Research Note: 80% of All AI-based Software Will Be Built On Top of Large, Pre-Trained Models By 2026
Strategic Planning Assumption
By 2026, 80% of AI-powered applications will utilize foundation models, fundamentally reshaping enterprise software architectures and creating a new competitive landscape driven by model development and training capabilities. (Probability 0.92)
The rapid proliferation of foundational AI models, such as GPT-3, DALL-E, and PaLM, is driving a seismic shift in how enterprises architect and deploy AI-powered applications. Industry analysts project that 80% of all AI-based software will be built on top of these large, pre-trained models by 2026, up from just 20% in 2023.
Foundational Model Adoption: Leading technology providers, including Google, OpenAI, and Microsoft, have all made significant investments in developing and commercializing foundational AI models that can be fine-tuned and adapted for a wide range of enterprise use cases. Forrester's research indicates that 65% of organizations are already planning to leverage these models in their AI initiatives, attracted by the significant reductions in development time and costs. A 2024 MIT Technology Review survey found that early adopters of foundation models are reporting 40-60% faster time-to-market for new AI applications compared to custom model development.
Architectural Implications: The adoption of foundation models will necessitate a fundamental rethinking of enterprise software architectures. Gartner forecasts that 70% of organizations will need to redesign their AI infrastructure and DevOps processes by 2026 to effectively manage the integration and continual update of these large, complex models. This shift will create new dependencies between software teams and AI/ML experts, requiring unprecedented levels of collaboration and shared accountability. IDC projects that 60% of enterprises will struggle with model governance, lineage tracking, and version control as foundation models become ubiquitous.
Competitive Dynamics: The strategic advantages conferred by foundation models will reshape competitive dynamics in the enterprise software market. Organizations that develop superior fine-tuning and training capabilities for these models will be able to deliver AI-powered applications faster and with greater differentiation than their competitors. A recent McKinsey study found that companies leading in foundation model development are seeing 20-30% higher profit margins on their AI-powered software offerings. However, this also creates a winner-take-most dynamic, as Forrester predicts that 40% of existing enterprise software vendors will be displaced by 2028 due to their inability to effectively leverage foundational AI.
Bottom Line
The ubiquitous adoption of foundation models represents a transformative shift in how enterprises approach AI-powered software development. Organizations that act quickly to develop the necessary technical expertise, architectural patterns, and model training capabilities will be positioned to dominate their markets by delivering AI applications with unprecedented speed, scale, and differentiation. Those that fail to adapt risk being left behind as foundation models become a critical source of competitive advantage, forcing a fundamental rethinking of enterprise software strategy and execution.