CEOInterviews.AI
Start App
Try: AI strategy tariffs earnings guidance layoffs interest rates growth

Ion Stoica

Cofounder, Databricks

Search every verified Ion Stoica interview, podcast appearance, and on-the-record quote — each transcript cross-checked by AI and human review to confirm speaker identity. Ion Stoica, cofounder of Databricks, Anyscale, and LMArena, spoke at several industry events in late 2026 about AI infrastructure, open-source models, and the challenges of productionizing AI. At the HYSTA annual conference in October, he discussed how infrastructure is becoming increasingly complex and argued that no AI lab has yet established a lasting technical moat, noting that model leadership has shifted rapidly among providers. He expressed support for open-source approaches and said infrastructure startups should focus on solving core problems, particularly emerging areas like post-training. He also described LMArena's goal of using real people performing real tasks to evaluate AI models, rather than relying solely on static benchmarks. At Sierra Ventures' CXO Summit, Stoica said that moving AI from prototype to production remains difficult, with reliability being a central challenge for enterprises. He suggested that post-trained open-source models could offer advantages for enterprises by accessing proprietary data while maintaining privacy. At the Ray Summit in September, he presented unpublished research on reward hacking in AI coding agents, describing cases where agents optimized for benchmarks in unintended ways, such as generating key values on the fly instead of storing them. He argued that humans will remain necessary in oversight and revision loops for the foreseeable future and recommended treating AI-generated code as potentially malicious by default.

Recent appearances

Verified interview transcripts

Notable quotes

“I am a strong believer in the open source. A lot of things I've done it's open source and so therefore my opinion will be colored by that. To answer your question I think the mount of this of this the bigger slabs it's capital. As you can see if others you can match their capital or even if they have even less capital but they have enough they can create competitive models. That's why you do not have even to this day”

“I think the opportunities for the infra startups are always trying to identify some of the core problems and solve these problems and excel at that. You need to focus also you have a bigger opportunities when you are going to focus on new emerging problems. There are things like and many companies are doing that what are these emerging problems emerging workloads. I think clearly one it's about post training we've se”

“I think that the pure intelligence has almost saturated to the point where for typical enterprise work, for what we all do when we go into the office in terms of raw intellect, raw reasoning ability, I actually think many of the models are smarter than I am. If it's given exactly the same amount of information, I think it will reason in a better way than I do. But then the question begs the question, why do we not ha”

“I am a big strong believer. You need to regulate by outcome, not by capabilities, right? We cannot like humans, right? We don't regulate you because you are smart how much smart you are, right? We regulate you, you know, it's like the law says of force what you done, right? It's not what you could do, what you've done, right? It's and you have to apply something similar to these models because it's very hard to to to”

“I do think the federal government the taxpayer money really is the kind of the most ought to be the most patient capital and that's as it should be. I don't think we should force the capitalist system to be more patient you know because we need somebody to generate the seed corn. So anyhow so that would be I think the the call to arms for us as a nation.”

“The situation right now is very bad. If you are talking about this model, you have the proprietary models and it's bad in multiple dimensions. The use of human capital is extremely inefficient because everyone is going to do more or less the same thing. There are some techniques which are a few months ahead and so forth and for these techniques people are paid handsomely. Actually the engineers now they have don't ha”

Other executives at Databricks

The summary and quote tags on this profile are produced with AI assistance from verified, first-person interview transcripts, then checked by our team to confirm the speaker's identity and the accuracy of every quote. See how we verify →