In the fourth installment of our AI – Payments, Commerce & Markets series, Executive Advisor Wayne Johnson III explores 2025 AI investments and the technologies shaping the next phase of automation. He highlights projected capital expenditures from major tech companies like Meta, Amazon, Alphabet, and Microsoft—totaling up to $320 billion—as part of their push into AI and data infrastructure. The installment also dives into the emerging distinctions between AI agents, agentic AI, and AI-native technologies, breaking down their capabilities, use cases, and strategic implications for industries moving toward more autonomous systems.
Announced 2025 Large Language Model (LLM) AI Projects
Capital expenditure remains a significant barrier to delivering end-to-end AI services. For context, tech mega caps like Meta, Amazon, Alphabet, and Microsoft are projected to spend up to $320 billion combined on AI technologies and datacenter buildouts in 2025. Overall, worldwide spending on technology to support AI strategies is expected to reach $337 billion in 2025. The following table summarizes announced 2025 LLM projects and their estimated costs.
AI Native Technology, Agents and Agentic
AI Native technology is designed for specific environments with deep system integration, achieving high performance but lacking interoperability with external systems (i.e., intra company only).
End users interact with AI agents via user interfaces, which may query specific data sets or integrate with an LLM platform through an API. There are now hundreds of publicly available agents, with potentially more in company specific use. AI agents are typically reactive, responding to triggers or requests, and are designed for targeted tasks. They may integrate tools like APIs or databases but operate within bounded scopes, with limited ability to adapt independently beyond their programming.
Agentic AI, on the other hand, refers to a more advanced, fully autonomous form of artificial intelligence capable of goal-directed behavior without human intervention. It proactively makes decisions, sets, or adjusts goals, adapts to new situations, and can coordinate multiple systems or sub-agents. Agentic AI embodies “agency” or the capacity for independent action and reasoning in complex, dynamic environments while often serving as a broader framework or underlying technology that enables sophisticated automation.
Key Differences
The terms are related and sometimes used interchangeably. AI agents are often seen as a query based component, searching for relatively limited and simple answers, whereas agentic AI represents more sophisticated reasoning capability which drives true goal oriented autonomy. Below is a functional comparison as we understand it:
Implications
The distinction matters for applications: AI agents excel in efficiency for routine automation (e.g., in customer support or data processing), while agentic AI drives innovation in areas like robotics, supply chain management, or cybersecurity, where adaptability is key. As AI evolves, agentic AI could pave the way for more general intelligence.
