When a tech company gives away a successful product, you can be sure they’re seeking profit from another part of the stack.
The strategic landscape of artificial intelligence is undergoing a significant, albeit subtle, transformation. Major technology players, long perceived as fierce competitors, are now collectively engaging in a seemingly altruistic practice: donating highly successful, widely adopted products to open-source foundations. This trend, exemplified by Anthropic’s donation of the Model Context Protocol (MCP) and Google’s contribution of its Agent2Agent (A2A) protocol to the Linux Foundation, is not an act of pure beneficence. Instead, it signals a strategic pivot, a calculated move to commoditize one layer of the technology stack and establish dominance in another, more lucrative domain. This approach mirrors historical patterns seen in cloud computing and software development, suggesting a predictable evolution in how AI value is captured.
The recent contributions of industry giants to the open-source community have raised eyebrows and prompted considerable analysis. Anthropic, a prominent AI research company, made headlines last year by donating MCP, a protocol that had achieved remarkable traction. At the time of its donation to the Linux Foundation’s newly established Agentic AI Foundation, MCP was reportedly experiencing close to 100 million monthly SDK downloads, with over 10,000 active servers leveraging its capabilities. This impressive adoption rate begged the question: why would a company relinquish such a valuable and successful piece of technology?
This move by Anthropic was not an isolated incident. Months prior, Google had made a similar strategic decision, contributing its Agent2Agent (A2A) protocol to the Linux Foundation. This initiative garnered significant support, with major technology firms like AWS, Cisco, Microsoft, Salesforce, SAP, and ServiceNow joining as founding members. Even OpenAI, a direct competitor to Anthropic and Google in the AI model development space, has aligned itself with the MCP ecosystem. OpenAI’s Responses API now supports remote MCP servers, and the company actively participates in the MCP steering committee. Furthermore, OpenAI contributed AGENTS.md to the Agentic AI Foundation, signaling a cooperative stance alongside its rivals on foundational protocol development.
This coordinated effort to standardize communication protocols within the AI ecosystem can be likened to strategic maneuvers in a geopolitical chess match, where perceived acts of generosity mask a deeper intent to reshape the competitive battlefield. For those who have observed the evolution of the technology industry over decades, this pattern is familiar. It echoes the strategies employed in the commoditization of cloud infrastructure and on-premises server technologies, where platform companies strategically open-sourced certain components to create dependencies and capture value elsewhere. The core principle remains consistent: companies do not relinquish technologies they deem critical for maintaining a competitive edge. Instead, they offload technologies that have become foundational or where the primary locus of competitive advantage has shifted.
The Shifting Sands of Technological Gravity
The history of technology is replete with examples of companies strategically open-sourcing technologies to establish market dominance. Google’s extensive involvement in open-source initiatives serves as a prime illustration. As far back as 2017, analyses indicated that Google’s contributions to projects like TensorFlow and Kubernetes were not driven solely by a desire to foster community. Rather, these open-source assets were strategically positioned as "on-ramps" to Google Cloud. At a time when Google was vying for market share against established cloud leaders like AWS and Microsoft, the company aimed to create what then-Google product manager Martin Buhr described as a "gravity well in the market for container-based apps [so] that a significant percentage of them will end up with us." This strategy effectively commoditized the container orchestration layer, allowing Google to compete more effectively on the higher layers of its cloud platform.
This pattern highlights a recurring theme: platform companies routinely work to commoditize one layer of the technology stack, thereby enabling them to focus their competitive efforts on areas where they possess a stronger advantage.
GitHub offers another compelling historical parallel. Git, the underlying version control system, is inherently open-source. In the past, various companies offered Git hosting services. However, GitHub emerged as the dominant platform for millions of software developers worldwide. While no direct payment is made for the use of Git itself, a vast ecosystem of developers and organizations pays for GitHub’s services, including its collaborative features, project management tools, and security offerings. This phenomenon, where the underlying technology becomes ubiquitous and free while the platform built upon it captures immense value, is precisely what is now unfolding in the artificial intelligence domain.
Trading Contributions for Control in the AI Era
The stated motivations of companies like Anthropic and OpenAI—often framed around advancing humanity’s good—do not fully account for their rapid embrace of open protocols like MCP. The deeper strategic imperative stems from the realization that the AI model itself, while central to current innovation, is becoming a less effective long-term differentiator. The frontier model leaderboards, which track the performance of the most advanced AI models, are in constant flux, with rankings changing almost weekly. This rapid evolution means that enterprises cannot reliably build their AI strategies on the assumption that any single vendor will maintain a permanent lead in model quality.
Instead, the true value proposition for enterprises in the AI era is emerging in the "dull reality" of integrating these sophisticated models with existing enterprise data, workflows, and business processes. AI companies, more than anyone, understand this. While they will continue to invest billions in training increasingly capable models—a necessary strategy to attract developers, generate headlines, and open enterprise doors—they are also implicitly acknowledging that benchmark leadership alone does not secure a durable platform.
The gravitational pull that retains developers and enterprises is rooted in the accumulation of tools, workflows, team collaboration, and completed work. For enterprises, this means leveraging existing data, established permissions, robust governance frameworks, and integrated business processes. Each new integration deepens the commitment to a particular platform, making it more difficult to migrate. Similarly, the development of new workflows increases the platform’s utility and stickiness. Protocols like MCP and A2A are designed to enhance this "gravity" around AI models, making them more accessible and integral to everyday operations.
The ultimate objective for every major AI company is to become the central hub for AI-assisted work. This involves deploying vast numbers of forward-deployed engineers and developing strategies to connect legacy infrastructure with their cutting-edge models. Enterprise incumbents, conversely, aim to achieve the same goal from a different vantage point. They do not necessarily seek to own the frontier AI models themselves. Instead, their focus is on seamlessly integrating these advanced capabilities with the systems that already reliably manage their businesses.
This perspective fuels skepticism towards confident predictions of AI completely displacing enterprise software. The historical precedent suggests a different trajectory. While developers may flock to the latest AI innovations, enterprises operate on a foundation of decades of accumulated applications, data, policies, and business processes. The newest AI models and agent frameworks are undoubtedly important, but their true business value is only realized when they are effectively connected to critical enterprise systems such as customer relationship management (CRM), financial systems, supply chain management, human resources data, and other core operational platforms.
It is precisely within these established systems that incumbent technology providers continue to wield significant gravitational influence. Companies like Oracle, Microsoft, SAP, Salesforce, and ServiceNow recognize this inherent advantage. Ironically, the proliferation of open protocols like MCP can actually strengthen the position of these incumbents rather than weaken it. When every AI model can communicate using MCP, and when agents can interoperate through common standards, enterprises gain the flexibility to adopt the best available frontier technology without the need for extensive and costly re-integration of their existing systems. The protocol itself becomes a standardized, interchangeable component.
Open Standards: Facilitating, Not Eliminating, Gravity
The rise of open standards like MCP is not a diminishment of their importance. MCP’s success, for instance, is directly attributable to its ability to address a genuine challenge: the need for developers to avoid building custom connectors every time an AI application requires access to enterprise databases or other business systems. Furthermore, neutral governance of such foundational infrastructure is crucial, as no company wishes to cede control to a direct competitor. However, it is vital to distinguish between open interfaces and truly open markets.
An enterprise might find it straightforward to swap one MCP-compatible AI model for another. Yet, this interchangeability does not necessarily translate into a reduced dependence on the platform where its prompts, evaluations, security policies, and employee work habits have become deeply embedded. This scenario mirrors past technological shifts. Kubernetes, for example, dramatically enhanced workload portability, but it did not render cloud providers like AWS, Azure, and Google Cloud interchangeable. Similarly, despite decades of SQL standardization, relational databases remain fiercely competitive and differentiated markets. Standards effectively reduce friction, but they rarely eliminate competitive advantages; they merely reposition them.
In this evolving AI landscape, Anthropic, Google, OpenAI, and other key players are strategically standardizing the communication protocols between models, agents, tools, and enterprise systems. Their expectation is not that the communication layer itself will determine the ultimate winner. Instead, they anticipate winning by becoming the indispensable platform where AI-assisted work naturally coalesces and accumulates. This transition will undoubtedly involve the widespread dissemination of open-source code, leading to enhanced developer productivity for the broader industry and substantial financial gains for a select few. The game is indeed on, and the rules are being rewritten through strategic standardization.
