'm not sure what you mean by busy or need attention. You should not hide them tho

Citation chips in Grok Web responses looks too noisy and takes too much visual attention ...

@a16z He has worn a raincoat indoors, for the last 2 interviews Why?

"I think our biggest risk as a country is suicide, not homicide." Palantir CTO Shyam Sankar joins a16z's Katherine Boyle and Erik Torenberg to discuss Shyam's new book, Mobilize, as well as defense, AI, the SaaSpocalypse, and more. 00:00 Introduction 07:53 Rebuilding the industrial base 18:01 Modernizing the Army 24:20 The SaaSpocalypse 29:42 Agency over automation 38:24 Beating China without self-sabotage 40:42 Film as cultural willpower 49:57 The story of Admiral Rickover @ssankar @KTmBoyle @eriktorenberg @PalantirTech

@Polymarket can you stop writing numbers like that and just put B? Its hard to read

JUST IN: U.S. and Japan announce $40,000,000,000.00 joint project to build nuclear reactors.

@RnaudBertrand Only a small section of pars was destroyed So that shouldn’t change much

I hate to be the bearer of bad news but if infrastructure like this 👇 gets blown up, as of this moment it will take at least a decade to recover from this war - and the truth is that the world's energy picture is probably changed forever. This single facility 👇produced roughly 20% of global LNG supply (aljazeera.com/news/2026/3/18…) and, as of 2011, had taken $70 billion to build (energyintel.com/0000017b-a7be-…). What makes this even worse is that Iran's strike on this was retaliation after Israel attacked their South Pars gas field which draws from the same natural gas reservoir, which is the world's largest by far (9,700 km² - about the size of Qatar itself). Heck, on the list of the 25 largest natural gas fields (en.wikipedia.org/wiki/List_of_n…) this single reservoir holds roughly 40% of their combined recoverable reserves - and is nearly 6 times bigger than the 2nd biggest field in the world. And, unlike many of the others on the list, it's only at 10% depletion (meaning 90% of the gas is still there). Which means that, probably for many years, a huge share of the gas from the world's largest reservoir simply won't be extractable, as infrastructure on both sides - Qatar's and Iran's - has now been blown up. From a global energy supply perspective, we're deep into worst-case scenario territory.

@Polymarket Why are all the bets either about 70% or below 34%? You should have more bets/coverage so that there are more bets around the 50% level Also why arent there longer time periods? ie. till Dec 2026. There should be bets for every month right?

JUST IN: Oil analysts are saying they “wouldn’t be surprised” if oil went to $200-$250 a barrel.

@BarakRavid What percentage of the pars gas field was disrupted as a result of the strike ?

🚨An Israeli official said the strike on the gas facility was coordinated between the Israeli prime minister's office and the White House 🚨The official said it was aimed at signaling to Iran that if it continues disrupting oil supply through the strait of Hormuz there could be an escalation in the targeting of its energy facilities and a worsening of the economic crisis in the country 🚨"It was a signal of what could come next", the official said

@jukan05 @deesontralize Do you mean its harder because of the language, or for some other reason?

@deesontralize It’s hard for me to cover it in that much detail. Since I’m Korean, it’s more difficult for me to understand U.S. companies at that level.

@TJTheWheelDeal Do you want to see the result of the lawsuit before investing? They could lose $135B in equity and you could be diluted

If you could invest in a round of Open AI, would you? Our contact at JP MORGAN brought the deal to us today. 880B valuation for 25B in revenue is steep. Yeah yeah yeah, but they will own the space. Will they? Then there is Sam. He’s just got an off vibe. What say you?

@a16z @vishalmisra Kind of a dumb waste of compute. Just train it on all modern physics and see if it can discover new ones Same test but actually potentially useful

The Einstein Test for AGI. @vishalmisra: “You take an LLM and train it on pre-1916 physics and see if it can come up with the theory of relativity.” “If it does, then we have AGI.” “It’s a high bar, but we should have high bars.” “At the time of Einstein, there were a lot of clues that Newtonian mechanics, there was something missing.” “But until Einstein came up with a new representation of the spacetime continuum, we were stuck.” “If you had a model that just looked at correlations and such... it would not have come up with the beautiful equation that Einstein came up with.” Vishal Misra with @martin_casado

@elonmusk Firstly, its not enriching yourself - its just getting back the equity you already deserved.. same as equity in your other co's Secondly, if winning the AI race is important for the survival of humanity, you should use the proceeds for that 90% of charity is wasted

Btw, the proceeds of any legal victory in the OpenAI case will be donated to charity. I will in no way enrich myself.

@AnishA_Moonka So polymarket in theory could be making $5B per year (all else being equal) Is it known why the fees/strategy differs so much between polymarket and kalshi? Presumably there is some tradeoff not captured in the pure revenue figures

polymarket’s $4.2M in 30 days annualizes to about $50M. kalshi’s $110M annualizes to $1.32B. so kalshi is generating roughly 26x more revenue. but polymarket charges 0.01% per trade vs kalshi’s ~1.2% average. if polymarket even moved to 0.1% fees (still 12x cheaper than kalshi) its revenue would jump to around $500M annualized at current volume.

Kalshi made $110M of the $122M total. One company accounts for 90% of the industry's revenue in the prediction market. Kalshi brought in $1.8M in 2023, $24M in 2024, and $260M in 2025. At its current monthly pace, the company is running at roughly $1.3B a year, according to Wall Street Journal. Almost all of it is sports. About 89% of Kalshi's 2025 fees came from sports contracts, and that share climbed above 90% toward the end of the year. The NFL season alone generated $138M in sports fees between September and November 2025. The product works through "event contracts," which are yes/no bets on outcomes like "Will the Eagles win by 3+ points?" Kalshi calls it trading. Sportsbooks call it competition. At $1.3B a year in sports revenue, Kalshi is now roughly a fifth the size of DraftKings, which expects $6.5-6.9B in 2026. Both DraftKings and FanDuel's parent company Flutter have lost more than half their market value over the past year. The #2 on this chart, Polymarket, charged zero fees for all of 2025. It brought in $0 in revenue that entire year. It only started collecting fees in January 2026. That $4.2M comes from a fee structure that's been live for about two months, charging 0.01% per trade. Kalshi charges around 1.2% on average. Roughly 100x higher. Kalshi's valuation doubled in seven weeks last fall: $300M raised at $5B in October, then $1B at $11B in December. Both Kalshi and Polymarket are now in talks to raise at $20B. Monthly active users on Kalshi went from 600,000 at the start of 2025 to 5.1 million by February 2026, per Sensor Tower. The chart's fine print notes that Kalshi and Crypto.com fees are estimated (Dune data). And the regulatory fights are piling up. Multiple states are suing Kalshi, a class action alleges it's operating unlicensed sports betting, and members of Congress introduced legislation in March to block prediction markets from offering sports and war contracts. The $122M/month "prediction market" is really a $110M/month sports betting exchange running under a different regulatory label.

@Polymarket Until now, your response to bets being potentially manipulated was "if you disagree bet on the other side" But clearly if bettors did that, they could lose all their money There is no defense for users, against manipulated or insider bets - for many possible bets

Polymarket condemns the harassment & threats directed at Emanuel Fabian — or anyone else for that matter. This behavior violates our Terms of Service & has no place on our platform. We've banned the accounts for all involved & will pass their info to the relevant authorities.

@iamgingertrash @alightinastorm Elon recently admitted some errors during the hiring process, appears open to fixing them Presumably he should be using the same hiring strategies now, that made his other companies so successful Its unclear what specific remedy you are requesting. Hire less capable people?

@alightinastorm This is correct It’s incredibly odd seeing Elon make the same mistakes time and time again

xai will remain behind as long as they publicly display this exceptionalism attitude hiring only stars never makes a good team, yet they keep hiring and reposting the stars why? how has it changed, at all? didn't zuck show hiring an all star team doesn't work?

@LxngevityLab Any time someone says "read until the end" or "I bet this will blow your mind", I automatically skip reading the post

Japan is testing a drug that could grow new teeth. Please read till the end – I bet this will blow your mind: Japanese researchers developed a drug that blocks the USAG-1 protein (USAG-1 is what stops your teeth from growing after your adult teeth come in). It shuts down two signaling pathways (BMP & Wnt) that control tooth formation. When you block USAG-1, those pathways turn back on. This allows dormant tooth buds (leftover structures from development) to grow into new teeth. Turns out humans might still have the ability to grow a third set of teeth and we just need to unlock it. It sounds insane, but let me show you some results: The drug is an antibody that neutralizes USAG-1. In animal studies, it: • Reactivated dormant tooth buds • Formed new teeth • Restored missing teeth in animals born without them Animal results: • The drug worked in mice, ferrets, and dogs. • These animals were chosen because their dental structures are similar to humans. • In every case, blocking USAG-1 allowed new teeth to grow. Human trials began in 2024–2025. The first phase is small (around 30 people) and focused on safety and dosing. Who’s being tested: • Adults missing teeth • Kids born without certain teeth (congenital tooth agenesis) (It’s for people with severe tooth loss or genetic conditions – not a cosmetic tool) Now the question everyone will be asking in the comments: When is this going to be available? Researchers hope for availability around 2030 (if trials succeed), but that’s optimistic, since most experimental drugs don’t make it to market. If this works, it will change dentistry completely (you could grow a new tooth instead of getting an implant) No drilling into your jaw (which everyone hates). No foreign material. And just to keep things accurate and informative (since there is a lot of misinformation from creators and other platforms), it is NOT: • Proven safe or effective in large human trials yet • Guaranteed to work beyond congenital tooth loss • Likely to replace implants anytime soon • Known to work for teeth lost to trauma or decay Most people think this will replace all dental work tomorrow. It won’t. We’ll know more by 2030. PS. Educational purposes only. Not medical advice.

@aakashgupta What are you basing the 50% attrition rate on?

Elon has mass-fired his way to the world’s most valuable rocket company and the world’s most valuable car company. He’s now doing it to his AI lab. 10 of xAI’s 12 cofounders have left. The CFO lasted three months. The general counsel lasted 16 months. Two more cofounders walked out in the same week last month. Yesterday, he posted that xAI “was not built right first time around” and is “being rebuilt from the foundations up.” Estimated attrition: 50%. That quote would be a death sentence at any normal company. At a Musk company, it’s a timestamp. SpaceX went through three consecutive Falcon 1 explosions and a founding team that looked nothing like the one that landed Falcon 9. Tesla burned through two CEOs before Elon took over and nearly died in 2008. The pattern is: chaos, turnover, near-death, then everyone else spends a decade catching up. The current gap is real. Grok 4 Heavy was the first model to hit 50% on Humanity’s Last Exam, a benchmark designed to be unsolvable. Perfect score on AIME 2025. 15.9% on ARC-AGI-2, nearly doubling the previous best. Then every competitor leapfrogged those numbers within six months. Gemini 3.1 Pro now scores 77.1% on ARC-AGI-2. Claude Opus 4.6 scores 68.8%. The infrastructure tells you which bet he’s actually making. 555,000 GPUs in Memphis. 2 gigawatts of power. $18 billion in silicon. Built the first 100K cluster in 122 days when the quote was two years. Roadmap: 1 million GPUs by late 2026. No other single-site AI facility on Earth is at this scale. Grok 4.20 shipped four days ago with a four-agent architecture where specialized models debate each other before answering. Still in beta. The full Colossus 2 cluster won’t hit capacity until Q3. xAI hasn’t even trained on the compute they’ve already built. The telescope line is early. But the track record says bet on the infrastructure, not the org chart.

@tbpn @markgurman doubled annually for 3 years = 8x price increase Is that what you meant

Bloomberg's @markgurman says that even though Apple partnered with Google Gemini for Siri, they actually run their business on Anthropic. "Apple runs on Anthropic at this point. Anthropic is powering a lot of the stuff Apple's doing internally in terms of product development and a lot of their internal tools." "They have custom versions of Claude running on their own servers internally, too. This Google deal just came together a few months ago. They were not going to use Google. Apple actually was going to rebuild Siri around Claude. But Anthropic was holding them over a barrel. They wanted a ton of money from them, several billion dollars a year, and at a price that doubled on an annual basis for the next three years." From his January appearance on the show.