Deep Dive
The Unprecedented Capex Boom
The creator opens by framing AI as an asteroid about to hit the planet, fundamentally different from all previous technology revolutions because humanity has never tried to industrialize intelligence before. While railroads and telephones were massive capex cycles, this one operates at a different order of magnitude — both in scale and in the radical economic restructuring it's causing. Google, Meta, Amazon, and Microsoft are collectively committing $725 billion to AI capex in 2026, dwarfing Tesla's $25 billion increase. To put this in perspective, $725 billion exceeds the military budgets of major nations and the GDP of entire countries. This is not just bigger numbers; it's a phase change in what the global economy invests in. Unlike past waves that built physical infrastructure, this one builds both the physical world and an intelligence layer on top of it. The demand for intelligence is infinite, making this a once-in-a-lifetime or once-in-human-history opportunity.
Google's Dominance in the AI Arms Race
Google emerges as the most interesting hyperscaler, with multiple simultaneous plays driving value creation. CEO Sundar Pichai revealed that 75% of Google's new code is now AI-generated, signaling deep internal adoption. Thomas Kurian's Google Cloud division has grown to nearly 20% of Alphabet's revenue and is unveiling aggressive new products in agentic AI. Google's backlog is nearly at half a trillion dollars and doubling each quarter, showing exponential demand growth. Beyond their core AI business, Google functions as an excellent venture capitalist with outsized returns — they invested $900 million in SpaceX that could return $100+ billion-plus, and they've backed Data Bricks, Stripe, Epic Games, Planet Labs, and Tempest AI. The creator estimates that half of Google's $60 billion Q1 revenue comes from these investments alone, making them a multi-layered AI beneficiary. As the creator puts it, if it's good enough for Google to invest in, it's probably a solid bet.
Semiconductor Dominance and Chip Supply Chains
Broadcom recently hit a $2 trillion market cap, scaling from relative obscurity a few years ago thanks to long-term deals with Google (TPU agreements extended through 2031), Meta (building custom AI chips), and Anthropic (compute access deals). Nvidia reached $5 trillion market cap while Broadcom hit $2 trillion, with Nvidia alone now larger than India's entire stock market despite India having 1.4 billion people. The creator argues that the best way to play this AI wave in the near term is through semiconductors — the picks and shovels. Beyond Nvidia, TSMC, and Broadcom, Google is building custom chips, Tesla will become a massive chip player, and Meta is entering the space. AMD and other semiconductor makers are creating unprecedented wealth. The creator notes this is a multi-year structural tailwind where chip manufacturers benefit from the foundational infrastructure race happening across all hyperscalers.
Real-World AI Deployment and Winners vs Losers
The narrative shifts from picks-and-shovels to actual real-world AI applications. Unsupervised robotaxi fleets are expanding with 25 of 649 tracked cars now fully autonomous (19 in Austin, 3 in Dallas, 3 in Houston), and the creator notes that once companies get comfortable, they can scale thousands overnight by flicking a switch. Humanoids are scaling faster than expected — Boston Dynamics-adjacent companies are 24x-ing manufacturing in 120 days from one robot per day to one per hour. Tesla is deploying two humanoid production lines starting July in Fremont and Austin with 10 million unit capacity goals. However, not all AI companies are winning. Anthropic is crushing revenue targets and revising forecasts upward every four months, now at $30 billion annual revenue. OpenAI missed revenue and user targets while Sam Altman fights in court over ownership disputes. The creator warns against buying OpenAI's IPO, calling it a shrivel-and-die risk, because they need exit liquidity before problems compound. Winner-takes-most dynamics are real.
The Cost Collapse and Future Scaling
AI inference costs have plummeted 91% in just eight months, dropping from $3.50 to 32 cents per million tokens — the creator calls this the most violent statistic of the day. This cost floor will continue dropping until intelligence becomes so cheap it can't even be metered or counted, essentially given away. Despite this price collapse, Anthropic is exploding in growth, revising their annual recurring revenue forecast from $9 billion to $30 billion in just four months — a 3.3x increase. Grok is scaling with 500 billion parameters today and plans to reach 10 trillion by year-end, a 20x jump, as part of XAI/SpaceX's anticipated mega-IPO in June. Some people already believe we're at AGI; others say it's imminent by year-end. The creator emphasizes that what's coming will be incomprehensible in scale — the inflection is accelerating, not slowing.