# Nach LLMs kommt das World Model: 1,1 Mrd. $ Pre-Seed für Ineffable Intelligence – Olaf Jacobi (Basic Ventures)

Page: https://stenobird.com/podcast/startup-insider-6744721/nach-llms-kommt-das-world-model-1-1-mrd-pre-seed-f-r-ineffable-intelligence-olaf-jacobi-basic-ventures
Text version: https://stenobird.com/podcast/startup-insider-6744721/nach-llms-kommt-das-world-model-1-1-mrd-pre-seed-f-r-ineffable-intelligence-olaf-jacobi-basic-ventures.md
Podcast: [Startup Insider](https://stenobird.com/podcast/startup-insider-6744721)
Published: 2026-05-05T02:00:00+00:00
Episode link: https://startup-insider.com/alle-podcast-folgen/
Audio file: https://afp-912057-injected.calisto.simplecastaudio.com/d2790bdf-c937-4878-8f59-941f6b6b4ef5/episodes/a21820a4-9ebf-48f9-9e8f-c3b108f74281/audio/128/default.mp3?aid=rss_feed&awCollectionId=d2790bdf-c937-4878-8f59-941f6b6b4ef5&awEpisodeId=a21820a4-9ebf-48f9-9e8f-c3b108f74281&feed=ZQdsoEnZ
Processing state: processed
JSON: https://stenobird.com/v1/public/podcasts/startup-insider-6744721/episodes/nach-llms-kommt-das-world-model-1-1-mrd-pre-seed-f-r-ineffable-intelligence-olaf-jacobi-basic-ventures
Duration seconds: 2669

## Resource

The era of Large Language Models may be hitting a data wall, as they are limited to distilling existing human knowledge. This discussion explores the massive $1.1 billion pre-seed investment in Ineffable Intelligence, a venture aiming to move beyond text prediction toward true 'World Models' that can discover new scientific truths.

## Highlights
- Main idea: LLMs are fundamentally limited by the 'data wall' because they only aggregate existing human knowledge without the capacity for true new discovery
- Main idea: The next frontier in AI is the development of 'World Models' that understand physical and logical reality rather than just statistical word patterns
- Failure mode: Relying solely on human-generated datasets prevents AI from achieving superintelligence or making breakthroughs in mathematics and science
- Practical takeaway: The massive $1.1 billion pre-seed valuation for Ineffable Intelligence signals a massive shift in capital toward deep-tech foundational models
- Main idea: True intelligence requires moving from reinforcement learning agents, like AlphaGo, to systems capable of independent reasoning and discovery

## Topics

Artificial Intelligence, World Models, Large Language Models, Venture Capital, Deep Tech, Reinforcement Learning, Ineffable Intelligence, Superintelligence

## Chapters
- 1:00 — Introduction to Ineffable Intelligence: An introduction to Olaf Jacobi and the unprecedented $1.1 billion pre-seed funding round.
- 8:10 — The Evolution of AI Architectures: Discussing how new architectures like Cloud Design could render current models as mere features.
- 11:40 — The Economic Impact of World Models: Analyzing the trillion-dollar potential of mastering the next step in AI development.
- 15:10 — Lessons from AlphaGo: Reflecting on how Reinforcement Learning enabled AlphaGo to master complex systems through self-play.
- 22:20 — The Competitive Landscape: Comparing recent massive funding rounds in the AI sector, including Amy Labs.
- 25:50 — Breaking the Human Data Dependency: The debate over whether AI can truly innovate if it is trained exclusively on human-generated text.
- 33:00 — The Future of Autonomous Agents: Speculating on the role of World Models in controlling humanoid robotics and scientific discovery.

## Actions

- request_transcript: `POST https://stenobird.com/v1/public/podcasts/startup-insider-6744721/episodes/nach-llms-kommt-das-world-model-1-1-mrd-pre-seed-f-r-ineffable-intelligence-olaf-jacobi-basic-ventures/transcription-requests` — Idempotently request low-priority transcript generation for this episode.
- read_markdown: `GET https://stenobird.com/podcast/startup-insider-6744721/nach-llms-kommt-das-world-model-1-1-mrd-pre-seed-f-r-ineffable-intelligence-olaf-jacobi-basic-ventures.md` — Read the agent-friendly Markdown representation of this episode resource.

A page view does not enqueue transcription. Agents should invoke `request_transcript` explicitly when they need this episode processed.

## Transcript

Full transcripts are not published on public pages unless there is a clear rights basis.