The next AI infrastructure race has nothing to do with chips
Investors have focused on AI infrastructure like chips and cloud capacity, but agentic AI—software that autonomously executes tasks—requires new systems for identity, permissions, payments, and audits. Early deployments like Travala’s AI travel protocol, launched June 10, demonstrate real-world applications, while experts predict 40% of enterprise apps will embed AI agents by 2026, highlighting trust and accountability as critical challenges.
Wall Street’s focus on AI infrastructure has centered on compute power, data centers, and cloud capacity, driving returns for early investors. However, a new gap is emerging as AI evolves from conversational models to agentic systems—software capable of autonomous task execution. The shift introduces legal, financial, and reputational risks, demanding infrastructure beyond traditional AI needs. Identity verification, permission frameworks, payment systems, and audit trails are now essential to ensure accountability. Sogni AI CEO Mau Ledford explains the transition: ‘AI is moving from *help me think* to *help me finish*.’ For example, while a chatbot may suggest a trip, an agent can book it. Travala launched the first end-to-end agentic AI travel protocol on June 10, enabling autonomous agents to search, reserve, and settle payments across 2.2 million properties, including Marriott, Hilton, and IHG. Transactions cost $0.01 per booking with near-instant settlement, though user approval remains required for final payment. CEO Juan Otero notes travel’s fragmented booking process makes it a natural fit for AI agents, eliminating manual steps like tabs and forms. Beyond travel, sectors like business procurement, customer support, financial management, and creative production are adopting agentic AI. Gartner projects 40% of enterprise applications will embed task-specific AI agents by 2026, up from under 5% in 2025. In creative production, agents can handle workflows from product photos to localized ad distribution, compressing multi-step processes into single instructions. Trust and identity remain the biggest hurdles. Unlike conversational AI, agentic systems must clearly define ownership, scope, and audit trails for every action. While solutions exist for some challenges, scaling accountability infrastructure will determine the pace of adoption. The focus is shifting from raw compute power to systems that ensure controlled autonomy.
This content was automatically generated and/or translated by AI. It may contain inaccuracies. Please refer to the original sources for verification.