When the Ground Shifts: How Adobe Is Fighting for Gravity in the Martech Infrastructure Wars
In late May 2026, Scott Brinker—the analyst Ad Age once dubbed the “Godfather of Martech”—declared that the martech infrastructure wars had officially begun. His argument is worth sitting with, because it reframes the entire competitive landscape that companies like Adobe operate in. And once you understand the board Brinker has laid out, Adobe’s flurry of recent AI alliances stops looking like routine partnership announcements and starts looking like a deliberate campaign to avoid becoming a casualty.
Brinker’s Thesis: From Apps to Infrastructure
Brinker’s starting point is a shift that has been building for over a year: martech platforms are evolving from applications into infrastructure. For decades, the value proposition of a platform like Adobe, Salesforce, or HubSpot was the opinionated application layer—a polished UI, a set of workflows, and a brand of software you logged into to get marketing done.
That world is ending. Customers, Brinker argues, no longer simply want the freedom to use their own AI agents—Claude, ChatGPT, and a proliferating cast of independent agents—they now demand it. They refuse to be confined to a single vendor’s interface and a single vendor’s flavor of AI. The vendors have noticed. Salesforce released a fully headless platform. HubSpot committed to full API parity so agents can run it. A long roster of others—Braze, Marketo, Pega, and more—have shipped interfaces built on the Model Context Protocol (MCP). Collectively, Brinker writes, they’ve concluded that resistance is futile.
But here’s the twist that keeps these platforms relevant: marketers still value them. They remain the center of gravity for how marketing actually runs—structured data models, persistent workflows, well-governed permissions, and a deep bench of reliable services. Their new aspiration is to deliver what Brinker calls context-as-a-service: becoming the coordinating hub that brings coherence to an organization’s sprawling, increasingly chaotic collection of AI apps and agents.
The Three-Body Problem
Brinker maps the terrain into six categories of infrastructure—cloud platforms, service platforms, data platforms, builder platforms, business platforms, and AI platforms. Adobe sits squarely in the business platform category, alongside Salesforce and HubSpot, the traditional backbone of the stack.
The real drama, though, is what Brinker calls martech’s “three-body problem”: the unstable contest between the three categories pulling hardest for strategic gravity.
AI platforms (Anthropic, OpenAI, Google) want to be the new front door to work—the place where people express intent and agents orchestrate action.
Business platforms (Adobe, Salesforce, HubSpot) want to remain the operating layer—where customer objects, campaigns, permissions, and records of engagement live.
Data platforms (Databricks, Snowflake) want to be the governed context layer—where enterprise data, semantic models, and machine intelligence converge.
“Gravity,” in Brinker’s telling, is when an organization orients itself around you: when a company calls itself “a Salesforce shop” or says it’s “building on Databricks.” And the question that decides who has gravity boils down to a single, deceptively simple one: where does context live?
For Adobe, this is the existential question. It is being squeezed from above by AI platforms that want to own reasoning and interaction, and from below by data platforms that want to own the governed semantic layer. Notably, in Brinker’s original roundup of vendors who had already gone headless or shipped MCP interfaces, Adobe was conspicuously absent. The implication was uncomfortable: a business platform that fails to open up risks being reduced to a commoditized execution engine—something agents simply call into, rather than the place where the work and the context live.
Adobe’s Answer: Alliances as Strategy
If Brinker’s framework is the diagnosis, Adobe’s April 2026 Adobe Summit was the company’s clearest public response. And the centerpiece of that response was alliance-building at a scale that maps almost point-for-point onto Brinker’s prescription.
At Summit, Adobe announced a major expansion of its agentic ecosystem, making its AI agents, skills, and developer tools available within offerings from Amazon Web Services, Anthropic, Google Cloud, Microsoft, and OpenAI. In one move, Adobe closed the very gap Brinker had flagged. Rather than defending a walled garden, Adobe chose to make its capabilities reachable from inside the AI platforms its customers were already gravitating toward.
This is the “resistance is futile” capitulation Brinker described—but executed as offense rather than surrender. The logic is straightforward: if marketers are going to work through Claude or ChatGPT regardless, Adobe would rather be the thing those agents call into than be bypassed entirely.
Coopetition in Its Purest Form
What makes this strategy so striking is that Adobe is allying with the exact AI platforms Brinker identifies as its biggest gravitational rivals. This is coopetition at its most naked—and Adobe knows it.
In its own FY2026 SEC filings, Adobe acknowledges the tension without flinching. It warns investors of intensifying competition from generative and agentic AI solutions, some of them embedded within or operating across the very third-party AI platforms Adobe now partners with. It even cautions that other companies could, in the future, limit or interfere with Adobe’s ability to use third-party models or sell its own AI solutions. In other words, Adobe is partnering with Anthropic, OpenAI, and Google while simultaneously naming them as risks. That is precisely the “intensifying coopetition” Brinker describes: cooperation the customer demands, layered over a contest for dominance neither side has conceded.
Anchoring the Context
Alliances alone don’t win the gravity war—they could just as easily train customers to treat Adobe as interchangeable plumbing. So Adobe’s second move is to make a hard claim on context, the very thing Brinker says the whole war is about.
That claim is embodied in CX Enterprise, the end-to-end agentic system Adobe unveiled at Summit. Adobe positions it as grounded in decades of domain expertise in data, content, and customer journeys—explicitly framing it as an anchor for agents that are reliable, auditable, and able to understand context. The vocabulary is deliberate. Where Brinker asks whether context lives in the AI agent’s memory, the business platform’s domain model, or the data platform’s semantic layer, Adobe’s answer is to insist that its domain model is the indispensable anchor the agents reason over—not a layer they can route around.
Embracing the Composable Canvas
Brinker’s broader prescription was that the future is not a rigid “stack” but a composable canvas, where data, context, decisioning, apps, and agents are assembled dynamically. Adobe has adopted both the architecture and the language. CX Enterprise is built on a composable architecture designed to work across varied technology stacks, extending agentic skills and workflows across Adobe and partner solutions alike. Adobe describes its strategy as combining intelligence at the application layer with orchestration, governance, and decisioning across workflows—almost a direct mapping to the “business platform provides domain-specific context and execution” node in Brinker’s canvas.
A Telling Hedge
For all its open-ecosystem rhetoric, Adobe is not betting on every protocol equally—and that selectivity is revealing. The company has leaned hard into MCP, where it has seen strong adoption, building out offerings to help brands manage it. But it has been more cautious with agent-to-agent protocols, where adoption has lagged, and has adopted only the product-discovery portions of Google’s commerce-focused protocol.
This is controlled openness, not unconditional surrender. Adobe wants to be open enough to satisfy the freedom customers demand, while still steering orchestration in directions that keep Adobe central. It is, in effect, trying to have it both ways—and given the stakes Brinker outlines, that may be the only rational play.
Implications for Adobe’s Partner Ecosystem
The infrastructure war doesn’t just reshape how Adobe competes with AI and data platforms—it reshapes the much larger constellation of companies that build on and around Adobe. Brinker’s framework implies that as platforms become infrastructure, the partner ecosystem stops being a distribution channel for an application and becomes the connective tissue of a composable canvas. Adobe’s April announcements reveal a partner strategy reorganized around exactly that idea, with several distinct constituencies and several distinct sets of implications.
Technology partners: from integrations to interoperability
Historically, an Adobe “integration” meant a connector that let a third-party tool exchange data with an Adobe application. In the agentic model, the bar moves to interoperability across agents, skills, and MCP servers. Adobe’s new in-application integrations with Acxiom, Demandbase, Genesys, RainFocus, and SAP are framed as letting teams analyze data and take action in a single continuous flow with no tool-switching, while specialized partners cluster around specific functions: 24/7.ai, Algolia, and Netomi for conversational Brand Concierge experiences, and Adyen, PayPal, and Stripe for agent-powered checkout. The implication for technology partners is that the integration surface is shifting from the UI to the agent layer—partners who expose clean, governed, agent-callable capabilities will be composable inside Adobe-orchestrated workflows, while those still betting on screen-level integrations risk being routed around. NVIDIA’s role is a notable variant: rather than plugging into Adobe, it supplies the secure, policy-governed runtime (NVIDIA OpenShell, via the Agent Toolkit) on which CX Enterprise Coworker runs—a reminder that some partners are moving underneath Adobe in the stack, not just beside it.
Agencies: standardizing on the platform
The most consequential shift may be with the holding companies. Adobe says leading global agencies—dentsu, Havas, Omnicom, Publicis, Stagwell, and WPP—are standardizing on CX Enterprise, layering their own IP and industry expertise on top to co-develop differentiated solutions for joint clients. This is the gravity dynamic playing out one level down: if the world’s largest agencies anchor their agentic offerings on Adobe, they become a powerful reinforcing force pulling brands toward Adobe as the operating layer. The flip side is dependency risk in both directions. Agencies that standardize on CX Enterprise tie a growing share of their delivery model to Adobe’s roadmap and governance, while Adobe leans on agencies to carry its infrastructure into accounts it can’t reach alone. It is a deeper, higher-stakes relationship than the old reseller or services model—more aligned, but also more entangled.
System integrators: packaging infrastructure by vertical
A parallel shift applies to the global SIs—Accenture, Capgemini, Cognizant, Deloitte Digital, EY, IBM, Infosys, PwC, and TCS—who Adobe says are packaging its agentic capabilities into solutions for specific industry verticals to help customers modernize their stacks and shorten time to value. In Brinker’s terms, this is how a business platform earns gravity at the enterprise level: not by selling an app, but by becoming the substrate that integrators build repeatable, industry-specific solutions on. The implication is that Adobe’s competitive position increasingly depends on mindshare inside the SI delivery machine. The same integrators, however, are platform-agnostic by nature and will just as readily package Salesforce, Microsoft, or Databricks—so Adobe’s presence in their catalogs is a contest, not a guarantee.
The strategic tension: open ecosystem vs. platform-of-choice
Running underneath all of this is a tension Adobe states almost explicitly. Its stated ambition is to become the “platform-of-choice for effective multi-agent collaboration,” yet the same announcement insists businesses need ecosystems that are open and interoperable rather than closed. Those two goals coexist uneasily. A genuinely open ecosystem invites partners to compose freely—including in ways that diminish Adobe’s centrality—while a platform-of-choice strategy works only if Adobe remains the default hub. Adobe’s resolution, consistent with the controlled-openness pattern seen elsewhere, is to be open at the protocol and integration layer while concentrating governance, trust, and first-party context in its own hands. Adobe Marketing Agent illustrates the play: it is going generally available in Microsoft 365 Copilot and entering beta across Amazon Quick, Anthropic Claude Enterprise, ChatGPT Enterprise, Gemini Enterprise, and IBM watsonx Orchestrate—meeting users inside rival surfaces—but it stays grounded in first-party data and content from Adobe Experience Platform. The partner ecosystem can sprawl outward; the context stays anchored at home.
For partners, the net implication is a clarifying one. Value in Adobe’s ecosystem is migrating toward whoever controls governed context and agent-callable execution. Partners who supply differentiated data, specialized agent skills, or vertical IP that complements Adobe’s context layer stand to gain; partners whose value was primarily UI-level integration or generic connectivity will find that ground commoditizing fast—the same squeeze Brinker says Adobe itself is trying to escape, now passed one rung down the ecosystem.
The Open Question
Adobe’s alliances are best understood as its answer to the three-body problem. By making its capabilities callable from every major AI platform, Adobe reduces the risk of being bypassed. By grounding those capabilities in proprietary customer data and governance, it tries to prove its context can’t be replicated by the AI platforms above or the data platforms below.
Whether it works hinges on Brinker’s unresolved question. If brands come to believe that truth and context live in their data platform or inside their agent’s reasoning environment—rather than in Adobe’s domain model—then these very partnerships could end up teaching customers to treat Adobe as a commoditized execution engine: present in the workflow, but no longer the gravitational center.
The martech infrastructure wars have begun, as Brinker puts it, and context is the force that binds the universe together. Adobe has made its opening move: open the doors, ally with the rivals, and stake everything on the claim that when the agents need to know what’s true about the customer, they’ll still have to come to Adobe to find out.
Sources & Citations
Adobe’s product claims below are drawn from company communications and SEC filings, which are inherently promotional or self-interested; the agentic-orchestration vision remains early-stage, and Adobe itself has signaled it is watching real-world adoption closely.
Source [1] — Scott Brinker, “Begun, the martech infrastructure wars have…,” chiefmartec newsletter, May 28, 2026. https://newsletter.chiefmartec.com/p/begun-the-martech-infrastructure-wars-have
Source [2] — “Adobe Expands Partner Ecosystem to Deliver Frictionless Workflows for Customer Experience Orchestration,” Adobe News, April 20, 2026. https://news.adobe.com/news/2026/04/adobe-expands-partner-ecosystem
Source [3] — “Adobe Summit: Adobe Redefines Customer Experience Orchestration Vision in the Agentic AI Era with Introduction of CX Enterprise,” Adobe News, April 20, 2026. https://news.adobe.com/news/2026/04/adobe-redefines-custome-experience
Source [4] — “How Adobe is building AI and agentic tools for brands—while keeping an eye on actual adoption,” Marketing Brew, April 28, 2026. https://www.marketingbrew.com/stories/2026/04/28/adobe-ai-agentic-tools-brands-adoption-strategy
Source [5] — Adobe Inc., FY2026 SEC filings (Form 10-Q for the period ended February 27, 2026; risk-factor language incorporated by reference from the most recent Form 10-K). https://www.sec.gov/Archives/edgar/data/796343/000079634326000056/adbe-20260227.htm
Source [6] — Adobe Inc., Form DEF 14A (Proxy Statement), FY2026. https://www.sec.gov/Archives/edgar/data/0000796343/000079634326000043/adbe-20260227.htm
Claim-by-Claim Citations
Brinker’s Thesis: From Apps to Infrastructure
Brinker declared the martech infrastructure wars had begun and framed the app-to-infrastructure shift. [1]
Customers now demand the freedom to use Claude, ChatGPT, and independent agents rather than living in one vendor’s UI. [1]
Salesforce released a fully headless platform; HubSpot committed to full API parity; Braze, Marketo, Pega and others shipped MCP interfaces; vendors concluded “resistance is futile.” [1]
Platforms remain the center of gravity via structured data models, persistent workflows, governed permissions, and reliable services; their new aspiration is “context-as-a-service.” [1]
The Three-Body Problem
The six infrastructure categories and Adobe’s placement in the “business platform” category alongside Salesforce and HubSpot. [1]
The “three-body problem” among AI platforms, business platforms, and data platforms, and the description of each category’s ambition. [1]
The definition of “gravity” (“a Salesforce shop,” “building on Databricks”) and the central question, “where does context live?” [1]
Adobe’s conspicuous absence from Brinker’s roundup of vendors that had gone headless or shipped MCP interfaces. [1]
Adobe’s Answer: Alliances as Strategy
At Adobe Summit (April 20, 2026), Adobe made its AI agents, skills, and developer tools available within offerings from AWS, Anthropic, Google Cloud, Microsoft, and OpenAI. [2]
Coopetition in Its Purest Form
Adobe faces increasing competition from generative and agentic AI solutions, some embedded within or operating across third-party AI platforms. [5]
Adobe warns that other companies could prevent, limit, or interfere with its ability to use third-party models or make/sell its own AI solutions. [5]
Note: The competition and third-party-model risk language is drawn from Adobe’s FY2026 SEC disclosures (the 10-Q incorporates the 10-K risk factors by reference). The exact passage originates in Adobe’s risk-factor disclosures; readers seeking the verbatim text should consult the “Risk Factors” section of Adobe’s most recent Form 10-K and Form 10-Q.
Anchoring the Context
CX Enterprise is grounded in decades of domain expertise in data, content, and customer journeys—an “anchor for agents” that are reliable, auditable, and can understand context. [3]
Embracing the Composable Canvas
CX Enterprise is built on a composable architecture that works across technology stacks, extending agentic skills and workflows across Adobe and partner solutions. [3]
Adobe’s strategy combines intelligence at the application layer with orchestration, governance, and decisioning across workflows. [6]
The composable-canvas framework and its mapping of categories. [1]
A Telling Hedge
Adobe has seen strong adoption of MCP and built out offerings to help brands manage it; adoption of agent-to-agent protocols has been lower; Adobe adopted only the product-discovery portions of Google’s commerce protocol. [4]
Implications for Adobe’s Partner Ecosystem
Adobe’s stated ambition to become the “platform-of-choice for effective multi-agent collaboration,” partnering with technology companies, agencies, and system integrators. [2]
Adobe connects CX Enterprise with partners across agents, skills, and MCP servers. [2]
New in-application integrations with Acxiom, Demandbase, Genesys, RainFocus, and SAP enable analysis and action in a single continuous flow with no tool-switching. [2]
Brand Concierge partnerships with 24/7.ai, Algolia, and Netomi for governed conversational AI. [2]
Transaction/checkout partnerships with Adyen, PayPal, and Stripe. [2]
Adobe partnering with NVIDIA to build CX Enterprise Coworker on the NVIDIA Agent Toolkit and the OpenShell secure, policy-governed runtime. [2]
Leading agencies (dentsu, Havas, Omnicom, Publicis, Stagwell, WPP) standardizing on CX Enterprise and layering their own IP. [2]
System integrators (Accenture, Capgemini, Cognizant, Deloitte Digital, EY, IBM, Infosys, PwC, TCS) packaging Adobe’s agentic capabilities into vertical solutions to shorten time to value. [2]
Adobe’s framing that businesses need ecosystems that are open and interoperable rather than closed. [2]
Adobe Marketing Agent generally available in Microsoft 365 Copilot and in beta across Amazon Quick, Anthropic Claude Enterprise, ChatGPT Enterprise, Gemini Enterprise, and IBM watsonx Orchestrate, grounded in first-party data and content from Adobe Experience Platform. [2]
The composable-canvas framing and the principle that infrastructure determines what can be composed, by whom, and under what governance. [1]
The Open Question
Brinker’s framing that context is “the force that binds the universe together” and the unresolved question of where context ultimately lives. [1]
Want more from The Wolf Tank? Click here to be notified when new posts are published.