Confluent has been leveraging IBM’s global sales footprint to accelerate its enterprise adoption in India. From 600 odd employees talking about Confluent earlier to 100,000 IBM employees taking Confluent to customers, the scale up has been massive. In an exclusive interaction, Rubal Sahni, VP, Confluent, an IBM company tells Rajneesh De, Group Editor, APAC Media & CXO Media how the IBM acquisition of Confluent has been a perfect example of a Better Together story.
IBM announced its $11bn acquisition of Confluent in December 2025 and it got operationalized in March. This was the second biggest acquisition made by IBM in its long history. How would you assess the impact of the acquisition today?
The IBM acquisition of Confluent has been a perfect example of a Better Together story. Since the time we have been integrated with IBM, our reach has increased a lot. Enhancing our enterprise scale, Confluent has been leveraging IBM’s global sales footprint to accelerate our enterprise adoption.
IBM has outreach across all the segments, all the industry sectors. It has been solving different problems for different enterprises right from security to infrastructure to AI to data. So as part of the Better Together story, when we now go and meet our customers, we start with data, trying to solve the data problems for them. But then it eventually leads to solving their compute problems, the sovereign AI challenges, and even integration with some of their legacy environments. And this is precisely where IBM contributes with so many solutions to solve these problems.
The integration has also expanded our hybrid cloud outreach, filled gaps in our portfolio and fueled our agentic AI deployment. Today IBM combines Confluent’s multi-cloud capability across AWS, Azure and GCP with Red Hat, integrates Confluent’s Kora engine to bridge real-time context gaps and even secures the real-time data flows needed for Watsonx and GenAI
No doubt, Confluent has been embraced really well within the IBM ecosystem. Our Chief Product Officer (CPO) Shaun Clowes, has been elevated to the ranks of General Manager of IBM’s entire data and AI business globally. Some of the other senior leaders of Confluent are also being elevated to run bigger roles in IBM.
Another way to look at it is we were 600 odd Confluent employees in India earlier. They were talking about Confluent to all the customers and to the ecosystem. Today we have more than 100,000 employees in the country who talk about Confluent. You can imagine the scale, the number of conversations, the number of opportunities we are engaging and trying to help our customers shift to be ready with AI at production scale.
Confluent has often been mentioned as a central nervous system for enterprises. How has been your experience in the last few months with the Indian enterprises especially since the acquisition by IBM? How are these enterprises looking to productionize their AI strategy?
Our cloud business in India has been growing more than 70% year over year. Out of our overall business, close to around 75% is coming from the cloud business in India. We have a dominant share in all the digital natives in the country. The remaining 25% comes from our on-premise solution called as Confluent.com. These are primarily for the regulated sectors like the banks and PSUs.
Now with the help of IBM, we are also venturing into the traditional enterprises, more into the public sector, defense and the likes. We are taking the Better Together story to our customers. Every enterprise needs real-time, fresh, contextual data not only to streamline their operations and be more efficient, but also to be ready to productionize their AI.
While Confluent started with streaming as our main pillar, most enterprises are today looking at Confluent as not just the data streaming platform but even as the processing platform where data materializes as soon as it is generated. As soon as the data is generated, Confluent helps enterprises process it for preventing fraud, for real-time offers, and giving a fresh context to an AI agent. The AI agent can then make a decision himself or herself, rather than leveraging a human in between, so that the human does not become a choke point. This is how AI is getting productionized now in enterprises.
From this productionization point of view, where are the enterprises still facing challenges and how can Confluent help them with those challenges?
Many organizations often start going into experimenting with AI without solving their data landscape. They start implementing towards the outcome of AI, rather than looking at how they should serve the data into the AI application. For this the context of the data is so critical.
Any historical data can only get you an accuracy of 90% in AI. And if you have an accuracy of 90% in AI, you cannot move to production. You need to get to 99% at least 98% for doing so. That is why fresh context is very important when you run it with a historic dynamic variability. Then too the data is collected from various sources and pipelines created so that the data can seamlessly flow in real-time.
This is where Confluent helps enterprise customers to bring the real-time data at scale from various sources continuously. On top of it, we have improved our MCP servers. With the help of that, an agent-to-agent communication has become much better. Plus, we have governance layer on top of it to ensure that the agents are not going rogue and getting some information, not hallucinating and always serving the right data. This is required to go into the agents to make any decision or show it to the end customer or end user.
Another challenge is that enterprises cannot use AI to solve a $10 million problem by investing $15 million on it. That is the biggest problem the enterprises and digital natives are facing in the country today. The cost of compute, the cost of these AI, the tokens which are being generated, is really killing them in terms of their budgets.
Confluent is helping them by launching a product called TableFlow. As part of that, you write the data once, but you read it multiple times. What happens today in the organizations is the same data gets transferred, moved multiple times and read multiple times across various stages and across various applications. And because of that, you pay on the compute and travel of the data networking cost multiple times.
But Confluent has solved that problem with the help of TableFlow. You read it, you write it once and the open format in Confluent platform and that same data can be read in your choice of lake houses. It is a format which is globally standardized called as Iceberg. So as part of Iceberg format, you can write the data in Confluent once, read it in any of the lake houses or any other systems you have. This is how we bring down the cost of your compute in a big way.
Another aspect is there might be certain use cases in AI where enterprises need data in milliseconds – like in case of a financial fraud, or in healthcare where it is about a patient’s life. But if one has to send a promotion out to a customer, that promotion can wait for one minute. For these one-minute use cases, we have a low-cost solution also which can happen at scale with the same kind of efficiency. This is called a warp stream and also freight clusters. Therefore, Confluent has solutions and offerings for our customers for various use cases.
Many of your customers would be concerned with how their AI agents behave and how agentic AI would impact their customer satisfaction quotient. How can you help your customers improve or enhance their customer satisfaction?
We are engaging with so many chatbots on so many platforms. What happens in most of those platforms are that these chatbots or agents are not able to understand the customer problem or solve the issue. Subsequently the customer experience goes down. This problem is happening because these agents do not have the real time context.
There is no proper governance because these agents are naturally built in such a way that they try to look into everything and solve a problem. But if you have governance layer on top of it, for a type of problem, you can train the agent to only look at those 2-3 specific areas. And then the agent has to decide when to involve the human.
With the sentimental analysis of the responses from the user I am talking to as an agent, I should be trained to involve a human when I am not able to solve. And for a criticality of the use case, these are the downstream applications or data sources I should be looking at and only get the response from there rather than taking too much of time. With the fresh context with the MCP servers of Confluent, we are able to solve that problem.
How does Confluent solution address the challenge of data fragmentation and data homogeneity in enterprises so that AI can be leveraged more effectively for more accurate insights?
We have what we call the data streaming platform, which is underpinned by a number of technologies. But in the context of data fragmentation, one of the technologies is an open-source project called Kafka. This is underneath the data streaming platform while the other technology is called Flink.
And then on top of that, we have built a number of capabilities. Confluent Intelligence handles data fragmentation occurs because the data is sitting in silos across many different systems in many formats — structured, unstructured, semi-structured.
The traditional way was to have a data warehouse or a data lake, and put everything into the data warehouse or data lake. As things have modernized over time, these two came together. The fundamental issue with data fragmentation still exists — being able to deal with data that is in motion or real time does not get captured by those two capabilities efficiently. Also the whole process is very expensive.
With Kafka you can create a topic, and that topic can have context around all the different data, and many different data sources can fill that one topic. And if you know that topic needs to be consumed by maybe different agents with different capabilities, they can take what they want.
What are the key pillars of Confluent’s GTM strategy and what are the key initiatives undertaken as part of the strategy?
Partners are a critical part of our go-to-market strategy. We work with all of the major cloud partners, such as Google Cloud, AWS, and Azure, local partners like Jio Platforms in India, and global SIs like Infosys or even the Big 4s like Deloitte. We collaborate with global systems integrators (GSIs), Indian GSIs, and regional partners who bring domain expertise in each area to deliver the best solutions for our customers.
Additionally, we have a direct sales team that works closely with our partners on joint sales efforts. Now with IBM coming in between, things have become even smoother. IBM itself has very big relationships with the likes of AWS, GCP, Azure — all the three large cloud providers globally. And this is only strengthening our go-to-market with our customers with a lot of other IBM products also run on these clouds. So now from a GTM perspective, it is a better coherence and better coordination with the large enterprise landscapes.
What are Confluent’s unique differentiators over other AI competitors?
Confluent helps customers leverage AI for real time insights. For any AI project, you need to use real-time data that is fed into large language models, small language models, or AI agents. These can either be built internally by businesses or made using pre-built AI solutions. Confluent plays a foundational role in the data layer which is like a composable, building-block model.
In this block model, Confluent is the underlying layer that can help businesses stream, process, connect, and govern data. On top of that, like building blocks, you can add agentic AI solutions—whether for industry-specific verticals, front-office customer interactions, middle- office operations, or back-office logistics and delivery.
