5 customer data predictions for 2026.

Before we get started, let’s see how I did with my customer data predictions for 2025.

2025 prediction review

Prediction 1: Data warehouses become the CDP.

This one was pretty easy. No outside companies are building on (which defines the term platform for me) the CDP. The main use cases for CDPs seem to be drag-and-drop synching for table builds in the data warehouse, SQL queries, and activating IDs into channels. Meanwhile the gravity well for first-party customer data has become Snowflake, Databricks, and GCP — where it can be managed and governed.

It’s probably the fourth or fifth inning of this cloud data warehouse migration, so strong growth will continue for all. Of course, certain people and companies will continue to use the term CDP. There’s been too much equity invested in the acronym by investors and consultants over the last decade for it to disappear all at once. But the writing on the wall is clear: for enterprise brands that need a platform for their data to solve business use cases and inform AI, the data warehouse is that platform. I’ll give myself a 5 / 5.

Prediction 2: CDPs become other things.

This prediction was all-or-nothing with the one above, but better exemplifies both. Growth Loop became a Compound Marketing Engine. BlueConic is now a Customer Growth Engine. Simon Data became Simon AI, an Agentic Marketing Platform. Amperity became a Customer Data Cloud. Bellwether composable CDP Hightouch does not use the term CDP on its homepage at all, but mentions AI three times. Others like Segment, Tealium, and mParticle still cling to their CDP identity — now “powered by AI.” There is clearly an identity crisis with CDPs. This one I nailed. 5 / 5.

Prediction 3: Snowflake and / or Databricks gets acquired.

This was a big whiff, but the day Databricks raised at a $134B valuation (now almost 2× SNOW) a CNBC analyst mentioned on-air that Snowflake is a potential acquisition target. Still, it didn’t happen in 2025. 0 / 5.

Prediction 4: Open source starts to make an impact on martech and adtech.

The big news in adtech was — and is — AdCP (Ad Context Protocol) being announced and hyped up this year as a standard. Right now the “impact” of it has been negligible except for the podcast and newsletter circuit. Qwen is also an important part of adtech and martech development that people like to keep a secret. Since almost all of adtech seems to be hanging its hat on AdCP for its AI future, and since I couched my prediction with starts, I’ll give myself 5 / 5.

Prediction 5: Martech or adtech will have its Klarna moment with AI.

I’m not happy about this one. There have been a lot of people who have lost their jobs the past year in advertising and marketing. Whether AI is the cause or the excuse feels unimportant. What is clear is we are in uncharted waters. Amazon itself cut 14,000 corporate roles. That’s before we even get into agency and consultant shrinkage. In retrospect I’m sorry I even made this prediction. No score.

2026 customer data predictions

Prediction 1: Snowflake will be purchased or merge.

I think I was a year early on this one, so I’m keeping it on the board. And since Databricks is now approaching $200B valuation, anything but an IPO seems implausible for them. That said, I think Snowflake is primed to be acquired by Google, AWS, Microsoft, Salesforce, or Oracle — or merged with a company like Adobe or ServiceNow. Oddly, the data in Snowflake seems to be worth more to others than what Snowflake is worth to the market. Something’s got to give.

Snowflake themselves needs to keep buying companies like they just did with Observe and focus on integration if they want to get the data + AI zeitgeist back from Databricks. Obviously any purchase would face regulatory hurdles, but the current environment is one where it could get approved.

Prediction 2: Agents will disappoint us into the trough of AI disillusionment.

Everybody and their mother is selling agents. Single agents, multi-agents, agent protocols, agent frameworks. Agents are perceived by many to be the low-hanging fruit of AI rewards, and we are in the peak of inflated expectations.

Agents right now don’t have agency.

I’ve been vocal about my concerns here. The idea of humans in the loop is being used to hide deficiencies and runs counter to agent self-learning, where the real value is unlocked.

For the enterprise there are major issues around how their first-party data is being used for training and improving agentic systems — many of which are not deployed locally but operate off downstream sub-processors. I took a quick look for Trust Centers for a number of the CDPs offering agents and I couldn’t find them. Those I did find were not addressing what models and infrastructure they are using for their agents. Not a way to instil confidence in an early market.

Expect a lot of bad press around agents (and possible data breaches) until mid-2027 — when advertising and marketing agents from the AI platforms themselves take us to the slope of enlightenment.

Prediction 3: Adobe makes a big acquisition.

Adobe got really screwed by the former FTC chair not allowing the Figma acquisition (whose shareholders also got screwed). Adobe built its entire marketing cloud — and pretty much its entire company — through acquisitions. While they still have some runway, clearly Adobe is no longer a future-forward technology company as the world moves from software subscriptions to consumption-based compute and AI.

Making matters worse, acquisition targets they might have bought a year or two ago — like Anthropic — are now worth more than Adobe. I know there is a ton of pressure on sales to sell products that are not that compelling anymore and require high service levels, like CJA and Real-Time CDP. Shantanu Narayen has been CEO since 2007, so he engineered their amazing 15-year run of increased shareholder value through acquisitions. In 2026 he will dust off the old playbooks, try to stop the bleeding, and find products relevant to new and existing customers.

Prediction 4: Google gets more cloudy.

This feels like a softball, but I think GCP / BigQuery does not get enough attention. The gap between how Google operates and the rest of adtech keeps widening. Google is currently rolling out AI Max. Soon you will not need keywords, you will not need creatives — you will only need a product, an ad budget, and a CPA goal. AI Max will do the rest.

This is all about customer data and GCP / BigQuery. BigQuery is being positioned as an execution plane for Google Advertising, providing advantages with activation, measurement, and optimisation that they will maintain to keep growing GCP (and Alphabet and YouTube and Waymo and everything else).

This leads to the last prediction.

Prediction 5: Multi-cloud has its moment.

Last year one of Snowflake’s largest customers started using Databricks for certain jobs, and sharing data back and forth via Lakehouse Federation. This is a strong sign that advances architecturally in moving data across clouds is one of the most important things that has been happening under the covers the past couple of years.

Unlike the walled gardens of customer attention, the tunnels that allow customer data motion are critical components to accruing additional value and revenue for businesses of all sorts (did I hear someone say “clean room”?). Each cloud has certain advantages for certain data formats, workflows, and processes.

What we saw in 2025 is the clouds become less competitors and more partners as they discover that data and compute is not a zero-sum game. For brands there are many advantages to multi-cloud infrastructure. For the clouds, ingress and egress is highly profitable — especially when it compounds with AI.

In 2026 there will be much more playing nice across clouds than the cut-throat competitiveness that defined their rise.

Well, back to work. We’ll check back in a year and see how I did. Hope you make it a great year for customer data.