CNBC's Jim Cramer speaks to Ali Ghodsi, co-founder and CEO of Databricks, to break down the company's fundraising and efforts to improve artificial intelligence.
What Ali Ghodsi said
Written from the verified transcript and checked against it. Every figure links to the moment it was said.
Ali Ghodsi, CEO of Databricks, said the company is in a strong position because it burns no capital, unlike competitors that need public markets to fund massive losses. He stated Databricks will go public but not this year, as it is better to build during the AI transition. Ghodsi argued that AI's main challenge is not intelligence but context, and that enterprises struggle to feed agents the right data. He highlighted Genie Ontology, a graph-based system that automatically structures data in the Databricks Lakehouse, enabling accurate, real-time answers. He contrasted Databricks' AI-first roots with Snowflake's data warehousing background, and emphasized the Lakehouse as the system of record for agents. He noted most Fortune 500 and Global 2000 companies are already in the cloud, and cited Prada as a customer that moved from Excel to Databricks for AI-driven insights.
Key takeaways
Databricks will go public, but not this year, because it burns zero capital and prefers to build during the AI transition.
Genie Ontology automatically builds a graph of data nuggets, enabling accurate, real-time answers without costly data looping.
The Lakehouse has become the system of record for agents, which Ghodsi calls SORA.
Prada moved from Excel to Databricks to ask real-time AI questions about sales and inventory.
3:16All right, so how about if I had Genie, what would I do with it?
4:43Isn't it too costly to go from Snowflake to Databricks?
6:01... have all their data on-prem and they can't prosecute it using Databricks, shouldn't I make a presumption that perhaps they are old school and not availing themselves of things that would make them more efficient and better at what they do?
6:49But isn't that the way it used to be before Databricks?
If you want to get a read on AI, it's not enough to follow publicly traded companies, because many of the most important players in this space are still private. Take Databricks, which helps companies take messy, fragmented data, organize it, analyze it, build AI on top of it, and frankly, figure out what you have. As of February, this company had $5.4 billion annual revenue run rate, and by now it's got to be substantially higher than that, as these guys are growing like a weed. Databricks is free cash flow positive. It's bringing in tons of business. There's a reason this company got $134 billion valuation in its last private fundraising round. All that helped send Databricks to the number three spot on this year's CNBC Disruptor 50 list. Now that AI has entered the agentic era, this one's become a heavy hitter. So let's take a closer look with Ali Ghodsi. He's the co-founder and CEO of Databricks. To learn more, Mr. Ghodsi, welcome back to Mad Money.
Hey, it's great to see you. First of all, I got to go through this rigmarole. I'm sorry, but, you know, we got all these people, you know. When is Anthropic going to come? OpenAI, is it filed at SpaceX? A lot of those companies need money. When I look at your company, I say to myself, here's someone who's in the driver's seat who can come public, not come public. Is that not a better way to look at Databricks than the other guys, where we play the guessing game all the time?
Yeah. I mean, the way you think about it is that if you're burning huge amounts of capital like ten, 20, 30, $40 billion, you need the capital, you need to be in the public markets. You have no choice. For Databricks, we don't burn any money at all, $0. So for us, we can choose the timing. I think this is a good year to be private, because I think this is a good year where you want to be building, because we're going through a huge transition in the market, in public, in private, all enterprises. This AI revolution is happening. So it's better to build in private right now. And we will be public and we will go public. But I don't think this year is a good time for us.
All right. So when I first met you, I said, well, wait a second. I see what's going on here. You can take the data, but it's your data. You can prosecute it. You can ask the question, you can inquire things. It seems like ever since AI, now there's like 100 things you could do if there were five things before. Do you advise your clients, look, you can do this now. And do they avail themselves of it?
Yeah. Look, I think the thing that is top of mind right now for every CEO on the planet, not just in tech, any CEO is, wow, these AIs are so amazing. They can do so amazing things, but secretly, and they don't want to admit this, my organization has none of the AI. It's not established. We don't have agentic coworkers doing lots and lots of work. How do I get AI into my organization? And they have plenty of intelligence in the AIs. But why are the agents not working in the enterprise? What's going on? And the simple answer to that is the AI is, if you give them all the data and the context, if you perfectly ask the question with all the context, they will nail the answer every time. So the big question is, AI doesn't have an intelligence problem. AI has a context problem. How do we feed it that context? And that context is in the data. So that's top of mind. Everyone wants to solve this problem. That no one on the planet right now doesn't want to have this problem nailed.
Ghodsi, A. (2026, May 21). Databricks CEO Ali Ghodsi: AI doesn't have an intelligence problem. It has a context problem [Interview transcript]. CNBC Television. CEOInterviews.AI. https://ceointerviews.ai/interview/923753/
MLA
Ali Ghodsi. "Databricks CEO Ali Ghodsi: AI doesn't have an intelligence problem. It has a context problem." CNBC Television, 21 May. 2026. Transcript, CEOInterviews.AI, https://ceointerviews.ai/interview/923753/.
BibTeX
@misc{ghodsi2026_923753,
author = {Ali Ghodsi},
title = {Databricks CEO Ali Ghodsi: AI doesn't have an intelligence problem. It has a context problem},
howpublished = {Interview transcript, CNBC Television. CEOInterviews.AI},
year = {2026},
month = {may},
url = {https://ceointerviews.ai/interview/923753/},
note = {Speaker-attributed transcript with timestamps}
}
