Last month, I wrote about the shift of martech platforms evolving from applications into infrastructure.
To point: Salesforce released its fully headless platform. HubSpot committed to full API parity with its UI to let agents run HubSpot. Many other vendors have launched MCP interfaces — Braze, Hightouch, Knak, Marketo, MoEngage, Pega, SAS, Treasure AI, etc. — and are expanding the functionality provided through them.
Collectively, they’ve realized that resistance is futile. As much as they would prefer customers to live within their UI and use only their own brand of AI agents, the market has spoken. Customers now want demand the freedom to use Claude, ChatGPT, and independent AI agents of their own to tailor their marketing operations and experiences, outside the constraints of those suite-platforms that once dictated their destiny.
Yet marketers still value these platforms. They remain the center of gravity for how most of marketing runs, providing structured and stable data models, consistent and persistent workflows, well-governed rules and permissions, and a deep bench of services that reliably execute marketing functions, whether invoked by a human, an agent, or a classic process automation.
These platforms serve as a coordinating hub for the myriad of AI apps and agents multiplying across the workplace, bringing lawful-good coherence to an increasingly vibe-ified environment that would otherwise melt into chaotic-evil mayhem.
They aspire to deliver context-as-a-service — what I think of as productized context engineering — to the broader AI ecosystem within an organization. (I’ve been writing about that here, here, and here, a narrative multiple vendors are now adopting.)
I fully believe there’s a ton of value they can deliver in this role as an infrastructure platform — arguably more than the opinionated app layer most were selling up to this point. If they succeed, this can be their road out of the SaaSpocalypse.
But here’s where the real battle begins. Because in the contest to dominate marketing’s infrastructure, some beefy new gladiators are storming the arena.
A Strategic Taxonomy of Martech Infrastructure
Infrastructure is a multi-layered thing. I’m going to avoid calling it a “stack,” because the relationship between these different elements is more fluid and entangled than the fixed leg-bone-connected-to-the-knee-bone architectures that epitomized martech stacks of yesterday.
From an industry perspective, I see six categories of martech-relevant infrastructure:
Cloud Platforms such as AWS, Google Cloud, Microsoft Azure that provide base-level compute and storage along with a smorgasbord of higher-level services
Service Platforms such as Glean, Okta, Stripe, and Twilio that provide specialized programmatic capabilities for search, identity, security, payments, telephony, etc.
Data Platforms such as Databricks and Snowflake that serve as a universal data plane across an organization, with semantic-layer governance and embedded AI/ML compute (AWS, Google, and Microsoft play here too)
Builder Platforms for engineers and non-engineers to create custom apps and agents, from Claude Code, Codex, and Cursor to Lovable, Replit, Workato, and Zapier (many of the iPaaS leaders of the previous decade have pivoted here)
Business Platforms such as Adobe, HubSpot, and Salesforce that have been the backbone of the martech stack, with all the services I described earlier
AI Platforms such as Anthropic, Google Gemini, and OpenAI that started with foundation models and chat interfaces but are rapidly expanding into higher-level business functions
These categories are locked in a struggle of intensifying coopetition.
They benefit from cooperation — or more accurately, their customers benefit from their cooperation. AI platforms work across business platforms, which in turn integrate and aggregate through data platforms. Builder platforms enable custom agents and automations to be built on top of them. All of them run on underlying cloud platforms.
But they’re also muscling in on each other’s territories, competing for dollars and dominance. Google has a dog in almost every one of these fights: Gemini, Workspace, Antigravity, BigQuery, Google Cloud. So does Microsoft: Copilot, Dynamics, Fabric, Azure. Agents are probably the most contested ground, the high ground, as everyone wants you to build your agents on their platform.
Here’s how I see these categories playing out at the moment:
A few quick disclaimers. This is through the lens of my martech glasses, although I believe these patterns are similar across the rest of a company’s infrastructure. My examples are only a tiny representative sample to make each category more tangible — not an attempt to recreate the entire martech landscape. Placement of a few on the boundaries could be debated. And the ordering of the rows only approximates a sense of “altitude” as a business user might perceive them.
Martech’s Three-Body Problem
Gravitational pull is a measure of how strongly a category becomes the strategic center that the rest of the organization orients around. Companies describe themselves as “a Salesforce shop” or “building on Databricks.” That’s gravity.
While all six categories matter, the most unstable contest is between the three bodies pulling the hardest: AI platforms, business platforms, and data platforms.
(Cloud platforms have astronomical mass, but most business users still don't think of their cloud provider as the center of their customer experience. What clouds do have is reach, with tentacles stretching into nearly every other category. Which is exactly why they show up in everyone else’s cross-competition column.)
AI platforms want to become the new front door to work: the place where people express intent, agents reason through tasks, and actions are orchestrated across systems. If the agent is where the work begins, the AI platform has gravity.
Business platforms want to remain the operating layer of the business: the place where customer objects, campaign objects, permissions, processes, channels, and records of engagement live. If business operations and customer experiences are anchored there, the business platform has gravity.
Data platforms want to become the governed context layer: the place where enterprise data, semantic models, identity, history, analytics, and machine intelligence converge. If the organization believes truth and context live there, the data platform has gravity.
That’s the real contest: where does context live?
Is context primarily in the AI agent’s memory, tools, and reasoning environment? Is it in the business platform’s domain model, processes, and guardrails? Or is it in the data platform’s governed semantic layer and universal data plane?
The unsatisfying but accurate answer today is: yes.
Context is becoming distributed across AI platforms’ reasoning environments, business platforms’ domain models and processes, data platforms’ governed semantic layers, and builder platforms that stitch them together into custom apps, agents, and automations.
This is why I keep coming back to the idea of a composable canvas rather than a traditional stack.
In The New Martech “Stack” for the AI Age, the report I produced with Databricks, we described a new architecture in which data, semantics, context, decisioning, apps, and agents are composed together more dynamically than the old layer-cake model of martech.
You can see the rough mapping: data platforms anchor the universal data layer; business platforms provide domain-specific context and execution; AI platforms provide reasoning and interaction; builder platforms compose the custom apps, agents, and workflows across them.
The key word is compose.
In this world, infrastructure doesn’t merely support the apps. Infrastructure determines what can be composed, by whom, with what context, under what governance, and at what speed.
The martech infrastructure wars have begun. And context is the Force that binds this universe together.
Scott
P.S. Have you read the State of Martech 2026 report? Download it any time, free and ungated. And for the next few weeks you can still catch our #MartechDay keynote and interviews with seven martech leaders, free and on-demand.
