SpaceX’s $2T Case, Nvidia’s Shock Selloff, America Turns on AI, Trump Pulls AI Order, Bond Crisis?

| Podcasts | May 22, 2026 | 69.6 Thousand views | 1:42:00

TL;DR

Andre Karpathy joins Anthropic to pioneer recursive self-improving AI systems, while the industry faces a critical need to prove economic viability—now evidenced by EBIT-positive quarters—and combat rising public fear through concrete success stories rather than alarmist narratives.

🧠 Recursive AI Development Accelerates 3 insights

Karpathy joins Anthropic for self-improving models

The renowned AI researcher will lead a new pre-training team focused on recursive self-improvement, where models enhance themselves with minimal human intervention.

Weekend projects demonstrate viral influence

Karpathy's open-source auto-research tool garnered 82,000 GitHub stars and inspired widespread experimentation with recursive LLM training.

Architectural shift toward smaller networked models

Experts predict future systems will use networks of specialized small models rather than single massive ones, dramatically reducing cost per token.

💰 AI Economics Reach Inflection Point 3 insights

Anthropic achieves EBIT profitability

The Wall Street Journal reported Anthropic turned EBIT positive with roughly 80% gross margins on inference, validating AI unit economics.

LLM market projected at $200-400B ARR

Combined annual recurring revenue across major language model providers could reach $200-400 billion by year-end, excluding recommender system revenues.

GPU infrastructure shows measurable returns

The hyperscaler AI investment is now generating returns beyond ad targeting, with standalone LLM businesses proving the economic viability of massive compute spend.

🎭 Combating the AI Perception Crisis 4 insights

Public backlash hits tech leaders

AI has become so polarized that figures like Eric Schmidt were booed during university commencement speeches, reflecting growing public fear.

Regulatory capture strategies questioned

Chamath Palihapitiya suggested some CEOs promote alarmist narratives to establish regulatory moats that would entrench incumbents while excluding competitors.

Google deploys Gemini Nano quietly

Google installed a 4GB Gemini Nano model in Chrome browsers for autocomplete and proofreading without prominent user notification, sparking privacy discussions.

Industry must champion positive use cases

Advocates urged focusing on tangible benefits like a hedge fund manager using AI to cure his daughter's rare genetic disease rather than speculative fears.

Bottom Line

The AI industry must pivot from fear-based narratives to concrete stories of human benefit while recursive self-improvement technology matures, proving value through sustainable economics and measurable user outcomes rather than regulatory capture.

More from All-In Podcast

View all
Elon’s Anthropic Deal, The Next AI Monopoly?, “FDA for AI” Panic, Trading the AI Boom
1:22:02
All-In Podcast All-In Podcast

Elon’s Anthropic Deal, The Next AI Monopoly?, “FDA for AI” Panic, Trading the AI Boom

Elon Musk's xAI has struck a landmark infrastructure deal leasing Colossus 1 data center capacity to Anthropic, instantly creating a hyperscaler revenue stream while solving Anthropic's critical compute constraints. The arrangement positions Anthropic to potentially reach $100 billion ARR this year and $1 trillion by 2027, creating what could become the most valuable monopoly in history if exponential growth persists.

14 days ago · 9 points