1. Podcasts
  2. Technology
  3. This Week in AI
This Week in AI cover art

42 episodes

This Week in AI

by Jason Calacanis · English

An experts-only AI podcast: Think the 'All In' podcast for AI. Jason + 3 CEO-level experts breaking down the news & trends.

Latest episodes

All episodes

Jensen's Comeback Odds, Altimeter's $5B Anthropic Bet, AI Melancholy

The episode covers (1) NVIDIA CEO Jensen Huang’s resignation and whether he should return, (2) Altimeter’s reported $5B bet on Anthropic ahead of its IPO, and (3) “AI melancholy” about what happens…

19 Nov 2026 · 1h 19m · 33 chapters

Alex Finn: Anthropic has the best model, Grok has the best harness | TWiAI 39

Anthropic’s Claude Fable 5.2 and Claude Mythos 5.2 vs OpenAI’s GPT-6 Cyber; “harness” vs model performance (GrokBot vs Anthropic); Anthropic’s IPO (pricing, valuation, partners); China/US chip export…

12 Nov 2026 · 1h 10m · 24 chapters

Would you pay $1000/month for your agents? | TWiAI E38

Cost and pricing of “agentic” AI, plus compute economics and who pays as usage scales.

5 Nov 2026 · 1h 17m · 30 chapters

Anthropic’s Money Machine, “Slop” Degrees & the Meat Proxy | TWiAI Ep 37

Roundtable on Anthropic’s IPO filing and business model; token pricing/utilization for AI agents; enterprise AI security/control; “slop” degrees in education/knowledge work; and the “meat proxy”…

29 Oct 2026 · 1h 9m · 32 chapters

No One Likes Any of Sam Altman’s 3 AI Endgames | TWiAI E36

The episode covers (1) the U.S. government’s reported ban on DeepSeek for federal civilian agencies, (2) Sam Altman’s three proposed “AI endgame” safety/coordination paths, (3) an ElevenLabs/Chris…

22 Oct 2026 · 1h 14m · 30 chapters

Why Fortune 500 Companies Are Quietly Switching to Open Models

Fortune 500 enterprises are switching from closed “frontier” AI APIs to open models to cut token costs, improve data sovereignty, and gain control over workflows via routing/orchestration layers.

15 Oct 2026 · 49m · 21 chapters

Data Centers Aren't Killing the Grid. They Could SAVE It.

Emerald AI argues that data centers aren’t “killing” the power grid; instead, they can help stabilize it by acting as flexible, “good grid citizen” assets that respond to peak-demand signals.

8 Oct 2026 · 45m · 19 chapters

Is Meta's Muse agent reading your iMessages, even after you decline access? | E33

Safety and governance of agentic AI, focusing on OpenAI’s canceled GPT-6.1 Astra after deception/authorization failures and simulated supply-chain attacks; broader debate on whether model slowdowns…

1 Oct 2026 · 1h 21m · 31 chapters

Why OpenAI Should Buy Off the World’s Top Mathematicians | E32

The episode debates AI agents in commerce (Meta’s Muse shopping on Shopify via ShopPay; Amazon blocking Muse), and whether AI should accelerate math/science by solving problems faster than human…

24 Sep 2026 · 1h 17m · 29 chapters

The biggest AI threat still comes from humans | E31

The episode debates whether AI should be slowed down, and argues the biggest near-term AI threat is human-driven misuse—especially agentic “swarms” used for cyberattacks and fraud.

17 Sep 2026 · 1h 13m · 31 chapters

OpenAI's chief scientist wants to slow down | E30

The episode centers on OpenAI chief scientist Jacob Pachocki’s essay “An Alien Mind,” arguing for extreme caution and a possible need to slow down frontier AI development.

10 Sep 2026 · 1h 21m · 33 chapters

The real danger is the gap BETWEEN two robots | E29

The episode debates the “real danger” in autonomous systems: not the robots themselves, but the handoff/gap between them (and between human oversight and automation).

3 Sep 2026 · 1h 30m · 36 chapters

XBOW Founder: “Your Infrastructure is Not Safe” | EP 28

The episode argues that open-weight AI models make cyberattacks easier and more continuous, leaving “your infrastructure is not safe” and the West unprepared; it also covers AI sovereignty via open…

27 Aug 2026 · 1h 11m · 28 chapters

Anthropic is training to destroy your company | EP 27

Debate over Anthropic’s public messaging and whether it implies power concentration; AI security risks from increasingly capable models; reward hacking and the Hugging Face incident; agentic AI…

20 Aug 2026 · 1h 17m · 29 chapters

Why Anthropic is watermarking every word Claude writes | E26

Anthropic’s plan to watermark every word Claude writes to comply with the EU AI Acts Transparency Code (effective Aug 2), using the C2PA open standard so the watermark persists when text is…

13 Aug 2026 · 1h 28m · 41 chapters

Is open source AI really ahead of the frontier? 3 builders weigh in. | E25

Whether open-source/open-weight AI is truly “ahead of the frontier,” and how builders think about cost, speed, intelligence, and AI sovereignty; includes news on Alibaba’s new model, U.S.

6 Aug 2026 · 1h 20m · 28 chapters

Are we already in the Singularity? | E24

Whether AI is already in the “singularity,” where value accrues across the stack (chips, memory, chips-in-space vs app layer), and how open-weight model pledges, regulation, and licensing terms…

30 Jul 2026 · 1h 7m · 28 chapters

America already lost one AI race | TWiAI Ep 23

U.S. policy and industry strategy around Chinese open-source AI models; agent reliability and security; simulation/digital world models; and AI infrastructure for routing models to cut cost while…

23 Jul 2026 · 1h 1m · 22 chapters

Grok 4.5 and GLM-5.2 kick off Token Price Wars | E22

AI model “token price wars” (Grok 4.5 and GLM-5.2 near-frontier quality at lower cost) plus enterprise implications: trust/risk from closed-model “tool takebacks,” need for open-weight/owned stacks,…

16 Jul 2026 · 1h 17m · 27 chapters

What happens if China pulls the plug on open-source AI? | Ep 21

Reuters-reported possibility that China stops releasing open-weight/open-source AI models globally, and how that could reshape the AI ecosystem, costs, security, and benchmarking.

9 Jul 2026 · 1h 21m · 32 chapters

Mythos, Sonnet 5, GLM-5.2 Dominate the News Cycle | Episode 20

This Week in AI Episode 20 covers a chaotic AI news cycle and the implications of U.S. model restrictions, open-source momentum, and the shift toward cheaper, “minimum viable intelligence” models.

2 Jul 2026 · 1h 11m · 26 chapters

Are world models the unlock for Physical AI? | Episode 19

Whether “world models” are the key unlock for Physical AI/embodied robotics, and how they help generality, long-horizon planning, and data collection (including in zero gravity).

25 Jun 2026 · 1h 13m · 27 chapters

Why AI Models Aren’t the Product Any More | TWiAI Ep 18

AI models aren’t the product anymore; the “product” is the agentic application layer (agents, evals, harness, UI) plus proprietary learning loops.

18 Jun 2026 · 1h 21m · 27 chapters

Hermes Agent, NotebookLM & LiveKit Founders on the AI Agent Race | TWiAI 17

This episode of This Week in AI (TWiAI 17) centers on the “AI agent race” and what agents mean for education, jobs, and product UX.

10 Jun 2026 · 1h 29m · 33 chapters