Computing Infrastructure Revalued as Annual Capex Exceeds Hundreds of Billions
The market's enthusiasm for artificial intelligence is driving a colossal reallocation of capital, with global spending on AI infrastructure, networking, and data pipelines now exceeding hundreds of billions of dollars annually. A new report from Citigroup argues that this spending surge is a response to a fundamental constraint: the end of Moore's Law for classical computing. As the process of shrinking transistors approaches its physical limits, traditional architecture can no longer deliver the performance gains required for large-scale AI.
This reality is forcing a structural shift in investment away from general-purpose servers and towards specialized hardware. "Computing is at an inflection point," the report states, highlighting that governments and corporations are now prioritizing dedicated accelerators, neuromorphic designs, and quantum computing. For investors, the focus is shifting from which companies secure the most Nvidia chips to which can solve the long-term scarcity of computational power and energy. This pivot transforms data infrastructure into a strategic national asset, with hybrid systems combining classical and quantum computing seen as a key "capability multiplier" for the next generation of AI.
Agentic AI Dismantles the Internet's Trillion-Dollar Ad Funnel
While AI reshapes the physical world, its agentic form is simultaneously deconstructing the foundational business model of the digital world. For two decades, internet platforms in e-commerce, travel, and advertising have thrived on a simple monetization funnel: attract users via search, guide their clicks, and facilitate a transaction. This entire model, heavily dependent on search engine optimization (SEO), is now under threat.
Citigroup's analysis explains that "agentic discovery is fragmenting the top of the funnel." AI assistants are increasingly empowered to research, compare, and transact on a user's behalf, creating a "Do It For Me" economy. This bypasses traditional websites and apps, rewriting the economics of user acquisition and weakening brand power. For platforms, this presents a dilemma: open their data to AI agents and risk being commoditized, or resist and risk becoming irrelevant. In high-value categories like real estate and auto sales, AI is poised to compress the entire transaction process, integrating financing, insurance, and logistics into a single, seamless action.
Humanoid Robots to Form Multi-Trillion Dollar Market by 2050
For AI to deliver on its macroeconomic promise, it must extend beyond digital screens into the physical world. The report identifies embodied AI—primarily through autonomous vehicles and humanoid robots—as the core driver of this transition. Citigroup forecasts a dramatic reshaping of the global labor force, predicting the deployment of hundreds of millions of humanoid robots by 2050. This surge, fueled by falling costs and rising capabilities, is projected to create a multi-trillion dollar market.
The demand is structural, driven by persistent labor shortages in sectors like logistics, manufacturing, and healthcare, particularly in aging economies. As AI evolves from simple assistance to fully autonomous agency in vehicles and factories, it promises to solve critical efficiency bottlenecks. This shift is also expected to ignite a "blue-collar renaissance," increasing demand for a skilled workforce needed to build, service, and manage the very robotic systems and infrastructure that power the new AI economy. Investors are now focusing on ecosystems that master multimodal AI and low-cost sensor technology to build the next generation of physical world automation.
Edgen Stock·Mar 25 2026, 04:55
