Google Cloud Next 2025 | Yasmeen Ahmad, Google Cloud

Clips
More from Google Cloud Next 2025

Yasmeen Ahmad

Product Executive - Strategy & Product Management: Data, Analytics & AI

Google

play_circle_outline BigQuery data platform momentum and customer growth

play_circle_outline Enhancing Performance Through Unifying Data and AI: Exploring GenAI and Multi-Modal Data Use Cases

play_circle_outline Automation in data pipelines with AI agents

play_circle_outline Collaboration between humans and AI for analytics workflows

Info
Transcript

Yasmeen Ahmad, Google Cloud

Yasmeen Ahmad

Product Executive - Strategy & Product Management: Data, Analytics & AI Google

In this insightful video from Google Cloud Next 2025, Yasmeen Ahmad, a distinguished expert in data platform strategy and analytics, joins theCUBE to explore the innovative breakthroughs of Google Cloud. Ahmad, of Google Cloud, discusses the evolution and impact of BigQuery, highlighting data trends and significant enhancements in AI capabilities.

Ahmad shares their expertise in cloud data warehousing, revealing key innovations and strategic advancements in Google Cloud's data platform. With additional insights, this video uncovers crucial developments... Read more

explore Keep Exploring

What were they discussing about the number of customers on Google Cloud's data platform, BigQuery? add

What are some key considerations when integrating AI with data systems according to customer feedback? add

What new capabilities does Gemini 2.0 bring to data pipelines and how are they changing in the future or today? add

What capabilities does Gemini's thinking model provide in terms of explaining its answer and allowing collaboration on building a plan? add

bolt Powered by CUBE AI

Yasmeen Ahmad, Google Cloud

search

Savannah Peterson

>> Good afternoon, cloud community, and welcome back to Las Vegas, Nevada. We're here coming to the end of day two of our three days of coverage on theCube. My name's Savannah Peterson, bringing you all the insights and the data breakdown with Dave Vellante all week. Dave, I feel like this conversation we're about to have is very much in your wheel house.

Dave Vellante

>> Yeah, my favorite topic other than sports.

Savannah Peterson

>> Other than sports, is data?

Dave Vellante

>> Yeah.

Savannah Peterson

>> Yeah, I know.

Dave Vellante

>> Absolutely

Savannah Peterson

>> Something you and Robert Herjavec have in common, actually.

Dave Vellante

>> There you go.

Savannah Peterson

>> Obsession. An absolute obsession with data.

Dave Vellante

>> Wish we had more in common, if you know what I mean.

Savannah Peterson

>> Yeah. Holler. Oh, but here to tell us all about it, and share some very exciting new data that was just announced, is Yasmeen. Yasmeen, thank you so much for taking time to come hang out with us.

Dave Vellante

>> Thank you. It's a pleasure to be here with you discussing data, also one of my favorite topics.

Savannah Peterson

>> I feel like I'm in really good company. I might actually be the least data nerd on the desk right now, and that's saying something, because I'm really into it as well. So without further ado, you are able to share some numbers that were announced this week. Tell us all about that.

Dave Vellante

>> Absolutely. So for the first time, we shared the size, and scale, and momentum that we've built with our data platform, BigQuery. So we shared we have five times the number of customers than the other two data platform, data warehouse, AI platforms in the market. And that's super exciting.

Savannah Peterson

>> Five times?

>> Five times.

Savannah Peterson

>> That's pretty significant.

>> That is very significant. And part of that five times is just down to both Google's leadership in data and AI, which over the last year has meant we've seen an influx of customers coming to Google Cloud, from yes, on premises to modernize into cloud, but also from other clouds and other data platforms, as they want to come and leverage Google's best in class AI combined with our data technology.

Dave Vellante

>> And that's a customer number, like a counts, right? Is that correct?

>> That's number of customers.

Dave Vellante

>> You wouldn't care to share revenue comparisons, would you?

>> Unfortunately, they won't authorize me to share that. Just that stat alone is amazing for me to be able to share, because it shows Google's momentum in this space. And just over the past year, we've seen a 300% increase in the use of our migration services. We've seen a 30% increase in the data volumes we're managing. It's exabytes of data.

Savannah Peterson

>> Wow.

>> So it just shows the momentum of customers really moving to leverage our GenAI and data technology.

Dave Vellante

>> But I mean, BigQuery was a true cloud data warehouse native from the beginning. Separating compute from storage, obviously, was the big thing back last decade. Now the big conversation's around, essentially, separating data from any engine, right? The whole open table formats, and iceberg, and all that brouhaha. It seems to be of interest to customers, brings a lot of complexities. What's your thinking on that trend? What are customers telling you?

>> Absolutely. I think you're picking up on a key trend. So last year we announced the unification of a number of services into BigQuery, so including data, governance, AI, streaming. So that, last year, was that first step into a unified platform with all of these capabilities. We've taken it a step further this year to now truly unify data and AI together. So one of the key things we heard from customers is, AI bolted onto data systems doesn't work. I don't want to have to duplicate my data, take it out of my data cloud to go do AI. I need the AI brought to the data. And we also published numbers in a study we did, showing that at Google we're 8-16 fold cheaper to run AI, multi-modal AI over multi-modal data, because of how we're infusing data and AI at a foundational level. So we are seeing a huge number of GenAI multi-modal use cases now, over multi-modal data in BigQuery. And it's only because we are making it so easy to work with all of this different type of data, whether it's in an open data format or it's in a managed storage service from BigQuery. Exactly as you mentioned, we're separating that data foundation from the engines. So the engines now all operate through our unified catalog across the single data foundation.

Dave Vellante

>> So what's so smart about that, I mean, you guys obviously-

Savannah Peterson

>> It's a lot of things smart about that. Yeah....

Dave Vellante

>> know what you're doing here, but the trend is, the point of control is shifting from the database to that governance catalog layer. But the value actually is even further upstream, especially as you bring in AI. And so that's interesting... It creates an interesting dynamic for a lot of the players that rose up because they had, whether it was a great data platform, or maybe they had ML chops. So the whole world is now, you bring in agentic, it's really changing. How do you think about where the value is flowing for customers? And obviously, the role that data plays, harmonizing that data, unifying that data, being able to govern that data. But the value's moving upstream, isn't it?

>> Absolutely. And we actually see, there's innovation happening at every layer of the stack. So when I think about the data foundation, just yesterday, we launched a number of autonomous data foundation capabilities that are powered by agents, agentic AI. So for example, in that autonomous data foundation, we now have automated metadata generation, automated cataloging at scale. We've launched features around anomaly detection. All of this is in service of actually making the hard work of data much easier, so that organizations, and the humans and those organizations, can move up and focus on the business outcomes, instead of managing data. So as customers bring a lot of multimodal data to BigQuery, and overall our BigQuery BigLake services, we don't want them to have to be thinking about managing data, and instead be able to interact with some of our agent capabilities on the top layer that allow them to actually drive insights. You saw me showcase the data science agent, the data engineering agent, our new conversational analytics agent. We want humans to be focused on thinking about their business problems and how they can use data to solve those problems, versus managing the data infrastructure, if you like.

Dave Vellante

>> So said another way, sorry .

>> Right.

Dave Vellante

>> The database administrators live, which really was mundane, a lot of performance tuning, a lot of metadata management, gets dramatically simplified. That mundane stuff goes away, they can focus on other stuff. I've been surprised, actually that more companies haven't.... I mean, Oracle's obviously made a big deal about autonomous database. Google is there, and really driving that automation, which is game changer for the life of a DBA.

>> And we actually see that metadata generation, semantic generation, is also being critical in this AI era. So as we are building these agents and agentic capabilities, they have to work on a data foundation. And we have the largest customers in the world. They don't have small data foundations. They have millions of tables, billions of records, billions of columns of data. How is an agent or agentic capabilities supposed to make sense of which data to select from that vast sea? So actually, the work we've done on this automated metadata generation semantics builds into what we launched yesterday, the BigQuery knowledge engine. The knowledge engine is actually a dynamic knowledge graph that extracts its intelligence from metadata, from usage logs, from history. It allows it to create a context of the business, and how data is related. And that knowledge graph is the shared intelligence that powers all these agents. So when these agents get to work, we see by using the knowledge engine, they're 50% more accurate at identifying the right data sets. And through using semantics in this knowledge engine, we see two thirds better accuracy at answering business data questions.

Dave Vellante

>> That's how you're harmonizing all that disparate data, serving up the agents with confidence. And you've got the governance piece of it. You've got the technical metadata, the business and operational metadata all unified. Brilliant.

Savannah Peterson

>> And once you have your knowledge graph, you can really hit the races and hit the ground running, especially if that's well compiled and put together. I'm curious if you're noticing, especially given the incredible customer numbers that you have, are there specific industries that are jumping into this transformation faster than others? Are there any other trends that you're seeing, just in general?

Dave Vellante

>> It's been amazing for me to see just how fast industries have been able to adopt, whether it's the agent AI capabilities that we're building. This is faster adoption than any other technology we've seen. And part of it is, you've got the largest enterprises, they're innovation leaders, often ahead of the pack, able to invest in these new technologies. But actually, some of the agents in agentic AI are making it easier, even, for smaller organizations, startups, to SMBs, to actually invest in their data foundations and move faster, because it's giving them the skills that they would've traditionally had to go out and hire for, maybe didn't have access to. So it's, in some ways, GenAI has been a leveler, in terms of the playing field of data foundations for all organizations. The interesting use case, mixed though I do see, is yes, customers are doing the external use cases we've always seen with data analytics, like customer analytics, supply chain analytics, and so on. But the number of internal use cases, use cases to support inside the organization, operational workflows, helping the employee workforce go faster, those are use cases we traditionally did not see. And so Mattel, they were in the keynote, their CEO talking about how they're using BigQuery, Vertex AI, traditionally they had a lot of product feedback coming through a lot of retail channels. They would use, actually, spreadsheets to try and make sense and pull out signals from this feedback. Now that customer feedback, multimodal data, not very structured, comes into their multimodal data foundation in BigQuery, and they use Vertex AI to pull signals. And they shared a great example of a new product they released. It had a little elevator in it and there was a problem with the door. And they were able to pick up the issue with the product, and actually fix that in the same production run.

Savannah Peterson

>> Wow.

>> And have the product fixed. And I call this the real-time data activation flywheel. That's what we want to enable with customers. The speed at which they can actually land data, see it be processed, get to insight and activate it, we are shortening that cycle, and we're seeing that cycle shorten with customers on a lot of these internal use cases on unstructured data.

Dave Vellante

>> With all this automation-

Savannah Peterson

>> That's amazing....

Dave Vellante

>> and intelligence that you're injecting into the system, how do you see the data pipeline evolving, maybe changing completely? The work that Zhamak has done? Zhamak Dehghani, if you're familiar with her work, the data mesh, she'd laid out the problem brilliantly. How will data pipelines change in the future? Or how are they changing today?

Dave Vellante

>> So customers, even though we've been on years of journey on bringing together data foundations, bringing together the ideal customer 360, it's still been a huge challenge for organizations. In the keynote demo, you saw me show bringing together SAP, Salesforce, Google Ads data through a data engineering agent, through natural language, and asking it to build workflows. This is where I think the magic of Gemini really comes in. It was a game changer last year when Gemini 2.O came out with the thinking model. Because previous to that, we were working with GenAI models, but with complex tasks like data pipelines, we were having to try and figure out, it's not just a simple question-answer, I need you to do a multi-step thing. How do I break it into multi-step, and then ask the GenAI foundation model to build that step? Well, as soon as Gemini 2.O came out with its thinking mode, it can now think to create multi-step workflows and actually often create several options of them too. So you now have a technology that can create those workflows dynamically, whether it's three steps, or 10 steps, and you're not having to teach it how to do those steps. And within each step, we're teaching it actually how to use tools or other agents. So it's like, "Hey, you have access to a tool called BigQuery Machine Learning, where you can do lots of sophisticated data quality metrics." Or you have access to BigQuery anomaly detection, so as you're building this pipeline, you can be doing anomaly detection, and understanding if the data coming in today is different from the data patterns we've been seeing for the last month. Metadata generation, it can invoke metadata generation. So suddenly, having a model that can actually come up with multi-step workflows, and having a model that can activate tools, is super powerful. The combination of what you can do, the possibilities are endless.

Dave Vellante

>> Stay on that for a moment, because, unfortunately, we didn't get to see your demo because we were here. But you pulled data from SAP, from Salesforce, from Google Ads, you've got the harmonization, the semantic layer, some people call it. So you've got that piece covered. And then you're able to then allow the agents to work on that. And then the metadata piece, previously, you would've to pull data out of those systems, move it into a data warehouse or a data lake, and then at some point in time add the metadata in, bolt that on. You're saying all that is now done dynamically?

Dave Vellante

>> So as I wish the demos in the show floor here are showing in some of our sessions, the data engineering agents that we are building basically pull all of those pieces together with Gemini's model. So Gemini, the thinking model is what is able, we teach it how to use each of those tools. We just say, "Hey, you have a tool called Metadata Generation. You have a tool called Anomaly Detection." So it activates those tools based on what the user's prompt is. So depending on how I prompt the data engineering agent, it can begin to activate those tools, and help me dynamically create a pipeline. And we do actually see this as a collaboration between the user, and today, the agent. So with conversational analytics, that's a prime example where we put the agent to work, it builds a multi-step workflow to give you an answer. But what we're finding is the human user, often, if you think about real world analytics and how that happens, I don't just go to my analytics team and say, "Hey, do this for me, and get the answer in that space." Typically, you collaborate. You say, "Oh, well now that I see that, that's interesting. Could we actually, instead of looking at it by product category, maybe look at it by a different product segmentation?" So you'll collaborate. So with Gemini's thinking model, now Gemini can actually explain how it got to its answer so you can actually expose the plan that Gemini is building, and then have the user actually collaborate on that plan. So through our interfaces, what we're also giving is users the optionality to see that plan, collaborate, change things within that plan, and then come up with that perfect pipeline, come up with that perfect analytics workflow. So I wanted to labor on that point, because today, it's not just about the agent goes away and does it all, it's about creating transparency and collaboration.

Savannah Peterson

>> It is about that. And potentially, surprising and delighting yourself with what you're able to create or get done as a result of that. I can imagine you get to learn quite a bit from your customers as you go through these processes, and see what they do, or how they prompt things.

>> Yes.

Savannah Peterson

>> Or what they build. It's such an exciting time right now. All right, Yasmeen, I have one final question for you. When we're hanging out here at Google Cloud Next in 2026, same time next year, what do you hope to be able to say then that you can't yet say today?

Dave Vellante

>> What do I hope to be able to say then? So I spoke about some of the momentum with our customers. I'm just super excited about what our customers will do next with this technology. I think customers are only scratching the surface right now with what's possible. They're only discovering what's possible. So over the next series of months, we're actually doing a ton of innovation workshops with our customers where we're showing them demos of what's possible today. And you just see the eyes wide open.

Savannah Peterson

>> I bet that's so fun.

>> "Oh my gosh, if I could do this, then I can do..." boom, boom, boom. So I think by 2026 Next, you will just see a whole level of step change innovation with customers activating multimodal data foundations, being able... The next sophistication of these agents really working in harmony with their customers to do things that were thought impossible. And maybe just as a final comment, I'll reflect on the Wizard of Oz Sphere Event

Savannah Peterson

>> Yes. So many people have talked about that this week.

Dave Vellante

>> Yes.

Savannah Peterson

>> It's been really magical.

>> That was phenomenal. And if you watch the making of the Wizard of Oz, you'll see even our DeepMind teams saying three months ago, six months ago, "We thought this was impossible." And three months later, it's possible. So as we meet these customer challenges, what I love about Google is we love leaning into the impossible. We love leaning into the problems where it's, "Hey, this has never been solved before." So I'm also looking forward to how we make those impossible things possible in various industries, be it medication, pharmaceuticals, in movie making. I think there's just limitless possibility here.

Dave Vellante

>> So that is the so what of Wizard of Oz, because some of the analysts are saying, "Yeah, okay, what's the big deal?" Well, that was really hard. What else can Google solve that is really hard? And that's the point. That is the so what of the Wizard of Oz story.

Savannah Peterson

>> Yeah, we got there. Bringing it full circle from the beginning of the show. Yasmeen, thank you so much for this. This was really wonderful. We look forward to keeping up with you, and sharing some of those customers stories, and those fun examples next year. Dave, always-

Dave Vellante

>> Yeah, thanks for sharing.

Savannah Peterson

>> Yeah, that was awesome.

>> Thank you.

Savannah Peterson

>> Dave, thanks you for hanging out.

Dave Vellante

>> Yeah, of course.

Savannah Peterson

>> And thanks to our fabulous community for tuning into our three days of coverage here at Google Cloud Next in Las Vegas, Nevada. My name's Savannah Peterson. You're watching theCUBE, the leading source for enterprise tech news.