Skip to main content
Ali Ghodsi

Ali Ghodsi

CEO of Databricks

Nov 17, 2026

45:47
It's like,
45:47
Yeah.
45:48
So now it's writing itself.
45:49
And the next version it wrote itself.
45:51
Wait a minute.
45:52
Isn't that the recursive self-improvement? But wait a minute, then this is going to speed up and then we're going to get into this loop.
45:56
The books, the book called it super intelligence and the paperclip maximizer.

9 MINS LATER

54:55
Yeah.
3:28
How much do you think a company like NVIDIA lives or dies based on a Jensen or these companies can rewire and find new charismatic leaders?
3:38
I think the essence of the personality of the CEO continues to live in the company for long to come.
3:43
So I think Jensen's legacy will continue to be there at Nvidia for decades to come.
3:47
So he built an epic company, and I think that's going to continue.
3:50
Why do I say this? Look, Bill Gates was Microsoft, right? It's like he was Microsoft.
3:55
You could not disentangle the two, and the two got disentangled.
3:59
But still, many of the competitive sort of aspects of how he thought about how they strategize, how they bundle things, all of those things kind of lived on for many, many years to come.

8 MINS LATER

12:09
Yeah, then suddenly it's
5:10
[laughs]
5:10
"... for sure we've hit that." But, you know, and then there's always some, you know, stupid but.
5:14
We move the goalpost, or we wanna change it, or we wanna have some other definition, or it- it's-- or the AI at some-- there's some example that, you know, it couldn't count the number of Rs in strawberry or something, so therefore we don't have AGI.
5:25
Um, we already have AGI, okay? It's already smarter than many of the people that you interact with.
5:30
That is general intelligence.
5:32
It is artificial.
5:32
It's not exactly a human.

8 MINS LATER

13:12
You guys.
24:33
Because I, I think most, or at least, you know-
24:34
Yeah.
24:35
And the reason they're collapsing is that it used to be like, okay, we have like data and AI, the kinda stuff Databricks and these kinda companies used to do, which is like, okay, you have a bunch of data and you run AI and machine learning, and that just lives separately.
24:45
And then you have the cyber world.
24:47
Cyber world is, you know, we wanna detect if something bad, if, like, bad people are trying to hack us, if bad people are doing things, we need to detect that.
24:53
Okay? But now on the data and AI side, we have agents running internally in the company, people are having agents running, and the agents are also, like, doing things with other people's agents, and they're producing a lot of data, uh, logs, trails, you know, fingerprints that are being left.
25:07
Uh, and so, you know, then now we have internally these agents that are doing that.

23 MINS LATER

47:56
Hmm.
13:29
the use cases.
13:29
Hap- happy to do that.
13:30
I mean, look, let's start, since AI is sort of front page news today and, uh, people are talking about pacing it and so on, should do a benefit-cost ratio.
13:38
There's a lot of benefits.
13:39
So for instance, there's a company called Crisis Text Line.
13:42
They use us, they use AI, and they analyze, uh, you know, how, uh, you know, uh, teenagers are using AI and what they're, uh, typing and, you know, if they're a- about to do self-harm or suicide, it helps prevent that.
13:55
You know, and it-
15:34
Right.
17:59
Are you still doing that? And, and how does that apply in this AI safety context?
18:03
context?Yeah, I've always been one of these, you know, really, really paranoid CEOs.
18:07
We've been doing AI, you know, since the first days, 2013.
18:10
Our first use case was AI use case.
18:12
So yeah, we do those every year, and we wanna make sure that we're protected.
18:15
So as I said, I think cyber is real, so that is actually part of our exercises.
18:19
You know, what happens if someone gets access to the computer systems? Like, we have very, very sensitive data at Databricks, so we actually do a lot of these s- simulations of skies falling also on the cyber side, and we let basically attackers from outside try to break in and get all the way into the most, you know, uh, sacred secrets of Databricks, uh, and keep hardening those.
19:50
Does that factor into your calculus?
16:08
Very
16:08
aggressive.
16:10
Yeah.
16:11
And one of the first ADs that Ron had hired, Dave, got us a meeting with a customer that we couldn't get a meeting with.
16:19
And in the meeting, the exec that had gotten us the meeting said, that AE...
16:24
Get him out of here.
16:25
I don't want to see that guy ever again.

11 MINS LATER

27:42
Any thoughts on that in advising founders and CEOs about that kind of thing?
John Fortt
John ForttCORRESPONDENT
31:10
Tell me about what that says about agentic demand right now and why you think it's still not okay to jump into the IPO waters with that kind of an environment.
31:22
Yeah, no, I mean, the demand is crazy.
31:25
What's happening basically is everybody's using these agents, AI agents, and the whole world is laser focused on agents, AI, and sort of, you know, that core part of it.
31:36
But what they fail to look at is that that drives a lot of other things that are around it, that are adjacent to it.
31:42
So if you're using agents to write software, well, all the software needs a database.
31:47
So that's that lake-based database that we launched.
31:49
And we only launched it last year.
John Fortt
John ForttCORRESPONDENT
33:16
Because if models are open for a lot of work, they've got to be less expensive, right? Especially if you're adding context to them that makes them more effective.
Kate Rooney
Kate RooneyCORRESPONDENT
32:30
What do you make of the capabilities of Kimi, and i- how much of a threat is that to the frontier labs as we now look at almost trillion dollar valuations?
32:39
Well, I mean, look, I think the frontier labs are amazing, and they're gonna continue doing great.
32:43
And, uh, but what happened basically is that everybody wants to do, uh, cost controls, and they wanna control their budgets because as one executive of a Fortune 500 told me, uh, his costs of AI are increasing exponentially.
32:58
His revenue is not increasing exponentially.
33:00
So if this com- continues, he will go bankrupt, so he has to do something.
33:05
So then open source is how they wanna ...
33:08
You know, for the really difficult questions, let's ask the frontier American models and pay a lot for it.
Kate Rooney
Kate RooneyCORRESPONDENT
34:37
Mm
22:15
Here's what CEO Ali Ghodsi told us yesterday.
22:19
So, you know, we're not, we've not said that we're doing any fundraise.
22:22
I know that there's, you know, uh, rumors about this.
22:25
Uh, but we're always talking to investors.
22:27
Why? Well, we're investing heavily into these new product categories.
22:30
We just launching, uh, a completely new product for our s- that's called Customer Lake, which is in the area of marketing.
22:38
So we're, you know, targeting, uh, marketing folks, which is not an area where we used to, uh, be active.
23:32
Do you stand by that? Uh, has anything changed in your mind in the last two weeks?
14:05
You probably are in a place where you can evaluate it, Ali, aren't you?
14:08
Yeah.
14:09
So I think Jensen is spot on with both of those.
14:11
Uh, you know, I, I ...
14:13
We actually did the calculation.
14:14
We think in the next nine months, but let's say around one year or two years, we're gonna have more software being written than in all history of mankind.
14:23
You know, you ...
16:21
Yeah.
29:15
Isn't it too costly to go from Snowflake to Databricks?
29:20
Well, uh, I think that's a great company.
29:22
Um, you know, just the two companies have different backgrounds.
29:25
Uh, we grew up in AI.
29:26
So we started doing AI, uh, you know, in 2009, frankly.
29:30
And then when the company started 2013.
29:32
Uh, we're at this point synonymous with the data and AI company.

8 MINS LATER

37:41
Isn't it too costly to go from Snowflake to Databricks?
Deirdre BosaCORRESPONDENT
32:49
What's a SIEM?
32:50
A SIEM is a security information event management system that does all of your detections in your organization.
32:56
It detects if something is wrong, if something is suspicious, and all of that.
32:59
Uh, this is the year where it's gonna be disrupted.
33:02
You know, these old legacy systems are very expensive.
33:04
Getting data into them is super hard.
33:06
People are not even looking at all the data.
Deirdre BosaCORRESPONDENT
34:41
What does that mean? What's next?
22:01
You do? Yeah.
22:01
We do.
22:02
Um, but those things, they focus on how can we write, how can we help software engineers? What Genie Code really can do is it brings it to the knowledge worker, the people that create your dashboards inside of an organization, and they make sure that your revenue numbers are correct, or the people that make sure that the data that's coming in every day into the organization is correct, nothing breaks.
22:20
You know, there's not an outage.
22:21
You don't have like a blue screen, the dashboard is out, and you can't see it.
22:24
Or the people that are building machine learning models that can predict your prices or your costs or, you know, doing risk as-assessment with machine learning models.
22:32
It automates that portion, so it's sort of very complementary to cloud code and, you know, Cursor.
22:50
So how are you, how are you seeing the adoption of AI agents? What does it mean actually for the f- the strength of your data business?
Deirdre BosaCORRESPONDENT
41:02
And if what you're saying is that it's easier for non-technical people to make their own databases or the AI itself to make the databases, where does that leave this trade?
41:13
Yeah, I mean, markets are funny.
41:14
First, they ignore this problem for two, three years.
41:16
Now they wake up and they think it's going to happen in a week.
41:19
But the reality is that, look, this moat of it's very hard to move databases or it's very hard to use a different interface, teach humans to learn a new interface, that moat has decreased and almost is evaporating because humans can now just talk to an AI instead of learning a new interface.
41:39
Does that mean that everyone immediately is going to switch away? You know, it's going to take some while.
41:44
For instance, for databases, you need a really compelling reason to change database.
Deirdre BosaCORRESPONDENT
42:36
SAP is one that you bring up, but what about some of the more vulnerable software companies? What kind of companies are those, and when do you think they're going to be disrupted if they're not already being? I mean, look at Monday.com today, down some 20%.
5:28
And if what you're saying is that it's easier for non-technical people to make their own databases or the AI itself to make the databases, where does that leave this trade?
5:39
Yeah, I mean, markets are funny.
5:40
First, they ignore this problem for two, three years.
5:42
Now they wake up and they think it's going to happen in a week.
5:45
But the reality is that, look, this moat of it's very hard to move databases or it's very hard to use a different interface, teach humans to learn a new interface, that moat has decreased and almost is evaporating because humans can now just talk to an AI instead of learning a new interface.
6:04
Does that mean that everyone immediately is going to switch away? You know, it's going to take some while.
6:10
For instance, for databases, you need a really compelling reason to change database.
7:02
SAP is one that you bring up, but what about some of the more vulnerable software companies? What kind of companies are those, and when do you think they're going to be disrupted if they're not already being? I mean, look at Monday.com today, down some 20%.
18:24
Mm-hmm.
18:24
...
18:24
the quest for super intelligence, which is kind of intelligence that's like an almost, like godlike, it leads to recursive self-improvement of the AI, which then once you have that, it can cure cancer and solve all economical problems.
18:37
And we can probably 10X GDP over a few years period of time.
18:41
So what the hell are you talking about, that there's a physics problem? Like, it'll, anything, any of your cost equations are gonna pale in comparison to the economic value that this thing is gonna provide.
18:51
So that's like one camp.
18:52
And the way they're developing it is bigger and bigger clusters, more and more energy, and that's how they're going about it.
21:20
Mm.
175:34
Mm-hmm.
175:35
Number three, you need to put together amazing, stellar world-class team, right? That's all that matters 'cause if you're amazing, but your team is not, then you're not gonna go anywhere.
175:44
Even if you have all of these, it's not enough.
175:46
You need to also execute.
175:47
Like, you need to actually get the team to execute again and again and again and get to the results.
175:50
And that kinda matters more than anything else.
175:52
I mean, look at AWS, like, they're phenomenal at executing, right? And then, uh, finally, five, you need to do good culture.

6 MINS LATER

speaker_7UNKNOWN
181:33
Yeah.
18:23
Talk to me about your cash management strategy, how that factors into your raise here and your vision for the future.
18:29
Yeah.
18:29
So, you know, we've said that over the last 12 months, we've been free cash flow positive.
18:34
So, you know, we're efficient business, but we're reinvesting it.
18:38
We're not, you know, we're not trying to produce lots and lots of free cash flow.
18:41
We just want to stay at this break even and invest it back in agents, you know, in databases, lake base.
18:47
There's this new category of applications out there that are being built.
19:14
How much does that factor into your choice of whether you go public in 2026?
3:01
Where's the company now and how does the funding round help you move forward?
3:04
Yeah, there's really two places that we're seeing incredible demand, and that's where we're going to really focus the capital and the investments, both on go-to-market and R&D.
3:14
One of them is a product that's called Agent Bricks.
3:17
which we launched not long ago, two, three months ago.
3:21
It's in beta.
3:21
We only have it in two out of our 160 regions.
3:24
The pricing is not finalized.

13 MINS LATER

16:30
moving so quickly
31:29
And are they getting the most out of it? Because I got to believe that they don't really have someone like you at their own places to tell them.
31:37
Yeah, it's very interesting that you actually ask.
31:39
So I think there's been a lot of focus on super intelligence and building the smartest AI possible and asking it the most hardest AI math questions, programming questions.
31:51
But what we really need is for the AI to do mundane tasks, kind of boring AI.
31:55
Right, thank you.
31:57
extract this field from ServiceNow and enter it into Salesforce and do some basic summarization on it.
32:03
So we have really been focused on how do we really make sure that the agents can reason on these simple, mundane tasks.
32:54
Yes, you actually may be better at your job.
5:10
Talk about what it was like for you and in the context of what others can learn from it.
5:14
So we were in academia, so we were scientists.
5:16
And then, you know, I led engineering and product.
5:18
So you got to learn how to build a product and get product market fit.
5:21
And then I became CEO afterwards.
5:22
Each of these has different sort of challenges.
5:25
I think that in all of them, the thing that is in common is that you really have to understand and be extremely good at the task at hand.

28 MINS LATER

33:38
What is sort of your approach in terms of when to build versus when to buy? How do you think about acquisitions more broadly?
Mackenzie Sigalos
Mackenzie SigalosCORRESPONDENT
5:30
But when are we actually going to start to be able to measure a return on invested capital for these companies?
5:36
Yeah, it's happening slowly.
5:37
I think that, you know, people got too excited maybe in the last two years and it got kind of, you know, people were following the investments that were happening in the infrastructure and so on.
5:45
They were a little bit ahead of themselves, but actually we're now seeing the use cases land.
5:49
We're seeing them being automated.
5:50
You know, MasterCard actually has been using us and OpenAI for all of their product onboarding for all of their cards.
5:58
Adidas has been using us and using agents that actually can now measure the sentiment around the world for all of their shoes, and they can then improve those shoes and the designs in a way they couldn't do before with humans.
Mackenzie Sigalos
Mackenzie SigalosCORRESPONDENT
6:48
Do you think that this is a healthy build out or more of a concerning spending spree?
MacKenzie Sigalos
MacKenzie SigalosCORRESPONDENT
42:33
But when are we actually going to start to be able to measure a return on invested capital for these companies?
42:39
Yeah, it's happening slowly.
42:41
I think that, you know, people got too excited maybe in the last two years and it got kind of, you know, people were following the investments that were happening in the infrastructure and so on.
42:49
They were a little bit ahead of themselves, but actually we're now seeing the use cases land.
42:52
We're seeing them being automated.
42:54
You know, MasterCard actually has been using us and OpenAI for all of their product onboarding for all of their cards.
43:01
Adidas has been using us and using agents that actually can now measure the sentiment around the world for all of their shoes, and they can then improve those shoes and the designs in a way they couldn't do before with humans.
MacKenzie Sigalos
MacKenzie SigalosCORRESPONDENT
43:52
Do you think that this is a healthy build-out or more of a concerning spending spree?
14:28
What's driving it? Can it continue?
14:31
Yeah, so we're seeing lots and lots of momentum behind two of our products.
14:34
The first one is called Agent Bricks, and that product really helps you build agents that are helping out in the workplace to automate sort of work that you have to do everyday tasks.
14:46
And what's different here is a lot of the AI labs have been really focused on super intelligence, you know, math Olympiads and programming contests.
14:55
We're more focused on just everyday tasks that you need to do.
14:59
And we're seeing a huge uptick in the adoption of this.
15:01
So it's actually accelerating our revenue.
18:12
Can you give an example with LakeBase of how the adoption cycle with large customers plays out, how you expect to gain share?

Scroll for more appearances

We value your privacy

We use cookies to understand how you use our platform and to improve your experience. Click “Accept All” to consent, or “Decline non-essential” to opt out of non-essential cookies. Read our Privacy Policy.