Linas Beliūnas's banner

Linas Beliūnas

@linasbeliunas • 11,828 subscribers

Connecting & deconstructing all things Finance + Tech. FinTech, Financial Technology, and AI constantly meets here on X & in my daily newsletter👇

Videos

Anya Rossi

sweetdream.ai

SweetDream.ai•Sponsored•Livecam

Watch Anya Live

Anya is streaming live right now! Join her private show and enjoy exclusive content.

Exclusive private shows

1.2k viewers online

Private Show

Join now for exclusive access

Free preview available • Premium content

Instead of watching 1 hour of Netflix tonight, watch this ex-Google Chief Scientist Jeff Dean’s lecture. It’s the clearest explanation I’ve seen of the full AI engineering stack - from building LLMs from scratch all the way to one human coordinating 100 agents. The best part is that it’s useful whether you’ve never touched a model or you’ve been shipping agent systems every day for the past year. Bookmark it & watch the whole lecture this weekend, because it might end up being the most valuable thing you learn all week.

Instead of watching 1 hour of Netflix tonight, watch this ex-Google Chief Scientist Jeff Dean’s lecture. It’s the clearest explanation I’ve seen of the full AI engineering stack - from building LLMs from scratch all the way to one human coordinating 100 agents. The best part is that it’s useful whether you’ve never touched a model or you’ve been shipping agent systems every day for the past year. Bookmark it & watch the whole lecture this weekend, because it might end up being the most valuable thing you learn all week.

Linas Beliūnas

594,938 views • 4 days ago

Jeff Dean explaining Google built a ChatGPT-style chatbot roughly a year before OpenAI released theirs. Tens of thousands of Google employees were already using it and loved it. Leadership still killed any public launch because: • “What if people stop searching on Google?” • “What if it says inaccurate or ‘bad’ things?” Google invented the Transformer architecture, and still threw away the first-mover advantage for nothing. Innovator's Dilemma at its finest.

Jeff Dean explaining Google built a ChatGPT-style chatbot roughly a year before OpenAI released theirs. Tens of thousands of Google employees were already using it and loved it. Leadership still killed any public launch because: • “What if people stop searching on Google?” • “What if it says inaccurate or ‘bad’ things?” Google invented the Transformer architecture, and still threw away the first-mover advantage for nothing. Innovator's Dilemma at its finest.

Linas Beliūnas

388,923 views • 7 days ago

Google co-founder Sergey Brin rarely speaks publicly. He just sat down for an unscripted Q&A on Frontier AI and admitted something most lab leaders won’t: Even the people building these models do not fully understand what they have created. In this 29-minute conversation at AGI House, Brin walks through the surprises that actually matter right now: ↳ Specialized models are converging into one general system faster than anyone predicted. - Train on coding and math reasoning mysteriously improves. - Feed it images and geometric word problems get better. - The capabilities bleed into each other in ways nobody fully engineered. ↳ One of the biggest leaps came from the dumbest-sounding trick imaginable: - Just telling the model to “think step by step.” - Brin says there was no obvious reason it should work. It did. ↳ He pushes back on hype around superintelligence (it still can’t solve the impossible), notes that AI mastering a domain has never stopped humans from getting better at it (chess after Deep Blue, Go after AlphaGo), and says something close to transformers is probably enough to reach AGI. ↳ Inside Google, they are already using the AI to build the AI. - That self-improvement loop is where Brin spends most of his time. - World models and physical interaction are the missing piece for the version of AGI that can do anything a person can. Candid, technical, and free of the usual marketing. One of the clearest looks at how the people actually shipping frontier models are thinking in real time. Instead of another Netflix series tonight, watch this talk.

Google co-founder Sergey Brin rarely speaks publicly. He just sat down for an unscripted Q&A on Frontier AI and admitted something most lab leaders won’t: Even the people building these models do not fully understand what they have created. In this 29-minute conversation at AGI House, Brin walks through the surprises that actually matter right now: ↳ Specialized models are converging into one general system faster than anyone predicted. - Train on coding and math reasoning mysteriously improves. - Feed it images and geometric word problems get better. - The capabilities bleed into each other in ways nobody fully engineered. ↳ One of the biggest leaps came from the dumbest-sounding trick imaginable: - Just telling the model to “think step by step.” - Brin says there was no obvious reason it should work. It did. ↳ He pushes back on hype around superintelligence (it still can’t solve the impossible), notes that AI mastering a domain has never stopped humans from getting better at it (chess after Deep Blue, Go after AlphaGo), and says something close to transformers is probably enough to reach AGI. ↳ Inside Google, they are already using the AI to build the AI. - That self-improvement loop is where Brin spends most of his time. - World models and physical interaction are the missing piece for the version of AGI that can do anything a person can. Candid, technical, and free of the usual marketing. One of the clearest looks at how the people actually shipping frontier models are thinking in real time. Instead of another Netflix series tonight, watch this talk.

Linas Beliūnas

547,924 views • 10 days ago

Monzo & GoCardless co-founder Tom Blomfield just joined Anthropic. In just 13 minutes, he explains exactly how to build a self-improving, AI-native company. Tom recently served as YC’s General Partner, so he clearly walks through how to create recursive, self-improving AI loops, and why founders who get this right will run companies that improve while they sleep. One of the best videos available to date. From the person who founded two iconic fintech unicorns.

Monzo & GoCardless co-founder Tom Blomfield just joined Anthropic. In just 13 minutes, he explains exactly how to build a self-improving, AI-native company. Tom recently served as YC’s General Partner, so he clearly walks through how to create recursive, self-improving AI loops, and why founders who get this right will run companies that improve while they sleep. One of the best videos available to date. From the person who founded two iconic fintech unicorns.

Linas Beliūnas

56,696 views • 29 days ago

Kimi CEO Zhilin Yang nailed it: "Every AI lab like Anthropic thinks the model is what matters most. That's wrong. It's how you organize the people building it that wins. And that's about Kimi 3.” In this keynote, Moonshot AI founder breaks down exactly how to stop treating the model as the only thing that matters and start winning through superior organization instead. The key? Long context is basically the AI era’s version of RAM - the same jump from 128K to gigabytes, just compressed into a couple of years instead of forty. He traces the idea back to a 1970s Intel chip nobody wanted to buy to show that your biggest advantage, and perhaps your only advantage, is your organization. One of the clearest, most practical frameworks for AI strategy and org design I’ve seen. Instead of another Netflix series tonight, watch this talk.

Kimi CEO Zhilin Yang nailed it: "Every AI lab like Anthropic thinks the model is what matters most. That's wrong. It's how you organize the people building it that wins. And that's about Kimi 3.” In this keynote, Moonshot AI founder breaks down exactly how to stop treating the model as the only thing that matters and start winning through superior organization instead. The key? Long context is basically the AI era’s version of RAM - the same jump from 128K to gigabytes, just compressed into a couple of years instead of forty. He traces the idea back to a 1970s Intel chip nobody wanted to buy to show that your biggest advantage, and perhaps your only advantage, is your organization. One of the clearest, most practical frameworks for AI strategy and org design I’ve seen. Instead of another Netflix series tonight, watch this talk.

Linas Beliūnas

30,887 views • 22 days ago

Aswath Damodaran on AI bubble: “Dot-com was an equity bubble. AI is becoming an infrastructure debt bubble.” In the 1990s, many internet companies raised money, built websites, burned cash, and disappeared. When the bust came, shareholders got wiped out. AI is different, because this boom is not just apps and pitch decks. It is data centers, GPUs, power contracts, fiber, leases, project finance, and private credit. The numbers are wild: ↳ Goldman Sachs estimates hyperscalers may spend ~$5.3T on AI/data center capex from 2025–2030 ↳ Morgan Stanley estimates ~$2.9T in global data center construction from 2025–2028 ↳ Top hyperscalers are projected to spend ~$600B in 2026 alone, with roughly 75% tied to AI infrastructure Dot-com asked investors to believe in traffic. AI is now asking lenders to believe in utilization. If demand keeps compounding, the buildout looks genius. But if demand disappoints, the problem is not just “stocks go down.” It becomes stranded capacity. Refinancing pressure. Private credit stress. Defaults. AI may end up becoming the defining technology platform of our lifetime. But the bubble risk is not where most people are looking.

Aswath Damodaran on AI bubble: “Dot-com was an equity bubble. AI is becoming an infrastructure debt bubble.” In the 1990s, many internet companies raised money, built websites, burned cash, and disappeared. When the bust came, shareholders got wiped out. AI is different, because this boom is not just apps and pitch decks. It is data centers, GPUs, power contracts, fiber, leases, project finance, and private credit. The numbers are wild: ↳ Goldman Sachs estimates hyperscalers may spend ~$5.3T on AI/data center capex from 2025–2030 ↳ Morgan Stanley estimates ~$2.9T in global data center construction from 2025–2028 ↳ Top hyperscalers are projected to spend ~$600B in 2026 alone, with roughly 75% tied to AI infrastructure Dot-com asked investors to believe in traffic. AI is now asking lenders to believe in utilization. If demand keeps compounding, the buildout looks genius. But if demand disappoints, the problem is not just “stocks go down.” It becomes stranded capacity. Refinancing pressure. Private credit stress. Defaults. AI may end up becoming the defining technology platform of our lifetime. But the bubble risk is not where most people are looking.

Linas Beliūnas

41,948 views • 1 month ago

No more content to load