Why open infrastructure will shape the future of AI

AI-Assisted development fuels dependence on proprietary tools

AI-assisted coding is changing how software gets built, and it’s happening fast. Developers are using tools like Claude Code, Gemini Code Assist, and GitHub Copilot to write code more quickly and with fewer mistakes. According to JetBrains Research, 74% of developers worldwide now use these kinds of AI tools. The 2026 State of Engineering Management Report from Jellyfish found Claude Code is already the most popular AI coding assistant, and 91% of developers said their productivity has increased over the past year.

That’s the upside. The downside is emerging just as quickly, dependency. These AI tools rely on cloud-based, vendor-managed infrastructure. Every query, every line of code is processed through proprietary APIs controlled by commercial entities. As Gartner predicts, by 2028, spending on AI coding tokens could exceed developer salaries. This points to a deeper issue: companies may find themselves locked into specific platforms that dictate their costs, access, and operational flexibility.

Peter Farkas, CEO of Percona, highlighted that convenience blinds many organizations to the risks of dependency. It’s easy to deploy a model or service with a few clicks, but that “ease” comes with hidden commitments. When a platform controls pricing, runtime, and data storage, independence erodes.

Executives should view this dependency not as a technical issue but as a strategic one. The decision to integrate proprietary AI tools into workflows is also a decision about who controls the future cost and availability of those capabilities. Open architectures and ownership of critical systems aren’t just ideals, they’re business levers for resilience, cost control, and long-term value creation.

Open infrastructure as the backbone of the AI era

Open infrastructure is not about being idealistic, it’s about maintaining freedom of action. The next wave of AI innovation will favor organizations that can move fast and switch technologies without disruption. Open standards and frameworks create that option. They allow companies to combine tools from different vendors, adopt new AI models as they emerge, and avoid dependency on a single supplier’s roadmap or billing model.

Brian Alvey, CTO of WordPress VIP, said it clearly: “Open always wins, not because it’s a fancy ideology, but because it gives you total freedom to adapt, evolve, and stay in control.” Manik Surtani, CTO and co-founder of the Agentic AI Foundation, reinforced this view, noting that AI cannot achieve its full potential if controlled by a handful of companies charging perpetual rent for access.

Open infrastructure is a control mechanism. It ensures enterprises can own their AI strategies, swap components as needed, and avoid costly migrations. It also enables interoperability across ecosystems, letting data and models move freely while maintaining governance and security standards.

For executives, the takeaway is simple. Openness is a hedge against future volatility. It keeps strategic options open as AI costs and capabilities shift. Closed platforms may promise speed now, but open infrastructure builds lasting power, the ability to adapt without asking permission. In the long run, that freedom becomes a competitive edge.

Divergence between open and proprietary AI paradigms

AI development is expanding in two directions. One is open, built through collaboration, transparent codebases, and shared innovation. The other is closed, driven by proprietary platforms that prioritize speed and monetization. Right now, both paths are advancing rapidly, but their long-term implications are very different.

Open-source AI is moving fast. Projects such as Mistral, Ai2’s OLMo, and the thousands of open-weight models on Hugging Face are challenging the idea that innovation must happen behind closed doors. Austin Parker, Director of AI Strategy at Honeycomb, points out that “open-source models and tooling are hot on the heels of state-of-the-art,” showing that community-driven work is matching corporate R&D in capability and pace. Mark Collier, General Manager of AI and Infrastructure at the Linux Foundation, adds that open frameworks and orchestration tools are advancing at remarkable speed.

Yet proprietary ecosystems are strengthening their control. Big players, Anthropic, Google, Microsoft, OpenAI, and others, are embedding open components into closed APIs and specialized platforms. They deliver convenience but limit flexibility, trapping organizations inside fixed interfaces. Roman Shaposhnik, Co-founder and CTO at Ainekko, compares this shift to earlier periods when “democratized” development environments led to rigid and closed systems. LangChain’s Open Agent Platform, initially open-sourced and celebrated, was discontinued in favor of managed commercial offerings, an early sign of this consolidation.

Executives should understand this divergence clearly. The choice between open and closed AI ecosystems is not just about speed to market, it’s about who defines your technological limits. Open AI supports integration, experimentation, and long-term value control. Proprietary AI delivers short-term results but limits the ability to evolve. The companies that future-proof themselves will be those that maintain the capacity to innovate independently of any single vendor’s toolset.

Proprietary dependency heightens business risk and limits innovation

Platform dependency is more than an operational risk, it’s a direct business vulnerability. Organizations that standardize around one proprietary AI system may gain short-term efficiency but lose flexibility, negotiating power, and resilience. When a provider changes pricing, deprecates tools, or shifts priorities, those decisions ripple across the customer’s operations. Recovering from such changes can mean costly migrations, lost data control, and long adjustment periods.

Mark Collier from the Linux Foundation highlights that “as infrastructure consolidates, enterprises become more exposed when platforms change direction, raise prices, or fall behind technically.” Roman Shaposhnik of Ainekko calls this dynamic a source of fragility: when the underlying layers of infrastructure are closed, everything built above them inherits that instability.

The cost implications are measurable. Cloudaware estimates that migrating large enterprise software systems can exceed $100,000 per move. Moreover, Gartner’s projections around AI token spending suggest that financial pressures will intensify as AI consumption scales. These pressures tie directly to the degree of vendor control.

Business leaders should treat these dependencies as strategic liabilities. Closed systems limit an organization’s ability to innovate or pivot quickly. An open infrastructure model, where software and data layers are built around flexible standards, mitigates those risks. It lets teams respond instantly to market or technology changes and keeps intellectual property within company control.

In the end, corporate leaders should design for optionality, ensuring they can evolve, merge, or replace components without depending on a vendor’s next move. The enterprises that maintain control over their infrastructure will own their trajectory in the AI-driven economy.

Growing momentum for open AI frameworks and governance

The AI industry is moving quickly toward openness. This is happening through tangible contributions, formal foundations, and enterprise adoption. The creation of the Agentic AI Foundation (AAIF), Anthropic’s donation of the Model Context Protocol (MCP), and Block’s release of the Goose agent mark significant steps toward industry-wide collaboration. These moves show a renewed commitment to shared infrastructure and interoperability rather than isolated innovation.

Austin Parker, Director of AI Strategy at Honeycomb, points out that open standards have always been a core driver behind global technological progress. His expectation is that this momentum will strengthen as enterprise AI adoption scales. Roman Shaposhnik, Co-founder and CTO at Ainekko, agrees that while progress is visible, the AI ecosystem still lacks cohesive governance bodies similar to early digital standardization groups. This absence leaves room for misalignment among vendors, especially as they race to define protocols and interfaces that will shape future systems.

These developments are complemented by other major contributions, such as the donation of llm-d, a Kubernetes framework for large language model (LLM) inference, to the Cloud Native Computing Foundation (CNCF). Efforts like these strengthen the shared infrastructure required for AI systems to evolve without fragmentation.

Executives should pay attention to this shift. Open collaboration in AI infrastructure is becoming a key differentiator. Companies contributing to or adopting open standards improve their ability to integrate new tools, maintain flexibility, and ensure compliance across increasingly complex environments. Engaging with these open initiatives is a strategic move to remain adaptable, competitive, and resilient in the next decade of AI growth.

Interoperability is critical for sustainable AI growth

Interoperability defines whether AI systems remain flexible as they scale. The ability for models, frameworks, tools, and data pipelines to communicate across vendors ensures that organizations are not locked into rigid, isolated ecosystems. Open Application Programming Interfaces (APIs), metadata standards, and communication protocols such as the Model Context Protocol (MCP) and the Agent Client Protocol (ACP) are key enablers of this flexibility. They create a foundation where developers can integrate and replace AI components without destabilizing entire workflows.

Mark Collier, General Manager of AI and Infrastructure at the Linux Foundation, emphasizes that openness in connecting systems, through shared protocols, identity frameworks, and data standards, matters as much as what happens inside the models themselves. Neeraj Abhyankar, Vice President of Data and AI at R Systems, adds that open orchestration and execution layers are essential for enterprises to maintain agility. Roman Shaposhnik from Ainekko further stresses the importance of getting MCP right, arguing that without standardized interaction layers, proprietary lock-in risks will only grow stronger.

This need for interoperability extends to AI operations. As inference moves closer to edge environments, organizations must understand how AI models perform, utilize memory, and scale across different infrastructures. Open systems increase visibility and enable optimization across these variables. Peter Farkas, CEO of Percona, identifies Kubernetes as a reliable foundation for hybrid deployments, offering hyperscaler-grade convenience without tying businesses to a single cloud vendor.

For the executive audience, the message is direct. Interoperability mitigates risk, sustains innovation velocity, and lowers long-term cost. It transforms AI infrastructure from a static investment into a dynamic asset that evolves with business needs. The companies that commit early to open, interoperable architectures will retain control, scalability, and speed as AI technology and regulatory frameworks continue to evolve worldwide.

Historical precedent paves the way for enduring open standards

History shows that open standards consistently produce resilient and scalable technology ecosystems. The internet and the Linux operating system stand as clear examples of how openness fosters collaboration, interoperability, and durability over time. These platforms became the backbone of global digital infrastructure precisely because no single entity controlled them. They allowed innovation to occur at every layer, hardware, software, and services, creating opportunities for industries to grow around shared foundations.

In the AI era, the same principle is emerging again. Open infrastructure ensures that organizations can build, operate, and scale on technology they fully understand and control. Mark Collier, General Manager of AI and Infrastructure at the Linux Foundation, notes that Linux became a global default because it provided a vendor-neutral foundation that everyone could build upon. That neutrality remains essential now as companies construct their AI ecosystems. Roman Shaposhnik, CTO of Ainekko, stresses that without open protocols, industries risk recreating tightly controlled systems that restrict competition and resilience.

The numbers reinforce this momentum. The 2026 State of Open Source Report found that avoiding vendor lock-in is the top driver of open-source adoption globally. This aligns with the observed trajectory of cloud-native technologies, where projects like Kubernetes evolved into fundamental infrastructure through openness and broad community participation. Austin Parker, Director of AI Strategy at Honeycomb, believes that the same pattern will define AI adoption, gradual standardization leading to scalable, interoperable growth.

For executives, the message is straightforward. Open infrastructure is a long-term strategy for stability and independence. Companies that invest in open standards today are setting the stage for sustained control over their technology stack and the flexibility to evolve as the AI landscape changes. Openness in AI ensures that progress continues at the pace of innovation rather than the pace set by single vendors. It establishes the groundwork for a balanced, competitive, and durable technology environment that can support global enterprise growth for decades to come.

Concluding thoughts

AI is moving faster than most organizations can adapt, and leaders need to decide how they’ll build for that future. Relying on closed, vendor-controlled systems might seem easy today, but it limits choice, flexibility, and long-term control. The companies that maintain ownership of their data, models, and infrastructure will ultimately control their own innovation cycles.

Open infrastructure isn’t just about ideology, it’s a business strategy built on freedom, adaptability, and cost transparency. It allows enterprises to scale without being boxed in by shifting vendor terms or restrictive architectures. It ensures that when AI evolves, and it will, you can evolve with it, on your own terms.

Executives should view openness as a stability mechanism in an evolving economy. The organizations that invest now in open standards, interoperable architectures, and transparent governance will gain more than technology flexibility, they’ll gain strategic independence.

In the AI era, the true competitive advantage won’t belong to those who move fastest. It will belong to those who move freely.