Open Models Will Absorb Most of the AI Spend

2026-07-22 · Source: Gradient Flow · Field: Technology & Digital — Artificial Intelligence & Machine Learning, Emerging Technologies & Innovation, Cybersecurity & Data Privacy · Depth: Intermediate, long

Summary

Open models, encompassing both open weights and open source, are projected to absorb the majority of global AI spending, despite proprietary frontier models garnering headlines and high IPO valuations. This shift is driven by the rapid improvement of open models, which are quickly narrowing the performance gap with leaders, often within two to six months. Economic factors like the "Jevons paradox," where cheaper per-task costs lead to increased overall consumption, and the strategic use of premium models as planners rather than workhorses, further favor open alternatives. Additionally, open models offer critical benefits such as enhanced control over data and hardware, improved privacy, and a hedge against vendor lock-in, as exemplified by incidents like the Fable shutdown. The article highlights the growing importance of inference efficiency and geopolitical considerations, particularly regarding US-China AI development.

Key takeaway

For AI Architects and VPs of Engineering evaluating their AI infrastructure strategy, prioritize building resilient, multi-model stacks that leverage open weights for cost efficiency, data control, and vendor independence. Develop robust switching layers and evaluate models based on specific task performance and ownership, not just headline benchmarks, to mitigate lock-in risks and optimize long-term spend. Your prompt logs and agent traces are strategic training assets; manage them carefully.

Key insights

Open models offer superior economics, control, and resilience, driving a shift in AI spend away from proprietary frontier models.

Principles

• Frontier model leads are short-lived (2-6 months).
• Cheaper AI units increase total consumption (Jevons paradox).
• Control over data and weights is paramount for enterprises.

Method

Architect AI systems with premium models for planning and cheaper open models for bulk execution, building a swappable portfolio with a robust switching layer.

In practice

• Implement a portfolio approach for AI model usage.
• Develop a switching layer for model interoperability.
• Treat prompt logs and agent traces as strategic assets.

Topics

• Open Models
• AI Economics
• Vendor Lock-in
• Data Control
• Inference Efficiency
• AI Architecture

Best for: Investor, CTO, Entrepreneur, Director of AI/ML, AI Architect, VP of Engineering/Data

Related on AIssential

• The real AI race may no longer be at the frontier — The AI industry is shifting towards open-weight models for production due to cost, ownership, and control benefits.
• Open and closed models are on different exponentials — Open and closed AI models follow divergent economic exponentials, creating distinct market structures and value capture mechanisms.
• Open Source Growth Boosts Together AI, Hugging Face — Open-source AI models and supporting services are rapidly gaining traction due to cost efficiency and vendor lock-in concerns, driving significant industry shifts.
• we need to talk about where AI spend is actually going — Open models, not proprietary frontier models, are predicted to absorb most global AI spending and compute resources.
• Spend Like AGI, Lobby Like It’s a Toy — AI's rapid advancement creates market paradoxes and shifts value to context and operational efficiency.
• The Hybrid AI Stack Is Coming for the Pricing Power of OpenAI and Anthropic — Enterprises are adopting hybrid AI stacks, balancing proprietary and open-weights models for cost, control, and workload-specific optimization.

See Counsel's argued verdicts on the open AI decisions leaders are weighing →

Open in AIssential →

Editorial summary, takeaway, and curation by AIssential. Original article published by Gradient Flow.