Handelsblatt Today – Der Finanzpodcast mit News zu Börse, Aktien und Geldanlage · Solveig Gode, Nicole Bastian, Sandra Groeneveld, Nele Dohmen, Anis Mičijević, Ben Mendelson

Sonderfolge: Der COO des ChatGPT-Entwicklers OpenAI im Interview

·25 min·2 clips

OpenAI's COO Brad Leitkamp debunks the AI bubble myth and reveals revenue numbers in a Davos interview.

1. Handelsblatt Today presents a Davos interview by editor-in-chief Sebastian Mattes with Brad Leitkamp, COO of OpenAI, recorded at the World Economic Forum. 2. Leitkamp is the Chief Operating Officer of OpenAI, the company behind ChatGPT, and has been at the organization for nearly eight years. 3. The interview's central thesis is that AI progress has shifted from hype to an infrastructure problem, and the bottleneck is now organizations adapting to models that are already more capable than their systems. 4. Leitkamp says the biggest misunderstanding about AI is treating it as a 'one-and-done phenomenon,' when it is a continuously accelerating technology still in its very early stages. 5. On the enterprise productivity gap — where pilots look good but actual gains disappoint — Leitkamp introduces the concept of a 'capability overhang': models are already more capable than the workflows they are inserted into. 6. He says enterprises need to rethink their entire organizational structures and IT stacks to ingest AI systems that are performing fundamentally new work, not just automating existing tasks. 7. T-Mobile US is named as the lead success case: OpenAI worked closely with T-Mobile to redesign customer support interactions from first principles, bringing language models directly into contact center workflows and plans to expand to voice. 8. Germany is specifically cited as a top-three country globally by ChatGPT users, number of developers, and number of businesses — a detail Leitkamp says he finds 'staggering.' 9. On OpenAI's new device (referred to only as a hardware product), Leitkamp says the device category is still confidential but describes the underlying idea: every major platform shift produces new interaction models, and AI's natural-language paradigm likely needs hardware built around it. 10. His own newborn child and their peers, who instinctively talk to tablets expecting a response, are used as evidence of how interaction paradigms shift generationally. 11. ChatGPT Health is in an alpha launch: users can link medical records, receive lab result interpretation, and be directed to qualified health information sources, with the platform explicitly not providing medical advice. 12. Education is described as one of OpenAI's largest enterprise segments, with adoption by students being one of the earliest organic use cases. 13. On advertising, Leitkamp acknowledges Sam Altman's October 2024 statement calling ads a 'last resort,' but says OpenAI is now experimenting with ads while committing that paid subscribers and enterprise customers will never see them. 14. He frames ads as potentially additive to the experience because ChatGPT is already used for product searches, similar to how commerce was added to the platform. 15. On AGI, Leitkamp defines it as systems able to 'do most economically valuable work' — specifically, an AI that can use a computer to do anything a human could do at a high level of proficiency. 16. He declines to predict when AGI will arrive but notes that in his experience at OpenAI, things that seem two to four years away often arrive in six to eighteen months. 17. Software engineering is named as the concrete profession already transformed: Codex agents can autonomously spend five to six hours on a complex task — writing, testing, debugging — requiring only codebase access. 18. On the AI bubble question, Leitkamp states that every dollar invested in compute has consistently returned $3 in revenue, and user growth has been 'geometric.' 19. Technology investors, software engineers, and enterprise IT leaders evaluating AI deployment strategy will find the most actionable content here. 20. Listeners seeking technical AI research discussion, model comparisons, or policy analysis will find the conversation stays at a business strategy and market positioning level.

Listen to the show on

Apple Podcasts Spotify

Play clips from this episode More from Handelsblatt Today – Der Finanzpodcast mit News zu Börse, Aktien und Geldanlage