Data intelligence consolidated while everyone was watching the models

Data intelligence consolidated while everyone was watching the models

For two years the interesting news in enterprise data was happening one layer up. Model releases got the coverage, the benchmarks got the arguments, and the plumbing underneath — the tools that move data, describe it, clean it and decide who may see it — carried on being treated as solved infrastructure that somebody in the platform team already owned.

It was not solved, and over roughly seven months it stopped being independent. Three deals reshaped the middle of the stack, and by the time the second one closed the shape of the market a buyer was choosing from had changed.

What actually changed hands

Three transactions, all reported by TechTarget in its survey of what is shaping data management in 2026:

Salesforce acquired Informatica. Announced in May 2025 and closed in November 2025. Informatica is not a niche product — it is one of the oldest and most widely deployed data-integration and master-data-management estates in the enterprise, the kind of software that sits underneath a decade of pipelines.

Fivetran and dbt Labs agreed to merge, in October 2025. Those two occupy adjacent halves of the modern analytics workflow: Fivetran moves data into the warehouse, dbt transforms it once it lands. A very large share of teams that built an analytics stack in the last five years run both.

IBM acquired Confluent, in December 2025. Confluent is the commercial home of Apache Kafka, which is to event streaming roughly what Postgres is to relational storage — not the only option, but the one an architecture diagram assumes.

Each deal is defensible on its own terms. Read together they describe something different from three companies changing owners: ingestion, transformation, streaming and master data all moved inside larger platform vendors inside seven months.

Why a buyer feels this as risk rather than savings

The pitch for consolidation is always the same, and it is not dishonest. Fewer vendors means fewer contracts, fewer integrations to maintain, fewer support relationships, one throat to choke. Organisations really do carry too many overlapping data tools, and the cost of that sprawl is real.

The problem is who absorbs the transition. A buyer who chose Informatica in 2019 chose a company whose incentives were to integrate with everything, because integrating with everything was the product. The same software inside a platform vendor has a second set of incentives layered on top, and they point toward that vendor's other products. Nothing has to be done cynically for the effect to appear: roadmaps get prioritised, and the integrations that get the attention are the ones that help the parent.

There is evidence that buyers have already internalised this as their problem. Gartner's 2025 survey of Chief Data and Analytics Officers found that one in two CDAOs now considers optimising the technology landscape a primary responsibility. That is a striking way for a role to be defined. The senior data person's job used to be described in terms of what the organisation could learn from its data; increasingly it is described in terms of managing the estate of tools that data passes through.

The spending backdrop makes the stakes concrete. IDC puts global data and analytics spending approaching $420 billion by 2026. That is the size of the market being reorganised around fewer owners.

And the reorganisation is happening while the underlying data is not ready for what is being asked of it. Gartner reports 57% of organisations saying their data is not AI-ready. Consolidation is arriving at the moment the majority of buyers have an unfinished data problem — which is the worst possible moment to be changing vendor relationships, and the best possible moment to be sold a platform that promises to make the problem go away.

The counter-move: consolidating onto standards instead of suites

The interesting part of 2026 is that the market did not simply accept the suite answer. The counter-argument is that consolidation is fine as long as what you consolidate onto is an open format rather than a vendor.

That argument has teeth now because the table formats underneath the modern warehouse are genuinely open, and because the layer above them is converging on protocols rather than products. The debate in agent tooling is a good example of how fast that convergence gets decided. Donald Farmer of TreeHive Strategy is sceptical that every proposed protocol survives: "A2A has a tougher road ahead. The protocol is only needed when organizations actually run agent swarms that need to work together." Chris Aberger of Alation goes further on where it lands: "MCP won't coexist with A2A. It will absorb the useful parts because the ecosystem will demand one standard."

Whether or not that specific prediction holds, the framing is the useful part for a buyer. A market consolidating onto a standard leaves you able to change vendor. A market consolidating onto a suite does not. Those are different futures wearing the same word.

What our own catalogue shows about the long tail

TrustList's directory is not a market census, and the numbers below should be read as what a directory of vendors looks like rather than as a measurement of industry structure. With that caveat, the shape is informative.

The broad category — companies listing themselves as big-data and business- intelligence providers — carries 767 live listings. That is one of the largest categories on the site, and it has not thinned out.

The specialist categories underneath it are much smaller, and they map almost exactly onto the functions that just changed hands:

• ETL software: 20 live listings
• Data mining software: 20
• Data warehouse software: 18
• Customer data platform software: 16
• Predictive analytics software: 16
• Embedded analytics software: 15
• Data quality software: 14
• Master data management software: 14

Two things follow. First, the long tail is genuinely long — hundreds of firms still sell into this space, and consolidation at the top has not removed the option of choosing a specialist. Second, the specialist categories are thin enough that losing two or three independent vendors from any one of them changes a buyer's shortlist materially. A category with fourteen credible entrants is one acquisition away from having eleven.

That is the practical shape of the risk. It is not that there will be nobody left to buy from. It is that the specific alternative you were relying on as leverage in a renewal may not be independent by the time the renewal comes round.

There is a second reading of the same numbers that cuts the other way, and it is worth stating because it is the more optimistic one. A category with fourteen or twenty independent entrants is a category where the problem is still considered worth solving on its own terms. Data quality and master data management have both been declared solved-and-absorbed several times over two decades, and both still support a working set of specialists. The long tail persisting through a consolidation wave is evidence that the suite answer has not actually closed the gap it claims to close.

What to ask a vendor now

The questions that mattered in 2023 were about capability. The questions that matter now are about what happens to you if the vendor is bought, or if you want to leave.

Where does my data physically live, and in what format? If the answer is an open table format in storage you control, a change of vendor is a migration of compute. If the answer is a proprietary internal format, it is a rebuild.

What is the documented export path? Not whether an export exists — every vendor says yes — but whether the exported artefact includes the transformations, the lineage and the access rules, or only the rows.

Which integrations are contractual and which are goodwill? After an acquisition, goodwill integrations are the first thing to lose engineering attention.

If this product is acquired, what happens to my contract? Most buyers discover the answer after the announcement.

None of that is new advice. What is new is that the odds of needing it went up three times in seven months, in a market where the majority of buyers say their data is not ready for what they are trying to do with it.

The forecast underneath all of this

Consolidation at this particular moment is not a coincidence, and two Gartner projections explain the timing better than any deal rationale does.

The first: by 2027, 60% of data management tasks will be automated. The second: 75% of new data flows will come from non-technical users. Gartner also reports 71% of organisations planning to invest in data management technologies embedding generative AI within the next two to three years.

Put those together and the product being acquired is changing shape. For twenty years, data integration software was sold to a specialist who operated it. If a majority of the tasks that specialist performs are automated, and three quarters of new data flows originate with people who have never opened the tool, then the thing being bought is no longer an instrument for an expert. It is infrastructure that has to be safe in the hands of people who will never read its documentation.

That is a platform-vendor product, not a point-tool product. It needs the governance layer, the identity layer and the catalogue to be part of the same system, because there is no longer a skilled operator in the middle catching mistakes. Whatever else these acquisitions were, they were bets that the buyer of data infrastructure in 2028 is a different person from the buyer in 2018.

What consolidation asks of your team

The consequence that lands on staff is easy to miss in the deal coverage.

If 60% of the tasks are automated, the remaining 40% are not the easy ones. They are the judgement calls — deciding what a metric means, deciding who may see which rows, deciding whether an anomaly is a data error or a business event. Those are exactly the tasks that do not automate, and they are the ones a consolidated platform makes more consequential rather than less, because the platform will apply the answer everywhere at once.

This is the practical reading of Gartner's CDAO finding. When one in two senior data leaders describes optimising the technology landscape as a primary responsibility, the role has partly become vendor management. That is a rational response to a market where the leverage available to a buyer is decided by architecture choices made years earlier, but it is worth naming, because it is not what most data teams were hired to do and not what most are structured for.

The teams that will come through the next two years in the best shape are probably not the ones that pick the right platform. They are the ones that keep their definitions, their access rules and their lineage in a form they could hand to a different platform — because that is the only thing that converts a vendor decision from a marriage back into a purchase.

The questions that changed

The questions that mattered in 2023 were about capability: can it handle our volume, does it support our sources, how fast is the sync. Those are now mostly answered by any credible vendor.

The questions that matter now are about exit. Not because you plan to leave, but because the ability to leave is what determines the terms you get if you stay. In a market that has consolidated three times in seven months, the price of a renewal is set by how credible your alternative is — and a credible alternative requires that your definitions and your data are portable in practice, not just in a feature list.

Michael Ni of Constellation Research summarised the year's shift in emphasis neatly: "2025 was about building agents. 2026 is about trusting them." Trust is mostly a property of the plumbing. The plumbing just changed owners.

Sources

• TechTarget — 4 trends that will shape data management and AI in 2026 (the three acquisitions and their dates; quotes from Michael Ni, Donald Farmer and Chris Aberger)
• N-iX — Data management trends in 2026 (Gartner: 57% of organisations report data is not AI-ready; Gartner 2025 CDAO survey; IDC data and analytics spending approaching $420B by 2026)
• TrustList's own catalogue, queried 8 September 2026, for the live listing counts by category.