Midway through 2026, the AI economy appears to be very different from what most observers were characterizing eighteen months earlier. The early conception of generative AI as a productivity layer that businesses would implement consistently and methodically, increasing output across industries while labor markets gradually adjusted, has given place to something far more disorganized and challenging to map. Even the most knowledgeable observers are navigating this environment with genuine uncertainty due to the convergence of multiple factors at around the same moment and their interplay.
Probably the most significant change in the technical environment is the move to agentic AI. The chatbots that dominated the deployment cycle in 2023 and 2024 are not the same as systems that can plan, reason, and carry out multi-step tasks without human supervision at each level. Software engineering, legal analysis, financial modeling, and other white-collar jobs that were previously thought to be relatively sheltered are being automated, and businesses scaling these systems are discovering that the labor market implications are sharper and faster-arriving than earlier models predicted. While the direction is constant, the pace varies per industry.
The infrastructure figures are impressive and deserving of a clear statement. Through the end of 2028, it is anticipated that hardware and data center building would cost more than $2.9 trillion worldwide. That is physical infrastructure on the size of highway systems, requiring land, steel, water, and massive amounts of energy; it is not a technology investment in the strict sense. The bottleneck is starting to show up in the energy requirement. The need for data centers is already straining power grids in the US, Europe, and some parts of Asia, forcing governments and businesses to expand their nuclear and renewable capacity, which will take years to provide at the necessary scale. There is the computing power required for AI. Reliability in powering it is a completely separate issue.
The uniform growth story has been complicated by the fractures in corporate adoption. Compared to content-driven industries, industries with ongoing labor shortages, such as manufacturing, energy, and logistics, are implementing AI with a different level of urgency. Agentic AI provides operational continuity for a manufacturing firm that is unable to fill skilled staff. The calculus is nearly inverted for a media firm or creative agency that is swimming in inexpensive synthetic content. There is currently a “crisis of distinctiveness” in the market due to the abundance of AI-generated assets that are inexpensive to produce and hard to tell apart from human-created work. Investing additional money in the instruments causing the flood is certainly not the best course of action in that situation.
Although the geopolitical dimension has been developing for a while, it is now yielding tangible institutional outcomes. The US and European-led approach to AI regulation and standard-setting is directly challenged by China’s formation of the WAICO alliance, which aims to gather Global South countries into a coordinated framework for AI governance and development. The struggle is not only for market share but also for the digital and physical infrastructure—compute access, data sovereignty, and skilled labor—that determines whether economies can engage in the AI economy at all. Without those resources, nations run the risk of becoming fundamentally dependent on systems created and managed elsewhere, in addition to lagging behind in the adoption of AI.
From whatever angle, it appears that the tidy narratives around AI investment and implementation over the previous two years are colliding with a more complex reality. In some regions, the productivity benefits are genuine, while in others they are truly illusive. Even while some of the underlying presumptions are still being evaluated, the trillion-dollar infrastructure promises are set in stone. Additionally, the governance structures that were meant to act as safeguards are still lagging behind existing systems. Which of these forces moves the fastest will have a significant impact on what happens next, and nobody can confidently answer that issue at this time.
