State of the Data & Analytics Market – Q3-2026

00:00:00: We're seeing a lot of work being done on vendors looking to acquire these governance tools that we are talking about.

00:00:09: We could talk about specifics, but Park Research says the adoption agents is foreseeing vendor community in good way.

00:00:27: Hi Karsten!

00:00:28: Welcome our quarterly update podcast.

00:00:31: Hope you well?

00:00:32: Hi Sean and I hope so too.

00:00:35: Yeah, thank you.

00:00:35: We have a lot to cover.

00:00:38: as always on this program we're gonna talk about trends that we see in the user community for end users and Obviously talked about some of the interesting trends coming out at the vendor space And of course as always investments an M&A and things of that nature.

00:00:55: so let's dive into topic number one And it won't surprise anybody who's listening today.

00:01:02: AI is likely to be dominant in our conversation, and I thought it'd be great for us to start around all the governance things that we're seeing.

00:01:13: what AI agents are causing end users to be concerned about?

00:01:18: How trust is playing a really big role as people are rolling out more utilization of autonomous agents?

00:01:26: What your perspective on this?

00:01:28: For me, that's one of the hottest topics customers we talk to.

00:01:35: Many figured out their general AI governance sometimes driven by compliance.

00:01:41: topics like EU-AI Act sometime.

00:01:45: just basically get there hands around what is happening with AI in companies and now agents come into play And that changes a lot because here you need to think about new things.

00:02:01: You have, for example, lots of interactions and connections with security during let's say data and AI governance in general.

00:02:11: obviously it is an important topic but its just one many topics if we look at agents this where suddenly the importance rises, that you need to be really aware what these agents can do in your network.

00:02:28: In your applications even maybe going outside of your own network.

00:02:34: we saw more scary examples of open AI and entropic.

00:02:40: so something new there is no Blueprint, there is no template typically.

00:02:47: So we right now talk to many companies that really have to figure this out from scratch and That's where?

00:02:55: We have a lot of interesting interactions Where we can help or they can support them while even organize meetings between user companies yet too for To let them exchange their experiences in their approaches.

00:03:08: so a lot going on I think super interesting.

00:03:11: Yeah it adds The autonomous side adds this multiplier that I don't think most end users are ready for.

00:03:20: Even us as analysts, we give it a lot of thought...I was building an agent recently and wanted access to some of our enterprise

00:03:27: systems.".

00:03:28: No!

00:03:29: Not not for this.

00:03:33: You'd think these large companies have thousands people who've been given subscriptions uh, anthropic products and open AI products.

00:03:42: And they're out there building things in the wild.

00:03:44: I call it Agenix spaghetti.

00:03:46: um because pretty quickly you just really have a hard time understanding what are all these agents doing?

00:03:53: What were the goals they were given and Are we able to monitor not only their performance but as You said from a compliance of regulatory standpoint?

00:04:01: that's interesting As well.

00:04:03: i will talk about It next Time when We meet.

00:04:05: But also as You know The EU AI Act has kind of forced the big model companies to start doing watermarking, too.

00:04:14: People are transparent about how AI is being utilized inside of companies and as that matures I'm sure we'll talk about it next quarter.

00:04:23: but i think this trend happening our users feeling a pressure from Pretty interesting.

00:04:33: and I think that when we talk about vendors in a little bit We'll talk a little.

00:04:36: but what were seen there as well, but it's uh It's an absolute topic.

00:04:40: And for those of you listening today If you're not already thinking about how to govern the agent environment That your building?

00:04:49: We believe that it is not a nice-to-have.

00:04:51: It's a must have if you want to be successful going forward.

00:04:55: Carson You agree with the criticality of them.

00:04:58: Absolutely.

00:04:59: I mean, AI is happening.

00:05:02: AI is maturing and evolving.

00:05:04: especially agents are the big next steps for many companies.

00:05:09: we see the maturity level now.

00:05:11: it's rolling out its scaling.

00:05:13: It's not going out of a pilot phase into real production.

00:05:18: And that's where the problem start and whether governance now really has to work.

00:05:24: Yeah i agree.

00:05:25: let's talk about another maturity concept where your data is.

00:05:31: This idea of AI and data sovereignty, uh...is growing.

00:05:36: you've led some really great research that Park has done on this subject.

00:05:41: I like to think that park as quite an expert on the topic.

00:05:44: what's here take-on?

00:05:46: The evolution especially from the European side of the world?

00:05:50: But i also know That You found Some interesting Data Nuggets And What Those Of Us in the US Are Worryed About too.

00:05:55: so what are you seeing On This Topic?

00:05:58: Yeah, so I think you already touched on it.

00:06:01: It's a global phenomenon.

00:06:05: in the core of its meaning is really controlled over one's own data.

00:06:12: but also increasingly there are sovereignty discussions around AI models?

00:06:18: So do i have control over the models trained or where put my data?

00:06:25: and we see that whole topic broadening, and it's a global phenomenon.

00:06:30: Obviously from an European perspective.

00:06:33: there is a lot of talk around how we can decrease the dependency especially on US hyperscalers that are so dominant if something happens If the service shouldn't be available at some point anymore that you run into serious trouble as an enterprise being so dependent on that.

00:06:58: And this awareness is definitely now much bigger, and what we are seeing in terms of maturity?

00:07:15: Can we put some critical data or AI workloads more on local cloud providers?

00:07:24: Or should be move towards on-prem solutions, which is an interesting discussion because for the last ten of fifteen years.

00:07:32: We've definitely moved everything to the Cloud as much as we could.

00:07:36: and now there's a discussion on prem.

00:07:38: And it's interestingly I think interlinked with the AI cost discussion because we see more and more companies that tell us our AI business cases are not working anymore with the rising token cost, but we get from the AI providers.

00:07:58: And when you do something about it?

00:08:01: I see a certain pattern.

00:08:04: running your own models typically open source models and maybe running them on on-prem hardware even is a pattern we see increasingly.

00:08:17: And someone told me, well with that the only cost I really have to worry about is energy.

00:08:26: it's electricity i need to run this hardware.

00:08:29: obviously you have to buy To hire and have to have people that had to run this, I've too know how to operate an open source model.

00:08:40: So it's a bit more but obviously its not the worry about token cost.

00:08:45: so A lot happening in sovereignty.

00:08:48: The big shift surely was the fable moment we call it, yeah?

00:08:51: So when your government shut off access to fable through entropic fable I think that's something which is very highly noted everywhere.

00:09:04: so we had discussions with European companies and said okay now you cannot discuss this anymore as a theoretical thing but could happen right away.

00:09:15: But I also saw the same discussions with U.S companies that started to worry about access and how can ensure access maybe critically eye models, so thats why i see it as a global phenomenon With different aspects of what companies are worried or where they need to make sure they have access.

00:09:42: But it's definitely a topic that is rising with the criticality of AI for companies, data and AI for company.

00:09:52: The whole thing about control over it in sovereignty will also continue to rise.

00:09:57: It dovetails well into last subject uh, that we're not going to live in a single model world anymore.

00:10:05: And I think you made it great point.

00:10:07: um You also touched on cost and i think cost is the major driver for Uh companies as well as control As you just mentioned To control their own environment.

00:10:17: our own research shows That thirty six percent of company's responding to bark research?

00:10:23: I have plans to bring In what would call smaller domain level models like health care models or law models Or financial models to answer your questions better more accurately and at a lower cost.

00:10:35: And I think that's a trend we'll see going forward for quite awhile.

00:10:40: Yeah, no i agree

00:10:42: No all right.

00:10:43: well let's switch to what's happening in the vendor The provider side of the world.

00:10:49: uh...I wanted to touch on another piece Of our research.

00:10:52: they came out.

00:10:54: We talk about mature ai companies.

00:10:57: That number of maturity has been pretty low, a little bit startling that it hasn't moved.

00:11:03: So when Bart does research we have standard maturity model and we ask all respondents to fill out.

00:11:10: about nineteen-to twenty percent of everybody who's taken our research in the last two years is considered leaders.

00:11:18: they've built their act together for success.

00:11:22: Uh at the same time that means eighty percent of the people doing AI have not and that's a concern We just recently saw jump in companies.

00:11:30: That are ready to do.

00:11:31: uh, AI.

00:11:33: And we now are starting to see the maturity number closer to twenty nine percent.

00:11:37: Um what's your thought on that?

00:11:39: Do you think it's very regional?

00:11:42: Do You think it is about time?

00:11:44: What's?

00:11:45: ah, what're thinking on Being in the US, the U.S has a tendency of running off and doing things crazy.

00:11:52: And then where you are?

00:11:54: The EU tends to be little more conservative.

00:11:57: Yeah But we often worry that we're too slow.

00:12:02: so but first I think AI is happening everywhere.

00:12:08: In the North American market We typically see the pattern that adoption is faster and earlier.

00:12:17: I think that's still true here, but obviously AI is also a big topic for European companies.

00:12:23: So we see a bit of delay, but it's happening as well.

00:12:28: Let's shift to a topic.

00:12:30: you and i have covered for large portion our career business intelligence.

00:12:36: boy disruption has happened in there A lot work being done on what we might call the trust or context layer.

00:12:45: What's your take on what's occurring in the BI space right now?

00:12:50: Is it good, I mean is that a great thing or are we going to see some fallout from it.

00:12:56: I think its quite disruption.

00:12:58: so i think there is alot of things happening and the reasons are this or are they big AI platforms?

00:13:07: obviously.

00:13:09: So the capabilities have to analyze data create reports and anything you would expect from BI tools is increasing, clearly.

00:13:23: And that raises the question so what's happening now with the BI tools as we know them?

00:13:29: There are a lot of movement and I find it very interesting to see new vendors coming up!

00:13:44: who has the guts to found a BI vendor now.

00:13:50: But we talked with them, I mean... We are running right now the analytics and AI survey so our annual survey on the analytics users And in that we're also researching and have lots of interactions with vendors like Golden Analytics or Crack Tea And it's super interesting.

00:14:14: So yeah, so there are many new ideas to come up with.

00:14:17: they call that often AI native BI tool and the key idea is really either integrate into an environment or make a lot of use off.

00:14:29: AI capabilities for example come out with better user experience Yeah?

00:14:34: Or I used intelligence.

00:14:38: An AI platform can now bring, for example to find exceptions.

00:14:43: To explain what it found and so on...to the user.

00:14:50: So I think there's a lot of interesting things happening here.

00:14:54: Yeah!

00:14:58: We as analysts at BARC we attend a lot vendor events analyst programs And this year in the field I don't see the BI platform quite as often.

00:15:10: What i do see is, The investment the firm has made in semantics or context how they're feeding Context to the AI agents because that's something that Agents want very badly.

00:15:21: just understand the context of things and some bi companies have literally are okay with And have strategically given up the original UI.

00:15:34: A lot of times when we see demonstrations in the field now, it is like let's show you how Claude speaks to our BI platform and then what we're doing.

00:15:46: In our platform can then augment that process.

00:15:48: so there They're giving, they still want to have the car but their given up the steering wheel a little bit as far is how people will use the platform.

00:15:57: And I think it's kind of exciting and pretty interesting.

00:16:00: so for those listening from the user side today i'm sure everybody's experiencing some of these augmentations in new feature sets.

00:16:19: Executive level users, but you guys know who are and sometimes you need those stuff.

00:16:24: Those narratives an i think a lot of analysts or happy not to be the source about it anymore.

00:16:29: so like that's kind of an exciting thing thats happening in the vendor space.

00:16:37: I find that the governance topic emerges here again.

00:16:41: um... So we talked about what the user is seeing But I think it's worthwhile to talk about what the vendors are doing.

00:16:49: Um, i think that there is a lot of work being done in that category.

00:16:58: What you're thinking on

00:16:59: this?

00:17:00: Yeah We actually see it emerging as its own category.

00:17:05: So now we see more and more tools That focusing on the governance aspect.

00:17:11: And I mean, you can have a pretty broad definition of governance.

00:17:15: One would be including what we just talked about that if analytics and BI tools more or less give up their own user interface then they often really highlight on the semantic layer.

00:17:32: now many of these tools had for forever since their inception.

00:17:36: but this becomes so interesting that this capability is added to the capabilities of the AI platforms, typically over an MCP interface.

00:17:48: And I just want to mention at BARC we have a benchmarking capability for MCP servers and we did some super interesting work here where we compared what AI platforms can do by themselves when BI tools add their capabilities with an mcp server.

00:18:14: And it's too much detail to talk about here, but if anyone is interested please reach out and that's something we can talk.

00:18:22: more detailed very interesting findings there.

00:18:25: so in a way thats also part of the governance having semantics have context available to make sure that the user questions are answered in a right way.

00:18:40: Additionally, also looking at basically what I think there's a term emerging now it is called Harness.

00:18:49: so we see more and more companies say yes!

00:18:53: We're providing this maybe agentic harness or AI harness meaning yeah this governance called it maybe a layer around the capabilities of the applications, or of chatbots.

00:19:10: It's opening up a door which is a conversation.

00:19:13: we'll have more I guess in future... The harness and platform approach to governing AI and agents seems to be getting a grip.

00:19:24: You see it from the bigger data vendors, Snowflake Databricks and others are doing really solid job with harnessing their AI agents directly back into their governance wrapper.

00:19:35: And my takeaway right now is we're in that up-and down flow of point solutions.

00:19:42: They very exciting but at the same time The big platform vendors going you know hold on give us some moment We can help keep you out of trouble.

00:19:52: I think In the coming year with the population growth of agents in general, we're going to hear a lot of interesting stories.

00:20:00: Even more so than the big ones you referenced earlier.

00:20:03: was some agents doing some things at Anthropoc and OpenAI that got everybody's attention but were gonna see more of that especially from end-user companies who aren't putting into investment too really governs environment their AI or agents are working on.

00:20:21: I want to touch on one more, because we're on Agents and i like the topic.

00:20:25: What are your thoughts on that fact?

00:20:27: So for instance an enterprise has a thousand people using analytics tools And now they all have licenses to cloud or open AI Or they live in a token maxing environment.

00:20:42: We've suddenly multiplied our analytic environments By is a large multiplier.

00:20:47: they may have several agents attached to each individual and suddenly, the one from a thousand users two three thousand users are we gonna start breaking architectures.

00:20:57: We definitely do.

00:20:59: yeah so it's not only much yes its not on either number that is now increasing but also the intensity.

00:21:08: often agents do a lot of or more interactions with databases, whatever they are tasked to do.

00:21:17: And so we can assume that we see an explosion now off basically interactions with the system components that we have in their typical stack and that will surely break a few of the architectures.

00:21:37: Yep,

00:21:38: it certainly will.

00:21:40: Let's shift gears to our final category where we like to talk a little bit about M&A and investor perspectives.

00:21:47: We're seeing a lot of work being done on vendors looking To acquire these governance tools that were talking about night I-I...we could Talk About specifics but park research says That the adoption of agents is foreseen.

00:22:03: The vendor community in A good way To go find ways to serve this, I think we're seeing it more and more.

00:22:08: And i think uh... We'll get to talk about later in the year.

00:22:11: what's- What you are thinking from your perspective?

00:22:14: Yeah absolutely!

00:22:15: As be said as its becoming its own category.

00:22:19: obviously there is a lot going on with investments new companies or existing ones that now expanding And a lot of M&A activity.

00:22:28: and just wanna maybe highlight one, that's the acquisition of Natoma by Snowflake.

00:22:35: So they control over an MCP gateway and basically expand snowflakes capabilities for AI governance.

00:22:44: I think it is very prominent example in our industry to see there are lots going on.

00:22:52: I uh, i agree and we got to see the announcement for that when We were at this snowflake event.

00:22:58: This year?

00:22:59: I guess we're talking about snowflakes.

00:23:01: so let's talk About data brick.

00:23:02: um as well.

00:23:04: They Uh have had some interesting financial things happening recently That I think got everybody's attention with a new round of funding too what is by You know all evidence very successful company right now.

00:23:19: What's your take on this new This new valuation through this very large investment that happened at Databrush.

00:23:29: Yeah, it's interesting!

00:23:30: It is going on and actually everyone is waiting for the IPO.

00:23:37: but they add investment rounds now a five billion round.

00:23:44: so again for data company very significant amount of money And astonishing is surely the one hundred ninety billion valuation they got.

00:23:57: So if they do the IPO this year, it's really will be super interesting to see what that outcome would be.

00:24:07: so I think you can still happen.

00:24:10: Entropic is also.

00:24:12: the news are that they might do the IPO this fall, so they're getting ready.

00:24:18: why OpenAI might delay to twenty-twenty seven.

00:24:22: So that's also interesting too.

00:24:24: watch but I think we can see from data in the eye space will surely seem more things happening.

00:24:29: maybe even one or two prominent IPOs and valuation here should be as well including a lot of AI Maybe call it hype, I don't know.

00:24:41: But a lot of AI radiation in Databricks.

00:24:44: so they position them quite smartly as also an AI company which is important in the investment space here.

00:24:52: Yeah, I would say in the broader AI space and not focused on data bricks specifically but In the general AI space here in the US market.

00:25:01: there's quite a lot of discussion going out around.

00:25:04: you know Are we approaching a bubble phase?

00:25:06: And so once big vendors had earnings calls yesterday.

00:25:09: We'll talk about that with our listeners.

00:25:11: The next time we're on a call.

00:25:13: But sales force and Nvidia and others have had announcements recently.

00:25:18: and Going back to the Databricks thing We're going to see some record breaking valuations as some of these companies do find their way to the stock markets in the world.

00:25:29: So it should be kind of, I guess entertaining to watch for sure talking about other activities specifically in the analytics space.

00:25:40: Domo was acquired recently and equally by progress will be monitoring that.

00:25:49: what were your takeaways from the transaction.

00:25:51: Well,

00:25:51: it was a bit opposite of record breaking valuations because valuation is pretty bad for analytics and data analytics companies.

00:26:05: But it was, Domo in a pretty bad spot terms of how the whole overall set up off the company.

00:26:13: now we are not talking about product or anything but that's really more than the governance of the company.

00:26:19: I think you have some maybe want to share some more details.

00:26:23: Well, I think there's been disruption at the executive level.

00:26:27: They're pretty well documented and uh i'm sure There's no way it's not a distraction or wasn't a distraction within the firm.

00:26:35: so Yeah certainly A little bit of a surprise when it occurred but also at the same time for those of us that are really close in watch The space closely.

00:26:46: We weren't surprised that something had to happen there And then it was likely to happen.

00:26:51: actually they talked about A strategy in their very own corporate blogs about what they intended to do with the company.

00:26:58: so there is for warning, advance warning of what was likely to happen at Domo.

00:27:03: So that's underway as well.

00:27:06: more positive side I saw you wrote a really interesting piece on LinkedIn yesterday but an acquisition just happened this week.

00:27:13: You want talk about duck labs?

00:27:16: For sure.

00:27:17: So at BlackDB, from our point of view was one the most interesting database startups for last years and they had some real interesting innovation in their product that became more popular.

00:27:36: Now that they have been acquired by AWS, I think may is quite a smart move by Amazon.

00:27:43: Because they acquire very good technology and basically they acquired the team because DuckDB remains open source and IP is in foundation.

00:27:56: So they went to some length there, to keep this IP independent open source.

00:28:05: But obviously AWS now has the team and are surely going to commercialize it but can scale it technically because so far DuckDB is often used in more small-scale environments or at the edge of embedded scenarios.

00:28:26: But they can also obviously bring scale to customer access.

00:28:29: So I think it has a lot of potential for AWS, yeah?

00:28:34: Maybe add to Redshift maybe as the query engine with that may be even use the file format DuckDB has so... A lot interesting technology here that can now scale.

00:28:50: We surely will watch it, but I thought that was a good move on.

00:28:54: AWS acquired something really good out of Europe.

00:28:57: here we have to highlight that.

00:28:59: uh That was an innovation out of the Netherlands from CWI.

00:29:04: It's quite a prestigious institute on database research and so there is spin-off There And i can only congratulate the founders.

00:29:15: they made this step And we see what happens now.

00:29:20: I agree, um i enjoyed reading your piece on it and the rest of the news coverage on it.

00:29:26: It does seem like a great idea?

00:29:28: I almost missed one.

00:29:29: Um stripe bought open router and-and The reason I almost miss it is...I know who Stripe Is!

00:29:35: And..And I Know What They Do In The World Of Commerce.

00:29:38: And If It Wasn't An Immediate Fit To My Brain For Our Data Analytics & AI Coverage At Bark It Was Pretty Strategic

00:29:46: Absolutely.

00:29:47: I mean, everyone who bought them open router they bought and fits to the discussion.

00:29:53: we just had that company start to run more than one model.

00:29:57: so this gateway layer yeah?

00:30:00: That decides okay which models do are used for which use case becomes more and more important for companies, especially for cost reasons.

00:30:12: So that you're easy AI workloads use an open source model or something that's not so costly?

00:30:19: We all agree this layer and the technology here will gain in importance of companies.

00:30:26: And now Stripe acquired.

00:30:27: one is super interesting.

00:30:28: But on other hand This layers also controls what work loads go where.

00:30:36: And then these workloads have to be built.

00:30:39: They had to be ready for and that's where Stripe comes in, I thought it was very interesting again quite a maybe smart move also yeah?

00:30:50: To be on this intersection layer.

00:30:55: when the workload run through there is also financial stream has to be administered or handled and Stripe often doing that.

00:31:05: And now they control software that basically controls this.

00:31:10: So I think once you first thought who buys open router but then when you think about it, oh!

00:31:17: That actually makes a lot of sense.

00:31:19: i think

00:31:20: I absolutely agree and like how it ties back to some of our topics today.

00:31:25: We've mentioned everybody listening, you will have many models in your organization.

00:31:30: so optimizing traffic for the right model is going to become extraordinarily important.

00:31:35: .I don't know that there'll be a standalone segment on its own but It's gonna feature set.

00:31:39: You're gonna see almost every environment because we also know advanced users of AI through BARC research indicate cost Is their biggest headache their biggest challenge to success.

00:31:50: So, yeah almost missed it.

00:31:52: glad I didn't and i'm glad we got to talk about it.

00:31:55: thanks this is always a fun conversation to have with you And uh...I look forward to our next one-next quarter.

00:32:02: Absolutely!

00:32:02: Thanks Sean Talked To You

00:32:05: Soon.