Interviews
Leo Brunnick, Chief Product Officer at Cloudera – Interview Series
Add Unite.AI to your preferred sources on Google
Leo Brunnick, Chief Product Officer at Cloudera, is a veteran technology executive and entrepreneur with more than three decades of experience building and scaling software, digital media, and enterprise technology organizations. At Cloudera, he leads Product Management, Engineering, and Customer Support, overseeing the company’s product and technology direction as it expands its data and AI platform capabilities. Before joining Cloudera in 2025, Brunnick spent six years at Naviga, most recently as Chief Operating Officer, where he led more than 600 professionals across product, engineering, marketing, R&D, customer support, and professional services. Earlier in his career, he founded and served as CEO of Patheos, growing the digital media platform to approximately 15 million monthly visitors before a successful exit, and spent eight years as Chief Product Officer at enterprise content management company Vignette. His background also includes technology consulting at Andersen Consulting and service as a U.S. Marine Corps infantry officer.
Cloudera is an enterprise data and AI company focused on helping organizations manage, govern, analyze, and apply AI to data wherever it resides—across public clouds, private data centers, and edge environments. Its hybrid platform combines data management, analytics, machine learning, generative AI, and increasingly agentic AI capabilities while emphasizing enterprise security, governance, privacy, and workload portability. The company’s current platform strategy includes Cloudera Anywhere Cloud, an architecture designed to give enterprises a consistent cloud experience across different infrastructure environments, alongside an open data lakehouse, unified data fabric, real-time data processing, and tools for developing and deploying AI models and agents. Cloudera says its technology manages more than 25 exabytes of data and is used by large global enterprises seeking to modernize data infrastructure without requiring all workloads or sensitive data to move into a public cloud.
Your career has spanned technology consulting, leading product organizations at Vignette and Naviga, founding and scaling Patheos, and now serving as Chief Product Officer at Cloudera. How have those experiences shaped the way you think about building enterprise technology platforms at a time when AI is fundamentally changing how organizations use their data?
Across all of those situations, the same principle applies. Technology is most valuable when it makes something complicated easier for people to use. This point is even more important now, as AI creates significant opportunities while also adding another layer of complexity for businesses already navigating cloud data, data centers, and increasingly sovereign environments.
As we build this next era at Cloudera, I often consider how we can enable organizations to benefit from new technologies without requiring them to rethink everything they have already built. Over the past several years, enterprises have built up their data estates, applications, governance models, and operational processes. It wouldn’t be reasonable to discard all of that every time a new technology appears. The platforms that succeed in the AI era will build on what organizations already have while giving them the flexibility and control to adapt as their needs evolve.
Cloudera recently introduced Cloudera Anywhere Cloud to help enterprises build and operate AI across public clouds, private infrastructure, sovereign clouds, on-premises environments, and the edge. What problem were you seeing among customers that convinced Cloudera this type of unified architecture was necessary?
There is growing pressure on customers to put AI into practice, even as their data and infrastructure are spread across more environments than ever before. Data may live across public clouds, private data centers, sovereign environments and the edge, each with its own security, regulatory, performance and economic requirements. Managing all of these environments can create a significant operational burden for organizations.
The idea behind Anywhere Cloud is that businesses should not have to choose between the agility of the cloud and maintaining control, security, and governance over their data. Organizations need a consistent way to deploy, govern and scale data and AI services wherever it makes the most sense for the business. With a single control plane, companies can spend less time managing the underlying infrastructure and more time focused on the outcomes they want to achieve with their data and AI, no matter what environment it lives in.
You’ve said that enterprise AI has “outgrown the public cloud-only model.” What are the biggest limitations enterprises encounter when trying to run production AI entirely in public cloud environments, and which workloads are increasingly moving elsewhere?
The platforms that succeed in the AI era will build on what organizations already have and give them the flexibility to adapt as their needs evolve. Public cloud remains an important part of enterprise technology strategy, but it is not the only option. As AI moves into production, companies need to determine where their data and workloads should run based on factors such as regulation, security, and cost.
This shows that decisions about where to place workloads are now being made more deliberately. Certain AI applications would certainly be appropriate in the public cloud, whereas workloads that involve highly sensitive or regulated data might have to stay in a private data center or a sovereign environment. In other cases, workloads will be better off at the edge since latency is an important factor. The aim is not to move away from the public cloud; rather, it is to allow enterprises the flexibility to run each type of workload in the location that is most suitable for it, without being forced to create a separate operating model for each environment.
Cloudera Anywhere Cloud includes an agentic copilot that can translate plain-language requests into data workflows and infrastructure operations. How autonomous do you expect these systems to become, and where should enterprises maintain human oversight as agents gain greater control over critical data infrastructure?
Agents will continue to take on the repetitive operational work which currently occupies the time of data and infrastructure teams. If an agent can express an outcome in plain language and then convert that into a workflow or an infrastructure action, it will greatly reduce friction and make advanced data capabilities available to a larger number of people.
Greater autonomy also creates a greater need for effective governance. Companies must have clear policies specifying what an agent is allowed to access, what actions it may perform on its own, and when human approval is required. The greater the potential impact of an action, the more crucial those controls become. Autonomy and governance should advance in parallel to provide agents with sufficient freedom to achieve real productivity gains while ensuring the organization maintains oversight and control over its key data and infrastructure.
Cloudera is working with NVIDIA to bring native GPU acceleration through cuDF to Apache Spark 4.1 workloads, with potential performance improvements of up to 4x without requiring organizations to rewrite existing PySpark or SQL applications. Why is eliminating the need for code rewrites so important for accelerating enterprise AI adoption?
Companies have made significant investments in their existing applications, skills, and data pipelines. Switching to a new technology requires more than engineering work. It can introduce risk, delay results, and create disruption across systems that are already working.
Zero-code acceleration helps address that challenge. Organizations can use their existing PySpark and SQL workloads to gain the benefits of GPU acceleration without fundamentally changing how their teams operate. This makes the technology easier to adopt while also improving the performance of the data pipelines that support analytics and AI and protecting existing investments.
GPUs are increasingly being used beyond model training and inference, including for data preparation and analytics. Do you expect GPU acceleration to become a standard part of the enterprise data stack, and how could this change the economics of preparing massive datasets for AI?
In my opinion, enterprises will become much more skilled at matching their infrastructure to their workloads. GPUs have already changed the way model training and inference are carried out, and data engineering is the next obvious area for innovation, since preparing large datasets can be a major bottleneck when getting AI applications into production.
The economic impact is just as significant, as faster processing can do more than simply complete a Spark job more quickly. Reduced compute runtimes can lower infrastructure costs while helping data teams iterate faster, an increasingly important consideration as AI workloads grow. According to our research, 84% of respondents have seen AI workloads increase infrastructure costs, making it more important for businesses to look across the entire data pipeline for opportunities to improve both performance and cost. GPU-accelerated data processing could become an important part of that equation.
One of the promises of Cloudera Anywhere Cloud is to allow organizations to place workloads wherever regulatory, economic, or operational requirements dictate without continuously moving or copying their data. How important is bringing AI computation to the data, rather than bringing all enterprise data to the AI?
This is essential because businesses have spent years building vast amounts of data across different environments, and moving all that data into one place for AI is often impractical. It can be costly and time-consuming, while security, privacy, or sovereignty requirements may prevent organizations from moving certain data altogether.
At this time, it is not practical for enterprises to securely move all of their data to the different applications and environments where AI is being deployed. Bringing AI to the data allows organizations to use more of the data they already have while keeping it where it needs to be. It also gives them the flexibility to run workloads where it makes the most sense based on business and technical requirements. As AI becomes more embedded across the enterprise, the ability to operate consistently wherever data resides will become increasingly important.
Data sovereignty has become a major concern as governments and regulated industries adopt AI. How are sovereign AI requirements changing the architecture enterprises need, and do you expect sovereign clouds and private AI infrastructure to become significantly more important over the next few years?
Absolutely. Sovereignty is becoming an architectural requirement, not simply a compliance issue. Organizations need to know where their data is stored and where their AI workloads are running, as well as who has access to and how intellectual property and sensitive information are protected.
Organizations can no longer assume that everything can be centralized within a single public cloud. They need the flexibility to operate across sovereign clouds, private infrastructure, public clouds and on-premises systems while applying consistent governance across each environment. These requirements will only become more important as AI moves further into regulated industries and critical business processes. Organizations that can innovate while maintaining control over their data and AI environments will be better positioned to scale AI responsibly.
As autonomous AI agents gain access to enterprise data, applications, and infrastructure, traditional data governance may no longer be sufficient. How does governance need to evolve when the users accessing and acting on enterprise data are increasingly AI agents rather than humans?
In the past, governance has mostly been based on people and applications being able to access data. Agents add a new aspect, as they can access information and reason about it and take action based on what they find.
Because of this, businesses must apply ideas such as identity, permissions, lineage, observability, and zero-trust access to agents to ensure security and governance of their data. Organizations should know which agent accessed which data, what it did with that information, what actions it took, and whether those actions were in line with corporate policies. Governance can’t be introduced after an agent has taken action; it must be built into the environment so that policies and safeguards move along with the data and are continually enforced regardless of whether the entity in question is a person, an application or an autonomous agent.
Cloudera is emphasizing open standards such as Apache Iceberg, Polaris, Spark, Kafka, and Trino rather than requiring customers to operate entirely within a proprietary stack. As AI infrastructure consolidates around large platforms, how important will open standards and interoperability be in preventing a new generation of AI vendor lock-in?
The more AI becomes part of enterprise infrastructure, the more important open standards become. AI is evolving quickly, and organizations need to avoid making architectural decisions today that could limit the models, engines, infrastructure, or technologies they can use in the future.
Technologies such as Apache Iceberg, Spark, Kafka, Trino, and open catalogs give organizations the flexibility to use different tools for different workloads while maintaining interoperability across their data environments. This is particularly important as enterprises increasingly operate across public clouds, sovereign infrastructure, and private data centers, where they need the freedom to decide where their data and AI workloads run without rebuilding around a single provider.
An open approach also makes it easier to adopt new AI capabilities while building on existing investments, rather than repeatedly moving or copying data as technology evolves. As the AI landscape continues to change, open standards give organizations the flexibility to evolve their architecture over time while maintaining choice and control over their data, infrastructure, and technology decisions.
Thank you for the great interview, readers who wish to learn more should visit Cloudera.
