TBPN
TBPN

Sep 11, 2026 · 1h 46m

AI risk meets robot arms, agent investing and chip competition

Fruitfly Hard Takeoff, Washington on AI Risk, 𝕏 Timeline Reactions | Thijs Simonian, Alex Heath & Guy Oseary, Mitesh Agrawal

The episode connects abstract arguments about AI safety to hands-on robotics, emerging agent businesses and the hardware race challenging NVIDIA.

3 key takeaways
  1. 1AI risk proposals hinge on whether doomer predictions can produce testable market signals.
  2. 2A low-cost robot arm shows how models trade reasoning quality against latency in physical tasks.
  3. 3Agent investing and inference-chip competition reveal how AI’s next phase depends on distribution, financing and infrastructure.

Don't miss

Thijs Simonian explains how Codex controls a camera-equipped robot arm to paint while balancing image costs, latency and action planning.

The brief

The hosts open with a 2001 Wall Street Journal time capsule before asking whether public reactions to AI and data centers resemble earlier moments of technological fear.

A debate over compute caps and AI doomerism turns on falsifiability: if existential-risk predictions are serious, what observable market prices or actions should support them?

The fruit-fly connectome sharpens the ethical question: if an organism can be simulated, does interacting with its digital counterpart create new obligations around consciousness and harm?

OpenAI robotics engineer Thijs Simonian demonstrates a camera-equipped arm that paints, exposing the practical trade-off between model reasoning, image-token costs and real-time control.

Guy Oseary and Alex Heath discuss personal agents and venture pattern recognition, while Positron’s Mitesh Agrawal explains the financing, software and scale challenges behind an NVIDIA alternative.

The episode’s throughline is AI becoming physical and commercial: safety claims meet experiments, agents meet markets, and inference chips meet the demands of deployment.

What was said on this episode

54 statements · 37 positive · 8 negative · 1 mixed · 8 neutral

  1. John Cooganon AGI compute scaleNeutral16:28

    Large-scale compute is considered necessary for AGI

    “Scale. That scale is a prerequisite.”

    Listen at 16:28

  2. John Cooganon AI slowdown proposalsNegative19:45

    AI slowdown proposals might cause regulatory capture, a stock-market crash, or delayed gains.

    “It might create regulatory capture for a few major players. It might crash the stock market or delay economic gains”

    Listen at 19:45

  3. John Cooganon AI slowdown proposalsNegative19:45

    AI slowdown controls might create regulatory capture

    “It might create regulatory capture for a few major players.”

    Listen at 19:45

  4. John Cooganon AI slowdown proposalsNegative19:49

    AI slowdown controls might crash markets or delay economic gains

    “It might crash the stock market or delay economic gains”

    Listen at 19:49

  5. John Cooganon AI slowdown regulationPositive21:37

    Universal AI slowdown regulation would not be especially harmful

    “I don't think it would be the worst thing for companies and consumers and businesses and all sorts of different folks.”

    Listen at 21:37

  6. John Cooganon AI 2040 robotics risk viewPositive22:12

    AI 2040 is not concerned about a billion aligned GPT-6-level robots.

    “I don't think they're worried about a billion robots with GPT-6 level intelligence and years of alignment work that is currently happening. That's fine.”

    Listen at 22:12

  7. John Cooganon Future superintelligent AI modelsNegative22:51

    The main AI risk concerns models with independent goals

    “It's the model with its own volition, basically”

    Listen at 22:51

  8. Paul Christiano has expressed concern about AI risk

    “Paul Cristiano, uh, who, uh, is, has been worried about risk.”

    Listen at 32:29

  9. John Cooganon Paul Cristiano investment positionNeutral32:43

    Paul Cristiano is two-times leveraged long.

    “Paul Cristiano is, uh, two X levered. long”

    Listen at 32:43

  10. Paul Cristiano reportedly allocates 5% to Tesla and 90% to AI bets.

    “he has 5% of his net worth in Tesla, 90% of his net worth in AI bets”

    Listen at 32:57

  11. Jordi Hayson Fruit-fly brain simulationPositive35:23

    The fruit-fly simulation can theoretically replicate fly decisions and movements

    “in theory you can replicate all of the fly's decisions or whatever, it's like movements.”

    Listen at 35:23

  12. Consciousness is physically defined

    “our consciousness is physically defined”

    Listen at 36:10

  13. John Cooganon human simulationPositive38:24

    A sufficiently advanced simulation could reproduce human behavior and conversation.

    “you can simulate a human and you could talk to that human and it would do everything the human does.”

    Listen at 38:24

  14. John Cooganon LLM sentience and moral statusMixed38:57

    The moral status and sentience of LLMs remain debated

    “the real debate here is like the question of, you know, are LLMs sentient? Are they moral? Is there a moral weight to synthetic intelligence, to artificial intelligence? That's what people are debating here.”

    Listen at 38:57

  15. Thijs Simonianon DIY roboticsPositive41:53

    DIY robotics is expected to expand soon

    “that world's definitely, like, just around the corner.”

    Listen at 41:53

  16. Thijs Simonianon OpenAI models and physical robotsPositive42:17

    OpenAI models can be connected to physical robots for drawing tasks.

    “you could actually basically connect these into physical robots in the real world and wanted to see how they would do at drawing and tasks like that as well.”

    Listen at 42:17

  17. Thijs Simonianon RoboticsPositive46:22

    Planning then monitoring provides a good robotics speed balance

    “I found that has been a good balance of speed here.”

    Listen at 46:22

  18. Thijs Simonianon RoboticsPositive47:02

    The model has largely solved perception challenges in this robotics task

    “the model has pretty much solved a lot of the perception challenges here.”

    Listen at 47:02

  19. Thijs Simonianon Hugging Face SO100Positive49:59

    The Hugging Face SO100 is open-source and fully 3D printable.

    “this is a Hugging Face SO100 robot, which is open source, fully 3D printable.”

    Listen at 49:59

  20. Thijs Simonianon Hugging Face SO100Positive50:11

    The Hugging Face SO100 costs approximately $200.

    “it's around $200 or so.”

    Listen at 50:11

  21. Thijs Simonianon consumer robot armsPositive50:27

    Consumers will buy inexpensive robot arms for personal use before industrial equipment dominates.

    “Everyone's going to buy these sort of cheap, before we get really industrial equipment, for personal use and personal product.”

    Listen at 50:27

  22. Thijs Simonianon Robotic knife useNegative52:50

    Giving a robot a knife is probably unsafe

    “giving a robot a knife is like probably a very bad idea.”

    Listen at 52:50

  23. Thijs Simonianon robot assistancePositive52:54

    Thijs wants to explore robots assisting with cooking and everyday tasks.

    “I do want to like sort of see if I can help do like cooking tasks or do various tasks that are like things that you do in your life that would be really sick if you could have a robot help you out here and there.”

    Listen at 52:54

  24. Thijs Simonianon Robot painting modelPositive54:40

    The model improved its painting through limited guidance

    “The model just was able to figure things out with a few pointers here and there, how to get better and better at painting.”

    Listen at 54:40

  25. Alex Heathon personal agentsPositive56:59

    The personal-agent sector is currently highly attractive for investment.

    “Personal agent space. Okay. Very interested in. It's very hot, obviously.”

    Listen at 56:59

  26. Guy Osearyon OpenAI and AnthropicPositive59:14

    Guy Oseary viewed OpenAI and Anthropic as enduring foundational platforms

    “these were the companies. These were going to be longstanding foundational platforms.”

    Listen at 59:14

  27. Guy Osearyon OpenAI and AnthropicPositive59:15

    Guy Oseary believed OpenAI and Anthropic would become longstanding foundational platforms.

    “These were going to be longstanding foundational platforms.”

    Listen at 59:15

  28. Guy Osearyon Anthropic and OpenAI investmentNeutral1:01:26

    Sound Ventures has nearly $1 billion invested in Anthropic and OpenAI.

    “we have close to a billion dollars worth of money invested into those two companies.”

    Listen at 1:01:26

  29. Alex Heathon founder evaluationPositive1:13:31

    Experience with top founders helps identify people who are pretending.

    “when you see the people at the Apex who are crushing it and who are at high integrity, are beasts at the game, on the field, you can quickly see when someone is pretending.”

    Listen at 1:13:31

  30. Alex Heathon Founder evaluationPositive1:13:38

    Experience with top founders helps identify people who are pretending

    “you can quickly see when someone is pretending.”

    Listen at 1:13:38

  31. Alex Heathon Magnificent Seven CEOs interviewsPositive1:15:40

    Alex Heath expects to interview the Magnificent Seven CEOs

    “The mag seven, I feel pretty good about. Yeah. You're on that track for sure.”

    Listen at 1:15:40

  32. Alex Heathon AI inferencePositive1:16:26

    Inference is currently highly important in the AI ecosystem.

    “inference is just so important right now.”

    Listen at 1:16:26

  33. Alex Heathon Technology journalism careersPositive1:22:41

    Aspiring technology journalists should focus intensely on one important niche

    “You have to maniacally focus on one thing that matters.”

    Listen at 1:22:41

  34. Alex Heathon journalism career strategyPositive1:22:41

    Aspiring journalists should focus intensely on a niche that matters.

    “You have to maniacally focus on one thing that matters. But it's the intersection of niche and matters.”

    Listen at 1:22:41

  35. Alex Heathon Instinct valuation versus SnapPositive1:23:00

    Alex Heath would not be surprised if Instinct exceeds Snap's valuation

    “I wouldn't be surprised.”

    Listen at 1:23:00

  36. John Cooganon Agentic CommercePositive1:25:12

    Agent commerce with network effects and transaction fees could be highly profitable

    “If you get to network effect and you get to some sort of take rate on agenda commerce, like you buy your car through it and they make 500 bucks, like that's a very, very good business.”

    Listen at 1:25:12

  37. Alex Heathon agent network effectsNeutral1:25:30

    Instinct, Town, and Meta are pursuing network effects for agents.

    “And Instinct's doing it, Town is doing it, Meta's going to do it.”

    Listen at 1:25:30

  38. Alex Heathon Personal-agent networksPositive1:25:36

    Agent networks may become the next major network effect

    “maybe that is the next network effect.”

    Listen at 1:25:36

  39. Mitesh Agrawalon AI inference memoryPositive1:29:23

    Memory will become increasingly important for AI inference

    “Memory is going to get a big part of the story for inference.”

    Listen at 1:29:23

  40. Mitesh Agrawalon Nvidia GPUsPositive1:30:21

    NVIDIA leads the market while releasing a new chip selection every 12 months.

    “new chip every 12 months. NVIDIA is the absolute king.”

    Listen at 1:30:21

  41. Mitesh Agrawalon Nvidia chip releasesNegative1:30:24

    Nvidia's annual chip releases pressure competitors to match its pace

    “if they're coming out with a new selection every 12 months, you better get in that game”

    Listen at 1:30:24

  42. Mitesh Agrawalon Positron first product shipmentPositive1:31:24

    Positron shipped its first customer product within 15 months.

    “we got our first shipment to a customer in 15 months.”

    Listen at 1:31:24

  43. Mitesh Agrawalon AI inference hardwarePositive1:33:48

    Inference hardware combining low cost and interactivity should attract demand

    “if you can do and or, you're going to bound to have get demand.”

    Listen at 1:33:48

  44. Mitesh Agrawalon AI-chip startups versus NVIDIANeutral1:36:05

    AI-chip startups should reach 10% of NVIDIA’s revenue before claiming to rival it.

    “no one is rivaling NVIDIA. Like, get to at least 10% of their revenue before putting the tagline on”

    Listen at 1:36:05

  45. New chip companies are not currently rivaling Nvidia

    “no one is rivaling NVIDIA.”

    Listen at 1:36:05

  46. Mitesh Agrawalon hyperscaler and frontier-lab AI deploymentsNeutral1:38:26

    Hyperscalers and frontier labs require AI hardware proposals exceeding one gigawatt.

    “for hyperscalers and Frontier Labs, honestly, they will literally say gigawatt plus”

    Listen at 1:38:26

  47. Mitesh Agrawalon AI infrastructure demandPositive1:38:30

    Hyperscalers and frontier labs require proposals exceeding one gigawatt

    “they will literally say gigawatt plus, come up to us with a proposal of a gigawatt plus”

    Listen at 1:38:30

  48. Mitesh Agrawalon Positron production scalePositive1:39:00

    Positron expects to plan for hundreds of megawatts by 2028

    “In 2028, we better have a plan of how do we get to, like, hundreds.”

    Listen at 1:39:00

  49. Mitesh Agrawalon Positron revenuePositive1:39:48

    Positron could grow revenue through alternative customers to $200 million, $250 million, $1 billion, or $2 billion.

    “you can grow the company, you know, get 200 million revenue for 250 million, a billion, 2 billion through this other kind of channels as well”

    Listen at 1:39:48

  50. Mitesh Agrawalon Positron first-generation systemPositive1:42:49

    Positron's first-generation system can run new model drops within hours

    “within our team, you know, on our ad, let's first-gen could get it up and running within hours”

    Listen at 1:42:49

  51. Mitesh Agrawalon AI model deployment automationPositive1:43:04

    AI automation will probably handle model-drop integration

    “Model drops, comes in, can probably do it.”

    Listen at 1:43:04

  52. Mitesh Agrawalon AI model deploymentPositive1:43:04

    AI agents will soon automate bringing new model releases into production.

    “Model drops, comes in, can probably do it. And that's a very near future of it.”

    Listen at 1:43:04

  53. Mitesh Agrawalon Positron software stackPositive1:43:55

    Positron will build its software stack around open source

    “The software stack is going to be built around open source.”

    Listen at 1:43:55

  54. John Cooganon PositronPositive1:44:32

    Positron raised $875 million.

    “You raised $875 million”

    Listen at 1:44:32

Statements are attributed to the speaker as said on the episode and reflect their view at the time, not PodLume's. They are not advice.

Weekly email

State of the Pod, every Monday

The week's most-discussed names and stories from podcasts, with the moments behind them. Free, five minutes, unsubscribe in one click.

Listen to the full episode and explore every guest, topic, and moment on PodLume.

AI risk meets robot arms, agent investing and chip competition · PodLume