Morgan Stanley: A Transformative AI Leap Is Coming in 2026 — and Most of the World Is Not Ready
The investment bank warns that compute scaling laws are holding firm, and the accumulation of AI infrastructure at US labs is about to produce capability gains that will reshape the economy.
Morgan Stanley published a sweeping research report on Thursday warning that a transformative leap in artificial intelligence is imminent in the first half of 2026, driven by an unprecedented buildup of compute at leading American AI laboratories. The report argues that scaling laws — which predict that increasing compute inputs produces predictable gains in AI capability — continue to hold firm, and that the investments being made today are approaching an inflection point that most institutions, regulators, and employers are unprepared for.
The bank's analysts cited OpenAI's GPT-5.4 model, which recently scored 83 percent on the GDPVal benchmark — a measure of performance at economically valuable professional tasks — as early evidence of the transition from capable AI assistant to something that can match or exceed human experts in a wide range of knowledge work. GPT-5.4 also achieved 75 percent on OSWorld-Verified, which tests a model's ability to operate a desktop computer through screenshots and keyboard-and-mouse commands, surpassing the human benchmark of 72.4 percent.
Morgan Stanley described the emerging dynamic as the arrival of an 'intelligence factory' model, in which compute and power have become the primary inputs to economic value creation. The report projects a 12 to 25 percent deficit in US power supply relative to planned AI data center needs through 2028, identifying the energy grid as the binding constraint on the pace of AI expansion.
Data center developers are already converting Bitcoin mining facilities into high-performance computing centers and deploying natural gas turbines and fuel cells to meet demand.
The bank also identified 'transformative AI' as a deflationary force, warning that as AI systems replicate knowledge work at a fraction of human cost, prices in affected industries will compress, forcing restructuring across large portions of the economy. Several executives at major AI labs told investors to brace for progress that will 'shock' them, while Salesforce introduced a new productivity metric called Agentic Work Units to capture the value AI agents are delivering — an indication that conventional business metrics are already struggling to account for what is happening.
The report's conclusions were stark: the coin of the realm in the AI era is pure intelligence forged from compute and power, and the explosion is arriving faster than almost anyone predicted.
Read the original reporting at Fortune.
