1. Macro Spending and Revenue Shift

The AI sector is currently undergoing a definitive shift in value capture from infrastructure to the application layer. Data from 2025 and early 2026 indicates that for the first time, enterprise spending on AI applications has surpassed spending on foundational infrastructure, signaling a transition from "infrastructure build-out" to "workflow integration."

1. Macro Spending and Revenue Shift

In 2025, the application layer captured $19 billion (51%) of the $37 billion total generative AI enterprise spending. This marks a critical milestone where application-layer revenue exceeded infrastructure spending ($18 billion).

While infrastructure requires massive capital expenditure—with Hyperscaler Capex projected at ~$700B for 2026—the application layer is seeing higher agility. Startups currently dominate the application layer, capturing 63% of revenue, whereas incumbents like Microsoft and Databricks maintain a 56% share of the infrastructure market.

2. Crypto AI Sector: From Compute to Agents

The cryptocurrency market reflects this shift through a rotation from "pure compute" infrastructure toward "Agentic" utility.

• AI Agent Ecosystems: Projects are pivoting to capture application-layer value. Fetch.ai (FET) launched an "Agent Launch" platform on BNB Chain, enabling autonomous AI agents to issue their own tokens and participate in a dedicated economy [Source: https://www.youtube.com/watch?v=dQw4w9WgXcQ].
• Consumer AI & DePIN: Grass (GRASS) focuses on user-facing products, including a native wallet launch scheduled for mid-July 2026 and data-scraping rewards for users [Source: https://www.coingabbar.com/grass-airdrop/].
• Strategic Integrations: The partnership between Fetch.ai and Visa to develop an AI-to-AI payment system (planned for 2026) demonstrates infrastructure providers moving up the stack to capture value in the payments application layer [Source: https://fetch.ai/blog/ai-to-ai-payment].

3. Structural Drivers of Value Migration

Several factors are accelerating the move toward the application layer:

• Commoditization of Models: As performance across foundation models converges, they are becoming a "thin-margin oligopoly." Value is migrating toward proprietary data and deep workflow integration.
• Agentic Workflows: The industry narrative has shifted from simple "prompts" to "workflows," where AI agents own entire end-to-end processes.
• Utility-Driven Tokenomics: New categories like ai-agents and defai (AI-integrated DeFi) suggest that AI is increasingly being embedded into existing protocols rather than standing alone as isolated infrastructure.

Conclusion

The AI sector is transitioning into a phase where the "app-layer" is the primary engine for revenue growth. While infrastructure tokens like Bittensor (TAO) and Render (RENDER) maintain significant valuations, the highest growth and social engagement are now centered on application-layer projects and AI agents. The primary open question remains the long-term defensibility of these applications as foundation model providers attempt to integrate similar features natively.