Why this CEO thinks video games make better training data than the internet
Get the Tech newsletter
Daily tech — startups, AI labs, chips, the launches that shape the next decade. Free.
- General Intuition closed a $320 million funding round at a $2.3 billion valuation, betting that gaming data can train AI models that understand spatial and temporal movement — a gap the CEO says LLMs cannot close.
- The Bezos-backed New York startup drew Coatue, Eric Schmidt, and researchers at MIT and Google DeepMind as investors in the round.
- General Intuition spun out of Medal TV, a gaming platform, giving it a built-in pipeline of gameplay footage for model training.
- Pim de Witte, General Intuition's CEO, argued on TechCrunch's Equity podcast that models like ChatGPT and Claude are 'great at text' but lack the physical reasoning needed for AGI.
- The company is training 'world models' on gaming footage for what it calls 'physical AI,' positioning the work as the next leap beyond text-based LLMs.
- De Witte flagged ethical red lines around potential defense applications of the company's models during the Equity discussion.
Why it matters: General Intuition's $320M round — with Coatue, Eric Schmidt, MIT, and Google DeepMind on the cap table — puts institutional weight behind CEO Pim de Witte's argument that LLMs structurally lack spatial-temporal understanding. For text-first AI labs, it's an explicit challenge to the assumption that more internet text alone gets to AGI.
