The last decade of data infrastructure was built around pipelines. Move the data, transform it, store it, then do something useful with it. That approach became so normal that most teams stopped questioning it.
At Cloud Next 2026, Google seemed to be leaning toward a different way of working with data, one built more around direct access, unified platforms, and AI-powered execution. The bigger idea was not better pipelines – but fewer of them.
Most data systems rely on a chain of scheduled jobs and transformations. Each step depends on the previous one running correctly. As systems grow, these dependencies increase, making pipelines harder to manage and more expensive to run.
Data is ingested, transformed, stored, and then used for analytics or downstream systems. This approach still works and remains widely used, but over time it has created layers of ETL, ELT, and orchestration that add complexity as scale increases.
This is the problem Google is starting to design around. At this year’s event, BigQuery was presented as the place where processing and AI run on the same data. It’s not about where data lands at the end, but also about where models interact with live datasets. That removes a lot of the back and forth between systems that usually sits in the middle.
It also changes where transformation work happens. Instead of pushing data into other tools and then bringing it back, more of that work can stay inside BigQuery. According to Google, this translates to fewer transfers, fewer scheduled jobs, and less pipeline logic to maintain. No doubt that the pipelines still exist, but they are not needed for every step between raw data and actual use.
The lakehouse announcements point in a similar direction – data should not have to move every time a different tool needs it. At the event, Google introduced a cross cloud lakehouse built around Apache Iceberg, with support across services like BigQuery and Spark. The goal with this is to let multiple systems work on the same data without creating new copies each time.
Google expects that to cut down on the constant replication most pipelines exist to support. What this means is that instead of building more ingestion flows, the platform is trying to make those flows less necessary – aligned with the overall theme of data announcements at the event.
Google is taking a similar approach to AlloyDB. It added federation so AlloyDB can query data in BigQuery and the lakehouse directly. It also supports combining operational and analytical data without pulling everything into one system first. This replaces the common pipeline pattern of copying data, syncing it, then querying it. Here, the data stays where it is, and the query moves instead. This means fewer steps, fewer copies, and less pipeline work overall.
A lot of pipeline complexity does not come from moving data. It comes from keeping track of what that data actually means – the context. This includes definitions, structure, and lineage, all of that tends to get recreated across systems. That is where things often start to break.
(AI-generated/Shutterstock)
At the event, Dataplex was pushed further into Knowledge Catalog, with the goal of keeping metadata and business meaning closer to the data itself. Instead of rebuilding that context in every tool, it is meant to live in one place. That also makes it easier to apply consistent policies and governance rules without duplicating logic across multiple systems.
When you line this up with the rest of the announcements, the direction is hard to miss. BigQuery becomes the execution layer. The lakehouse becomes the shared data layer. Federation removes the need for constant copying. Governance holds the context together. Different pieces, same outcome, less duplication, less movement, and less orchestration. It also reduces the number of failure points, since fewer systems are involved in moving and reshaping data before it is used.
Pipelines still add value. There are too many systems that depend on them, especially where control and predictability matter. Batch processing, scheduled reporting, and regulated workflows will continue to rely on structured data movement. Those use cases are not being replaced. Data stays where it is more often. Work moves closer to it. Pipelines are still there, just a little less stressed. Over time, that changes where teams spend effort, less on moving data around, more on working directly with it.
If you want to read more stories like this and stay ahead of the curve in data and AI, subscribe to BigDataWire and follow us on LinkedIn. We deliver the insights, reporting, and breakthroughs that define the next era of technology.
The post From Pipelines to Platforms: Google’s Data Strategy at Cloud Next 2026 appeared first on BigDATAwire.
Go to Source
Author: Ali Azhar
