If you work in marketing technology and feel like the ground won't stop moving, Scott Brinker has a name for it: the chrysalis, the primordial soup that transforms caterpillars into butterflies.
The creator of the Martech landscape and former VP of Platform Ecosystem at HubSpot joined Jacqueline Freedman on Making Sense of Martech to pressure-test the 2026 State of Martech Report, and what emerged was one of the most honest reckonings with AI in marketing published this year.
The headline finding: AI is everywhere and integrated nowhere. And the organizations holding it all together are marketing ops teams that never signed up to be system administrators.
Start With the Customer. Not the Technology.
Before any conversation about AI strategy, stack architecture, or vendor consolidation, Brinker makes one thing clear: do not start with the technology.
"Marketing always needs to start with the customer, the business, the value proposition. The technology is much, much lower down the ladder of things we might implement to execute on that."
It sounds obvious. It is routinely ignored. Freedman has a name for what happens instead: Shiny Object Syndrome. "You do SOS," and Brinker notes, "you end up SOL."
The organizations making real progress right now are the ones that have maintained strategic clarity about who their best customers are, what those customers actually want, and how the business makes money serving them. Everything else, including the AI strategy, follows from that.
The Chrysalis Metaphor: You're Not Behind. Everyone Is.
The central metaphor of the 2026 State of Martech Report is the chrysalis — the messy middle stage where the caterpillar has dissolved, but the butterfly hasn't yet formed.
Brinker is deliberate about why: "If you were to cut it open in those early stages, you would have no hint that there was actually going to be this beautiful butterfly coming out."
That's where the industry is. AI is embedded in nearly every platform, pitch deck, and vendor conversation. What hasn't arrived is integration. The governance, taxonomy, and data infrastructure needed to support AI at scale are lagging behind the deployment of AI itself, and that gap is not temporary. It is the condition.
The practical takeaway: the feeling that you're behind, that everyone else is running 5,000 agents while you're still untangling your CRM, is not a reflection of your organization's failure. It is the actual state of the industry. "Let's just give ourselves permission that this is the actual state we're all in," Brinker says. "We've got further to go than we've already covered."
The Landscape Has Hit a Historic Plateau, and It's Going to Bifurcate
After fifteen years of relentless expansion, the Martech landscape grew by less than 0.1% in 2026. For an industry conditioned to exponential growth, that number demands interpretation.
Brinker's read: bifurcation, not peak.
The stack is splitting into two distinct trajectories. Core infrastructure, data warehouses, major CRM and MAP platforms, and orchestration layers are consolidating. The evidence is already clear: Salesforce's headless announcement and HubSpot's commitment to full programmatic access are both signals that incumbents are adapting to a world where the UI is decoupled from the backend. Neither move was a strategic preference. Both were strategic realities.
Above that infrastructure layer, Brinker predicts an explosion of ephemeral, app-like software, tools built fast, used well, and replaced without ceremony. His analogy is clarifying: there are approximately 450,000 agencies worldwide, and marketers navigate that infinite choice without setting their hair on fire. Software, he argues, is heading in the same direction. Durability is no longer the value proposition. Utility is.
Content marketing led all subcategories in net removals from the landscape this year, and it's the clearest early signal of what that bifurcation looks like in practice. First-wave AI-native tools built solely on content generation couldn't survive once foundation models commoditized that core feature. What remains, and what will continue to matter, is context. As Brinker puts it: "Content was king. Now, the context is the monarch."
Pace Layering: How to Make Stack Decisions When the Ground Won't Stop Moving
The most actionable framework in the conversation is pace layering, the idea that different components of the stack change at fundamentally different rates and should be evaluated accordingly.
Infrastructure decisions, cloud data warehouses, and core CRM platforms warrant deep evaluation and long holding periods. Experimental tools at the top of the stack should be trialed fast and replaced without sentiment. The evaluation criterion that matters most across the board, according to Brinker, is: "How well does it set you up to support change?"
Optionality is not indecision. It is the leverage that makes future decisions feasible without requiring, as Brinker puts it, an act of Congress.
The governance dimension of MCP is the most underappreciated corollary. Nearly 30,000 MCP servers have emerged in 18 months, a pace that took 15 years to achieve in the Martech landscape itself. The openness that makes MCP powerful is also what makes it risky. Bad actors and incompetent actors can do equal damage in an unvetted ecosystem. The platforms that build governed, curated ecosystems around MCP will have a structural advantage over those that don't.
Marketing Ops Is Becoming What It Always Wanted to Be
The 2026 report describes marketing ops shifting from system administrator to context engineer. Brinker is clear that this is not a threat to the profession; it is the opportunity it has been waiting for.
"Most people in marketing ops didn't really aspire to be system administrators. The aspirations were always: how do we transform how marketing works? How do we bring the value we deliver to the company to a higher level?"
The practitioners who invest now in learning to orchestrate AI, build data infrastructure, and develop governance frameworks and taxonomies that make AI outputs reliable are building the most defensible skill set in the stack. Marketing ops is not being replaced by this transition. It is becoming the foundation upon which everything else gets built.
The Stack Is Stratifying, and Nobody Has Won Yet
The battle for the center of gravity in the modern Martech stack is the most consequential strategic question in the industry right now, and Brinker is honest that the outcome is genuinely unclear.
Salesforce and HubSpot are claiming the orchestration layer through platform openness. Data warehouse companies like Snowflake and Databricks are moving up the stack with compute, semantic layers, and agent governance. Frontier AI Labs believes it should be the one coordinating what it built. Independent startups are pitching themselves as neutral orchestrators across all of the above.
"Unlike the previous generation of martech, where competitors were very apples to apples, you've now got apples, bananas, mangoes, and oranges all coming towards the same conclusion." Brinker's game-theoretic read: none of them may ultimately win outright, which means the pressure falls squarely on the brands and operators making stack decisions in the middle of the collision.
The answer, as always, comes back to value engineering. Does the stack deliver the right capabilities for your customers and your business? Any architecture that does that is the right one, regardless of which layer ultimately wins.
What This Means for Martech Operators
Three things to take from this conversation:
Start with the customer, not the tool. Stack decisions made under AI pressure without strategic clarity about business value will cost more to unwind than they saved to implement. SOS (Shiny Object Syndrome) leads to SOL (Sh*t Out of Luck).
The chrysalis is the strategy. Accepting the messy middle: AI everywhere, integrated nowhere, is not a failure posture. It is the accurate read of where the industry is. The organizations that build governance and infrastructure now will be the ones that emerge with stacks that actually work.
Design for change, not for certainty. Pace layering, composability, and strategic optionality are not signs of indecision. They are the architecture that makes future decisions possible without requiring heroic effort or executive approval for every experiment.
Special thanks to Claude for helping to summarize this conversation.
