When a technology company gives away a valuable piece of software, the first instinct is to look for altruism. In artificial intelligence, the better explanation is strategy. Over the past year, several leading AI companies have transferred important protocols and specifications to neutral foundations or made them broadly available. The pattern is familiar from earlier platform battles: companies commoditize a layer they no longer need to control so they can compete on a different layer where they believe they can win.
The clearest example is the Model Context Protocol, or MCP. It began as a way for AI models to connect to tools, databases, applications, and other external systems. It became popular quickly, with nearly 100 million monthly SDK downloads and more than 10,000 active servers at one point. Rather than keep MCP as a proprietary advantage, Anthropic donated it to the Linux Foundation’s Agentic AI Foundation. That move looked generous, but it also reflected a strategic calculation: if MCP becomes a universal standard, the battle moves away from the connector itself and toward the platform where AI-assisted work actually happens.
Key facts at a glance
- Anthropic gave MCP to the Linux Foundation’s Agentic AI Foundation after the protocol reached nearly 100 million monthly SDK downloads and more than 10,000 active servers.
- Google transferred its Agent2Agent protocol to the Linux Foundation, with AWS, Cisco, Microsoft, Salesforce, SAP, and ServiceNow as founding members.
- OpenAI supports remote MCP servers in its Responses API, sits on the MCP steering committee, and contributed AGENTS.md to the same foundation.
- The competitive frontier in AI is shifting from raw model quality toward integration, workflow, data access, governance, and platform gravity.
- Open standards can reduce switching costs at the interface layer, but they often relocate vendor advantage rather than eliminate it.
Google followed a similar playbook months earlier when it handed its Agent2Agent protocol to the Linux Foundation. The list of founding members was not a collection of small startups. It included AWS, Cisco, Microsoft, Salesforce, SAP, and ServiceNow. OpenAI, meanwhile, supports remote MCP servers in its Responses API, sits on the MCP steering committee, and contributed AGENTS.md to the same foundation. The competitive choreography is striking. Companies that compete fiercely in models and cloud services are cooperating on the plumbing that lets AI systems talk to one another. It resembles a power struggle in which the main players seek influence through apparent acts of generosity rather than direct confrontation.
For anyone who has watched the software industry for a long time, this is predictable. Platform companies do not usually give away technologies they have stopped caring about. They give away technologies they no longer need to own because competitive advantage has moved to new ground. The same script played out in cloud computing, on-premises infrastructure, developer tools, and data management. The gift is real, but it is also a way to shape the market’s default assumptions.
Google’s open-source history offers a useful precedent. The company was an exceptionally active contributor to open source long before AI became the center of its strategy. TensorFlow and Kubernetes were not released simply out of generosity. They became on-ramps to Google Cloud at a time when Google was trying to catch up with AWS and Microsoft. As a Google product manager put it at the time, the company hoped to create a gravity well in the market for container-based applications so that a significant percentage of them would end up on its platform. That is the logic of platform economics: commoditize one layer of the stack so you can compete somewhere you hold a stronger hand.
GitHub may be an even better example. Git is open. Anyone can host a Git repository, and once upon a time different companies did exactly that. Yet GitHub became the default place where software development happens for millions of developers. Nobody pays for Git, but many people pay for GitHub. The value is not in the underlying protocol or the repository format. The value is in the accumulated workflows, collaboration habits, integrations, and social graph that make the platform hard to leave. The same phenomenon is now playing out in AI.
Trading contributions for control
Anthropic and OpenAI have both presented themselves as working for humanity’s benefit, but that is not a sufficient explanation for why they are racing to give away protocols such as MCP. The deeper reason is that the model itself has turned out to be a poor place to build a lasting moat. Frontier model leaderboards change almost weekly. A lab can be ahead on one benchmark and behind on another within a matter of days. Enterprises should not build their AI strategy around the assumption that any one vendor will remain permanently ahead on model quality. AI may be the exciting part, but the dull reality is connecting those models to enterprise data, workflows, permissions, and business processes.
The AI companies understand this better than anyone. They will continue spending billions training ever more capable models because frontier models attract developers, generate headlines, and open enterprise doors. But they are also quietly acknowledging that benchmark leadership alone does not create a durable platform. Developers return to the places where their tools, workflows, teammates, and accumulated work already live. Enterprises double down on the systems where their data, permissions, governance, and business processes are already connected. Every new integration makes that destination a little harder to leave, and every new workflow increases its pull. That is what MCP, A2A, and similar efforts are about: increasing gravity around the models.
Every major AI company wants to become the place where AI-assisted work naturally happens. They are amassing armies of forward-deployed engineers and using other means to get legacy infrastructure to tie back to their frontier models. Enterprise incumbents want the same thing, but from the opposite direction. They do not need to own the frontier. Instead, they need to connect the frontier to the systems that already run the business safely and reliably.
That is why predictions that AI will sweep away enterprise software deserve skepticism. Developers absolutely live on the frontier, but enterprises do not. Enterprises create value by connecting new capabilities to decades of accumulated applications, data, policies, and business processes. The newest model matters, and so does the newest agent framework. But neither creates much business value until it is connected to customer records, financial systems, supply chains, HR data, and everything else enterprises already depend on. The frontier is where experimentation happens. The enterprise is where operational trust is earned.
That is where incumbents still possess enormous gravitational pull. Large enterprise software vendors believe this deeply, and for good reason. Ironically, open protocols strengthen that position rather than weaken it. If every model can speak MCP and every agent can interoperate through common standards, enterprises gain the freedom to adopt whichever frontier technology looks best without rebuilding every integration. The protocol becomes interchangeable. The place where prompts, evaluations, security policies, and employee habits accumulate remains sticky.
Open standards do not stop gravity
None of this diminishes the importance of open standards. MCP succeeded because it solves a genuine problem. Developers should not have to build a custom connector every time an AI application needs access to a database or another business system. Neutral governance also matters because nobody wants foundational infrastructure controlled by a direct competitor. But open interfaces should not be confused with open markets. A standard can make one part of the stack easier to swap while leaving the rest of the platform as entrenched as ever.
An enterprise may find it easy to swap one MCP-compatible model for another while still remaining deeply dependent on the place where its prompts, evaluations, security policies, and employee habits have accumulated. The same pattern has appeared before. Kubernetes made workloads dramatically more portable without making AWS, Microsoft Azure, and Google Cloud interchangeable. SQL has been standardized for decades, yet databases remain fiercely differentiated businesses. Standards reduce friction, but they rarely eliminate competitive advantage. They simply move it.
In the same manner, Anthropic, Google, OpenAI, and others are happily standardizing how models, agents, tools, and enterprise systems communicate because they do not expect the connection itself to determine the winner. They expect to win by becoming the place where AI-assisted work naturally accumulates. Along the way, the market will see copious quantities of code given away, increasing developer productivity for everyone and creating outsized financial rewards for a few. The real competition is not over the protocol. It is over the gravity that the protocol creates.
Source: InfoWorld News