Ihtesham Ali
@ihtesham2005 • 43,970 subscribers
investor, writer, educator, and a dragon ball fan 🐉
Videos
Sergey Brin retired in 2020 with enough money to do anything on earth. Within weeks he felt himself mentally spiraling. The thing that pulled him back became Gemini. He said staying retired would have been a big mistake. The plan was simple sit in cafes, study physics, decompress after decades of building one of the most consequential companies in history. He had earned it. Then COVID shut every cafe on earth the month after he retired. He was stuck. No intellectual outlet. No creative problem to solve. No team to build with. Just time. And he said something happened to him in that time that scared him. He felt himself losing sharpness. Spiraling. Not being as sharp as he was used to being. So he went back. Not because he had to. Not because Google needed him. Because he needed it. He started coming into the office when almost nobody else was. Spent more and more time on a project that at the time did not have a name. That project eventually became Gemini, Google's flagship AI model, the thing they are now betting their next hundred years on. He said it plainly: staying retired would have been a big mistake. Most people think the goal is to stop working. Sergey Brin had the money to stop working and learned within months that the work was not the burden. The absence of it was. --- Watch the full interview YT. Search: "Big ideas begin here: Sergey Brin at Stanford"
Ihtesham Ali230,652 views • 3 months ago
The President of Anthropic said something at Stanford that no other AI CEO will say publicly. She talked about what AI is doing to the people who use it, why she walked out of OpenAI, and what skills will actually matter when AI can do most of what we do today. 1/ What AI safety actually means
Ihtesham Ali150,207 views • 3 months ago
A student asked the CEO of Google DeepMind what he does not want AI to touch in his lifetime. He did not say art. He did not say relationships. He did not give the answer anyone expected. Here's everything he said at Stanford: 1/ AI shouldn't touch consciousness
Ihtesham Ali129,359 views • 3 months ago
Satya Nadella was asked whether most people will earn more or less after AI. He did not dodge it. His answer was two words that will either reassure you completely or keep you up at night: glue work. He has been thinking about this carefully. His framing: wages have never been about which tasks you perform. They have always been about what society decides to value. The last 250 years rewarded the accumulation of expertise. AI creates abundance of expertise. So the question becomes what humans build next that cannot be trained away. His answer is glue work. The connective tissue between automated systems. The judgment calls that sit outside any loop an agent can close. The work that holds everything together that nobody has written a spec for yet. He was not dismissing the disruption. He said plainly that real displacement is coming and the workflow will change. But his argument is that humans have always migrated toward the layer above whatever just got automated. Typists became knowledge workers. Knowledge workers become something else. The problem with his answer is the one nobody in that room pushed on. He could name the pattern. He could not name the jobs. Glue work is either the most honest answer a tech CEO has given about the future of employment. Or it is the most elegant way to say I do not know dressed up as reassurance. (Watch this on YouTube at
Ihtesham Ali74,472 views • 3 months ago
Meta signed a 20 billion dollar compute deal in April and by June was quietly trying to sell the leftovers, and Dan Niles went on CNBC to explain why that flip means the AI building boom may have already peaked. That timeline should stop you. For two years, the whole AI bet ran on one idea: that nobody has enough compute, and everyone wants more, and the demand never ends. Then Meta signed a 20 billion dollar deal in April. By June they were shopping the extra capacity to their rivals. Niles explained what broke. Companies used to burn as many tokens as they could to look busy. Then they started cutting. His example is the one that sticks. Using the smartest model to summarize your emails is like driving a Ferrari to the corner store for milk. You do not need it. Coinbase cut its AI spend in half while it used the tools more. Uber burned its whole budget in four months, then slashed it. When the biggest spenders suddenly have compute to sell, the story cracks. Everyone is still pricing these stocks like demand only climbs. The people paying the bills just found the off switch.
Ihtesham Ali65,575 views • 2 months ago
A reporter told Jensen Huang that his Uber driver wished he had just gone back to the farm instead of facing AI. Jensen flipped the entire fear in one move. The driver was genuinely scared. He looked at AI and saw the end of his work. Jensen did not argue with the fear. He pointed at where the man was sitting. That Uber job did not exist a generation ago. It only exists because of the last technology wave. Mobile. Cloud. The internet. The data centers running the routing and the matching in real time. The science behind a single Uber ride is enormous. None of it was possible before the revolution that came before this one. So the man driving for Uber, worried the next wave will erase him, is living inside a job the previous wave created. That is the whole reframe. The thing he is afraid of is the same kind of force that gave him the work in the first place. Then Jensen went further. The next wave creates new opportunity too. You cannot name it yet. Nobody could name Uber before the internet existed either. He has watched this happen his whole career. Personal computer. Internet. iPhone. Every time, people thought it would leave them behind. It changed everything. It left almost no one behind. The farm was not the safe choice. It was just the one he could already see. (Watch the full interview on YouTube at Associated Press)
Ihtesham Ali62,810 views • 3 months ago
Chamath said something that reframes the entire Anthropic-Fable story. The scandal was never about a jailbreak. It was about Anthropic personally handing Amazon, Microsoft, and Google the justification to become the permanent gatekeepers of AI. Think about what that means. AI is the single most economically leveling technology ever built. The thing that could give any person on earth access to the kind of intelligence that used to cost millions to employ. And the frontier labs have managed to bungle it so badly that they handed the three largest corporations in the world a bulletproof argument to sit between you and that technology forever. Chamath laid out the mechanism. Governments around the world now need someone to trust. The frontier labs proved they cannot be trusted. The hyperscalers step in and say let us be the gatekeepers. We will wrap it in KYC. We will verify identities. We will build the audit trails. We have the infrastructure. You cannot ask a neoscaler to build what Amazon and Microsoft have spent decades and trillions constructing. So you hand the keys to the hyperscalers. Chamath called it plainly. They doom trolled their way into it. Every press release about existential risk, every congressional hearing about AI danger, every public hand-wringing about safety created the political environment that made gatekeeping not just possible but inevitable. The most powerful tool for human economic mobility in history. Locked behind a driver's license check on Amazon Web Services. Own goal. WATCH THE FULL PODCAST ON The All-In Podcast
Ihtesham Ali47,277 views • 3 months ago
A Stanford PhD student built a system that turns any research paper into a working AI agent. It's called Paper2Agent. I watched her demo it live and couldn't believe what I was seeing. Here's exactly what happened. She pasted a 40-page NeurIPS paper into the tool. Within seconds it extracted the core method, identified the dataset, and started scaffolding agent code that actually implements the paper's approach. But the wild part was what came next. She typed: "Apply this method to my dataset and answer questions like the paper's author would." The agent didn't just summarize. It ran the methodology. On her own data. And when she asked it why it made certain decisions, it cited the exact sections of the paper. What normally takes a PhD student 3 weeks of implementation just happened in under an hour. She's not smarter than other researchers. She just stopped reading papers and started running them. (Link in the comments)
Ihtesham Ali67,959 views • 6 months ago
Demis Hassabis said that 90% of the breakthroughs the entire AI industry runs on came out of Google. Every competitor is building on work his teams did first. He said it without flinching. Over the last decade, he claims 90% or more of the big breakthroughs that underpin modern AI came from Google Brain or DeepMind. Transformers. AlphaGo. The reinforcement learning everything now depends on. Transformers are the T in GPT. The architecture that made every large language model possible came out of Google in 2017, in a paper called Attention Is All You Need. His point was simple. Every rival lab is running on foundations his people poured. But here is what he left out. Of the eight authors who wrote that transformer paper, all eight left Google. They went on to found or join OpenAI, Anthropic, and their own startups now worth billions. Google invented the engine. Then watched the inventors walk out the door and build the competition. He says Google still has the biggest research bench of any lab on earth, and the talent war is the most ferocious tech has ever seen. Everyone treats this race like the labs are evenly matched. The man who started fifteen years early does not see it that way. The people who left with his blueprint might be the reason he has to keep proving it. (Watch the full interview at Semafor on YouTube)
Ihtesham Ali35,961 views • 3 months ago
Unreal Engine asset packs are about to be worthless. Tencent just shipped HY-World 2.0. You type one sentence. It outputs a real 3D scene with meshes, point clouds, and 3DGS that drags straight into your game engine. Renders in real time on a consumer GPU. Phone video in, digital twin out, one forward pass. The marketplace economy of selling 3D assets just had its Napster moment.
Ihtesham Ali45,956 views • 4 months ago
Chamath accused Sam Altman and Dario Amodei of running a three-act playbook where apocalyptic AI warnings were timed to fundraising rounds, not safety concerns. If you think this is a fringe take then you are so wrong. This is one of the most connected investors in Silicon Valley saying the quiet part out loud. Act one: a lab needs money. So they seed the conversation with apocalyptic language. AI could end humanity. Regulation is urgent. The threat is real. The press runs it. Policymakers panic. Act two: the agitators are now fully wound up. Critics of the other lab pile on. Their fundraising process gets complicated. Their model release gets scrutinized. The chaos benefits whoever started it. Act three: the same lab that warned about extinction flips the narrative. Here comes the mythical new release. It will solve everything. It could also end the world. Investors clamor to get in before it does both. Sam Altman said it himself in April. He called Anthropic's safety rhetoric fear-based marketing used to justify concentrating AI control among self-declared trustworthy companies. The CEO of the rival lab gave the same diagnosis. Chamath called it deeply selfish. His reasoning was simple. These labs took the most important economic tool of our lifetime and wrapped it in their own personal agendas. Every person who could benefit from this technology watched the people building it turn it into a weapon for settling scores. The doom cycle was never about safety. It had a release schedule. Watch the full show here on The Axios Show:
Ihtesham Ali28,776 views • 3 months ago
David Sacks laid out the cleanest theory about why Anthropic keeps calling for government regulation of AI. The answer has nothing to do with safety and everything to do with market structure. Anthropic spent months writing blog posts warning that AI was dangerous. Dario gave interviews about existential risk. He published a piece calling for an FAA-style agency to approve all AI models before release. He primed government officials to treat frontier AI as a threat requiring oversight. Then one of Anthropic's own most trusted partners reported a credible jailbreak from Fable 5. And the government did exactly what Dario had spent months conditioning them to do. They rolled it back. Sacks called it on the All-In podcast. Dario got exactly what he wanted. The FAA for AI is not a safety mechanism. It is a moat. A government approval process for new model releases does not hurt Anthropic. They already have the models. It hurts every competitor who does not. It hurts open source models that cannot be regulated because there is no company to regulate. It hurts the Chinese labs only insofar as they care about the American market at all. The only companies that benefit from a labyrinthine government approval process are the ones already at the frontier who can afford to wait out the review cycle. That is Anthropic. That is OpenAI. Nobody else. The proof is in what they did not do. Chimath pointed it out directly. If you are genuinely worried about misuse, you implement know-your-customer verification. You make people identify themselves before accessing the most powerful models. Anthropic could have done that tomorrow. They did not. They do not want KYC. KYC is transparent. KYC can be audited. KYC gives users due process. What they built instead was an invisible surveillance system that profiles you, degrades your access without telling you, and asks the government to make sure no one else can offer you an alternative. If you thought this was safety then you are wrong. That is capture. Sacks said the response should be simple. Fix the jailbreak, come back to market, and do not reward Dario with the regulatory architecture he has been engineering for years. We will see if anyone is listening. WATCH THE FULL PODCAST ON The All-In Podcast
Ihtesham Ali25,284 views • 3 months ago
Dario Amodei just fired back at everyone calling his job warnings doom marketing. He did not take it quietly. "I want to be really clear and push back hard against this. In every interview I talk about the possible ways to address these risks. I have five pages where I lay out the difference between tasks and jobs." Jensen Huang said he was scaring people. Others said it was cheap marketing that benefits Anthropic. His response was sharp. "The idea that this is cheap marketing is itself cheap marketing. It's part of the disease of Silicon Valley. It's caught up in this social media world of three seconds." That last line landed differently than I expected from a Silicon Valley CEO. He is calling out the entire culture he operates inside. The three second clip economy. The incentive to reduce careful thinking into a take. Then he said what his message actually is. "My message is definitely not doom is coming. My message is this is something we should see coming, that we're worried about, and that we need to actually respond to positively." There is a difference between warning people and scaring them. He thinks he is doing the first. His critics think he is doing the second. Both of them are right about something.
Ihtesham Ali21,856 views • 3 months ago
Demis Hassabis has a test for whether AI is actually creative. Give it only the data Einstein had in 1901 and see if it invents relativity. He thinks today's text models will never pass it. This is his bar for true creativity. Not extrapolating something already known. Generating a genuinely new hypothesis about reality that nobody has reached before. The thing Einstein did in 1905 with a handful of papers that rewrote physics. Hassabis wants to hand an AI the same starting data Einstein had and see if it gets there on its own. No peeking past the cutoff. Then he explained why he doubts text alone can do it. Einstein daydreamed his way to relativity. He was a patent clerk imagining what it would look like to ride alongside a beam of light. He ran the experiment in his head first, then proved it with math. Yann LeCun, who won the Turing Award and ran AI at Meta, says the same thing in blunter terms. He says today's models don't understand the physical world as well as a housecat, and that text alone is "complete BS" as a path to real intelligence. Two of the most decorated minds in AI agree the missing piece is the same. Hassabis thinks an AI needs to understand the world of atoms, not just the world of bits. Most of the industry is racing to scale text models. Hassabis just told them the most important kind of thinking might live somewhere text can't reach. --- Watch the full interview at Semafor on YouTube
Ihtesham Ali18,905 views • 3 months ago
Alex Karp went on TV and said his enterprise customers are done letting AI companies quietly steal the alpha out of their business, and the CEO of Together AI explained on CNBC exactly how that theft works. Karp said his customers are livid. They feel they pay for tokens that create nothing while handing over the secret sauce of their own company. Then Vipul Ved Prakash laid out the mechanism. Every time you send your data to a company with very smart models, you hand them the recipe to rebuild what you do. Your data is the thing that makes you money. And you are pushing it straight into a competitor's system. The smarter the models get, the more they learn from what you feed them. This is why companies are moving to open models, and it was never really about price. You can download an open model, run it inside your own walls, and never let your data leave the building. It scores 81 against the top closed model's 85, and runs up to sixty times cheaper. Half the Fortune 500 is already running open models from Hugging Face. The story was that the labs win because they own the best models. The companies paying them just realized they owned the best data the whole time.
Ihtesham Ali17,433 views • 2 months ago
Anthropic's safety warnings are marketing. That is what Clem Delangue said. He runs Hugging Face, the largest open source AI platform on earth. And he said it is fair for regulators to start cracking down, because Anthropic and OpenAI have been doing doom marketing for years. He pointed to GPT-2. Five or six years ago. Already deemed too dangerous to release. The same playbook has been running ever since. His point was simple. These are now some of the fastest growing companies in the world, on track to be among the most valuable by year end. They can handle scrutiny. But he drew a hard line. He warned against spreading the same restrictions to startups, universities, and small labs that do not have the legal teams these giants have. His read on the bigger picture was blunt. The most dangerous capabilities sit behind closed doors at a handful of frontier labs. Open source is easier to evaluate, harder to hide problems in, and spread across thousands of smaller players instead of a few companies concentrating power. Watch the full interview on Bloomberg Technology
Ihtesham Ali13,198 views • 2 months ago
A Navy SEAL admiral gave a graduation speech that has been watched tens of millions of times. His first piece of advice: make your bed. Admiral William McRaven served 36 years as a SEAL. He oversaw the raid that got Osama bin Laden. And the lesson he opened with was about straightening sheets. If you make your bed every morning, you have completed the first task of the day. One small win triggers the next. By evening, one task done has become many. And if your day is miserable, you come home to a bed that is made. Proof that tomorrow can be better. Changing the world does not start with big moves. It starts with proving to yourself, every single morning, that you do what you say.
Ihtesham Ali13,681 views • 3 months ago
In September 1997, Apple was 90 days from bankruptcy, the stock was at $3.30, and Michael Dell had publicly said the company should be shut down and the money returned to shareholders. Steve Jobs had been back for 8 weeks. No title. No salary. Technically just an advisor. He walked on stage that month, slept three hours the night before and gave a 16-minute speech that almost nobody has watched. It is the speech that saved Apple. He did not show a product. He did not show a chip. He did not show a roadmap. He spoke about one idea. Marketing is about values. Not features. Not specs. Not megahertz. He said the world had become so noisy that no company on Earth was going to get a chance to tell people more than one thing about itself. So you had to be very clear about what that one thing was. Then he said the line almost nobody quotes from that morning. Even a great brand needs investment and caring if it is going to retain its relevance and vitality. The Apple brand had clearly suffered from neglect. He admitted on stage, to his own employees, that the company they worked for had stopped caring about the thing that made it matter. Then he ran the ad. Here is to the crazy ones. The misfits. The rebels. The troublemakers. The round pegs in the square holes. The ones who see things differently. When the tape stopped, the room was silent for a few seconds. Then they stood up. The thing he did next is what most people miss when they tell this story. He had personally called Yoko Ono to get permission to use John Lennon. He had called the estates of Einstein, Gandhi, Picasso, Edison, Amelia Earhart, Martin Luther King. Almost none of them had ever appeared in an advertisement before. Almost all of them said yes to Apple specifically, when they had said no to everyone else who had ever asked. He said on stage that morning that he did not think any other company on Earth could have run that campaign. He was probably right. The campaign broke on Sunday night during the network premiere of Toy Story on ABC. The ad ran twice. Print followed in the Wall Street Journal, the New York Times, USA Today. Billboards went up in five cities. Buses with Rosa Parks' face on them started driving through Manhattan. Apple did not announce a new computer that quarter. They announced who they were. 18 months later they shipped the iMac. 3 years later the iPod. 6 years later the iTunes Store. 10 years later the iPhone. The most valuable company in the history of capitalism was rebuilt on a 16-minute talk where the founder did not show a single product. Everyone quotes the Stanford commencement speech from 2005. The one about staying hungry and staying foolish. That one made him a philosopher. The 1997 speech is the one where he saved the company. He told his employees the company had lost its soul. He told them what the soul was. He told them they were going to spend a fortune reminding the world. Then he walked off stage and went to work. The difference between a company that dies and a company that becomes the most important company in the world is sometimes one person, on three hours of sleep, willing to stand in front of his own team and say we forgot who we are. The crazy ones changed things because somebody believed they could. That somebody was him.
Ihtesham Ali13,884 views • 4 months ago
Elon Musk said humanity is currently using less than a trillionth of the sun's power output. A trillion is a million times a million. That is how far we have to go. He was explaining the Kardashev Scale on a SpaceX livestream. A Russian physicist named Nikolai Kardashev invented it in 1964 as the most objective way to measure how advanced a civilization actually is. Not by population. Not by GDP. By energy. Type 1: you harness everything your planet produces. Type 2: you harness your star. Type 3: you harness your galaxy. Where does humanity sit right now? We are not Type 1. We are not close to Type 1. We are a tiny fraction of the way to Type 1. Musk said if aliens visited us today and measured us on this scale, we would not be registering. His exact words: "One microKardashev would be an epic achievement relative to where we are right now." The sun is 99.86% of all mass in the solar system. Everything else, Jupiter, Saturn, Earth, every planet and moon and asteroid combined, is the remaining 0.14%. Earth is in the miscellaneous category. Of the sun's total power output, Earth intercepts roughly half a billionth. And of that half billionth, most hits ocean. Most of the land is Siberia or Antarctica. The actual usable surface where you can capture solar energy is tiny. To get to even a millionth of the sun's power, you cannot stay on Earth. The math does not work. You have to go to space. That is not a dream. That is a physics constraint. And that constraint is exactly why SpaceX exists.
Ihtesham Ali11,145 views • 3 months ago
Every rocket ever built before Starship was thrown away after one use. Elon Musk just explained on a SpaceX livestream why that single fact made space exploration practically impossible until now. His analogy was the one that landed hardest. Imagine if you had to throw away a commercial airplane every time it flew. Not maintain it. Not refuel it. Throw it away and build a new one. What happens to the price of a ticket? What happens to the airline industry? What happens to global travel? It collapses. Nobody flies. That is exactly what every space program in history has been doing with rockets. Building them once, using them once, throwing them away. The cost was not a failure of ambition. It was a structural consequence of how the hardware worked. Starship is the first rocket ever designed for full and rapid reusability. Not just reusable. Rapidly reusable. The rocket lands, gets caught by the tower, goes back on the launch stand, and flies again without refurbishment. Like an aircraft. No laborious inspection between flights. Starship B3 is already more than double the thrust of the Saturn 5 moon rocket. Version 4 will be roughly three times. And Musk said they expect Starship to be flying more than once per hour eventually. That number is hard to absorb. The most powerful moving object ever built. Flying more than once per hour. The cost of getting mass to orbit does not drop gradually when that happens. It collapses. And when the cost collapses, everything that was previously impossible becomes a question of how fast you can build. Reusability is not a feature. It is the entire unlock. Every other problem in space is downstream of this one.
Ihtesham Ali10,431 views • 3 months ago