What you need to know:

• Martech isn’t consolidating, it’s combusting, according to the latest State of Martech report from Scott Brinker and Frans Riemersma. The long-promised rationalisation has become a mass extinction, with generative AI and agentic automation seeding a chaotic rebuild from the ashes.
• AI agents are replacing old integration tools. Scott Brinker says AI agents now serve as the new iPaaS (integration platform as a service), stitching together workflows across apps, data sources and platforms – both internal and external – at speed and scale.
• Customer data platforms are being dethroned and bypassed as AI pushes cloud data warehouses to become the new core of B2C marketing architectures.
• Churn is up, and growth still strong. Despite 1,211 martech products being removed – 774 of them pre-ChatGPT – the landscape still grew 9 per cent to 15,384 solutions. No single "hot category"; AI is fuelling broad-based innovation.
• Generative AI is becoming embedded. While 87.5 per cent of marketers now use generative tools, only 21 per cent report deep AI integration across workflows. The transition from experimentation to embedded architecture has begun..
• Anthropic’s MCP is the new backbone. The Model Context Protocol, now adopted by Google and OpenAI, is emerging as foundational infrastructure for agentic communication, tool use, and data access across stacks.
• Already, billions of apps are blinking in and out of existence. AI assistants like ChatGPT and Claude are quietly spawning temporary, task-specific software at scale – transforming digital operations with ephemeral, invisible infrastructure.
• Governance is still catching up. Agent permissions, observability, and compliance controls remain critical gaps as marketers enter a world of autonomous workflows and emergent behaviours from AI agent networks.

Martech isn’t simply transforming. It’s blowing it up. Fifteen years of stack-building, platform-wrangling, and category-defining looks like its about to be blasted apart by an invisible army of generative agents, LLMs, and “instant software” – apps so lightweight and fast they blink in and out of existence as required. That's our take on the State of Martech 2025 report by Scott Brinker and Frans Riemersma which was released last week.

The long-anticipated consolidation of martech is real, but it resembles less an orderly changing of the guards and more the beginning of a mass extinction event followed by an AI-driven Cambrian explosion. The stack isn’t shrinking. It’s mutating. And for marketers, that’s both exhilarating and existential.

"In many ways, AI is already changing SaaS apps, especially in martech. Three ways in which it seems clear that AI, particularly generative AI, will change these apps," Brinker, editor in chief of Chiefmartec, told Mi3.

"They'll have conversational user interfaces that make it easier for marketers to build and analyse content, campaigns, programs, and experiences. As martech got more sophisticated, it also got harder for people to learn how to use all of its capabilities. By being able to describe what they want in natural language, and let the software translate that into work it does, will be a big democratisation of the power of martech tools."

Next, Brinker believes martech is going to be very good at interpreting unstructured data such as calls, emails, forum messages, RFPs, feedback forms, and chat transcripts. That means it will be better equipped to uncover insights for marketers and adapt more personalised communications with customers. "This has been a huge, untapped sea of possibilities that is just now opening up."

Thirdly, he says, "In many ways, AI agents are the new iPaaS (Integration platform as a service – basically the way to make applications work together). They're going to make it easier for marketers to think of jobs-to-be-done – or, at a simpler level, tasks-to-be-done, and then the agent will be able to work across multiple apps and data sources, within the company's tech stack but also out on the internet at large, to execute on it."

He believes the speed by which Anthropic's MCP standard has caught on is a clear sign that integration in the age of AI agents is going to be much more fluid than the previous generation of martech.

According to Brinker, "When evaluating SaaS apps, the most important thing for CMOs is that the products and platforms they purchase are open, making it easy to share data and services with AI functionality, whether it's in their product or someone else's."

Darwinian economics

At the beginning of the last decade when Scott Brinker started mapping the marketing technology landscape, it was a scrappy little graphic with 150 logos. This year, it swelled to 15,384. A year ago Brinker told Mi3 the point would eventually come when it will be meaningless to try and count the apps.

That said, an interesting trend in the report is the arrival of a much-anticipated period of consolidation, although Brinker and Riemersma prefer to describe it as a "period of renewal".

Per Brinker: "The previous generation of martech products, especially those launched between 2010-2020, are definitely consolidating in the market. I wouldn't necessarily use the word 'weaker' or even underfunded – a lot of martech tools have been organically grown. But, yes, in head-to-head competition in their category, we're finally seeing a separation between winners and also-rans."

There's been a blossoming of thousands of new AI-native startups these past couple of years that are just beginning their journey, he says.

"To be sure, not all of them will survive either – and some might exit faster than we'd seen in the previous era, just given the increased tempo in our AI era. But some, maybe even many, will grow over the next few years. I expect we will see new AI-winners in all categories, as well as AI-rebirths of many of the leading incumbents."

Longtail goes hyper

Welcome to the hypertail – a digital ecosystem where AI isn’t just a feature, it’s a force of nature. And according to Brinker and Riemersma's report, that force is reshaping everything: stacks, strategies, and even the shape of the companies that populate the map.

Still, something is clearly dying. The report details how 1,211 products were removed from the martech landscape this year – a churn rate of 8.6 per cent. But here’s the kicker: “The majority of the products removed – 774 – were born before the ChatGPT inflection point of November 2022.” AI didn’t just make new things possible, it made many old things irrelevant.

An emerging case in point is the customer data platform (CDP). Once the belle of the martech ball – with Australia only recently described as being "on a CDP bender" – CDPs are now being absorbed or bypassed.

The consolidation that happened with CDPs this past year is probably the most talked about example, the authors note, with companies like ActionIQ, Lytics, and mParticle either acquired or diminished. The centre of gravity is shifting – up to cloud data warehouses or down into CEPs. Either way, the once-dominant CDP is being supercharged by AI or quietly sunsetted by the market.

Growth everywhere, and nowhere in particular

Despite the churn, the martech map is growing – albeit more slowly. After last year’s 28 per cent growth, 2025’s rise of 9 per cent might seem tame. But don’t be fooled. “None of them shrank. And the fact that they all grew by about 7-10 per cent is testament to the fact that startups are experimenting in every corner of marketing technology. There’s not one 'hot category'. The whole field is hot," they wrote.

Sales automation, content marketing, and ecommerce remain the busiest zones of innovation, thanks largely to the plug-and-play nature of generative AI. And although a few subcategories like ABM (account-based marketing) and native advertising saw contraction, the overall signal is clear: the martech fire still burns.

Interestingly, smaller martech firms – those between 11–50 employees – earned the highest satisfaction ratings from users, according to G2 data in the report. That reflects a broader trend: bootstrapped or modestly funded startups are thriving because “The truth is you don’t necessarily need a lot of capital to build something of value in this environment, and AI is only making that more true every month now,” the report notes.

AI morphs from experimentation to embedded architecture

The report’s second act is all AI, all the time. And it makes a clear distinction between early-stage usage – typified by standalone tools like ChatGPT – and deeper stack integration.

“87.5 per cent of our respondents reported they use them in their marketing now. That crossed the chasm fast, eh?” per Brinker and Riemersma.

But true transformation is just beginning. Only 21 per cent of companies reported broad or full AI integration across workflows. And while tools like ChatGPT, Claude, and Copilot are widely used, only 59 per cent of companies with a cloud data warehouse are using AI directly on that data. The implication? There’s still a massive runway ahead.

Yet the architecture is bending and new standards like Anthropic’s Model Context Protocol (MCP), already embraced by Google and OpenAI, are accelerating that journey. Think of MCP as the connective tissue of tomorrow’s martech because it will allow AI agents to communicate with each other, use software tools, and reference external data sources.

What’s especially telling is how AI is being used today. Content production (78.1 per cent), audience segmentation (42.7 per cent), and data analysis (38.5 per cent) lead the way. But as AI creeps into orchestration, analytics, and customer experience, the stack begins to shift shape – not just in its tools, but in its centre of gravity.

“The rise of the cloud data warehouse is remarkable,” the authors wrote. For B2C marketers, it has already overtaken CDPs and CRM as the central organising hub. That’s a fundamental re-architecture.

If you think martech has peaked, think again since what Brinker and Riemersma are suggesting is not the profile of a maturing market, but an exploding one.

"And it gets wilder. Popular AI assistants such as ChatGPT, Claude, and Gemini – that have millions of non-technical subscribers – are often creating software programs behind the scenes to do the bidding of their users without those users even knowing that software was built and run on their behalf."

The result, they say, is "Not millions but billions of custom software programs proliferating companies’ digital operations. Probably trillions, many blinking in and out of existence on demand".

Work to be done

The future outlined by the authors needs to be viewed with an important caveat. Despite the rise of MCP there is still much work to be done putting in place the basic scaffolding required to sustain a global agentic martech ecosystem.

On this point Brinker told Mi3: "Everything that was important for managing human users and manually configured automations applies in an agentic ecosystem. It's just happening at a larger scale".

He believes it's best to conceive of an agent as a user. "It needs to be granted specific permissions for what it can and can't access, what it can and can't do. Same with a more always-on agentic workflow. Those permissions for things like data access need to be controlled through your core platforms that are enforcing regulatory compliance."

He also notes that monitoring and observability will be important.

"That's probably where we need vendors to make more investment. We're going to have a lot of cases of emergent behaviours, where AI agents – or networks of AI agents – start doing things that weren't necessarily breaking any of the rules they were given, but it also wasn't what we expected them to do. In many cases, this could actually be a good thing."

Though probably not in all cases. It's a system of continuous combinatorial innovation, says Brinker. "But we need to watch it closely."