McKinsey Technology Trends Outlook 2026

McKinsey & Company | Michael Chui, Roger Roberts, Tanguy Catlin

Global energy technology investments reached nearly $200 billion in 2025 as the rapid scaling of artificial intelligence drove an unprecedented surge in physical infrastructure and data center power demands. This massive capital influx accompanied a doubling of AI infrastructure spending within a single year, shifting the technological frontier from digital screens to physical hardware.

To navigate these physical bottlenecks, the McKinsey Technology Trends Outlook 2026 tracks 14 critical trends across the AI revolution, compute frontiers, and cutting-edge engineering. Physical constraints now dictate digital progress. The report highlights agentic software development and AI-driven scientific discovery as two newly emerging domains accelerating breakthroughs in robotics, cybersecurity, and semiconductor design. However, deploying these innovations requires substantial workforce development and grid modernization. Ultimately, future scaling depends on which nations and organizations can build the physical hardware and secure the massive energy resources necessary to sustain these computational demands.

Comment

The transition of generative artificial intelligence from cloud-based software to physical infrastructure exposes a critical vulnerability in state-level industrial capacity. Computational sovereignty no longer depends on algorithmic design, but on the physical acquisition of advanced hardware like ASML's Twinscan EXE lithography systems. This dependency concentrates strategic leverage within a highly fragile, geographically centralised semiconductor supply chain. Consequently, the rate of technological scaling is bound to the physical limits of precision manufacturing and specialised industrial workforces in East Asia.

This hardware bottleneck directly constrains the deployment of high-performance computing clusters, such as the Frontier supercomputer at Oak Ridge National Laboratory, which are vital for simulating advanced materials and nuclear physics. Downstream, these physical limitations restrict the rapid prototyping of next-generation autonomous systems and cryptographic tools within the U.S. Department of Defense. Ultimately, the pace of military modernisation under initiatives like the U.S. Replicator programme will be dictated by the speed at which national power grids can integrate gigawatt-scale data centres.

Strategic Question for Discussion

If the physical supply of ASML's Twinscan EXE lithography systems remains constrained by geopolitical friction, does the operational viability of rapid hardware-centric initiatives like the U.S. Replicator programme break down, or can software-level optimisations compensate for the deficit?

My assessment is that software-level optimisations cannot fully bypass the physical throughput limits imposed by hardware scarcity. The historical trajectory of high-performance computing indicates that algorithmic efficiency gains eventually plateau, leaving raw hardware capacity as the ultimate determinant of strategic capability. Consequently, a prolonged supply disruption of ASML's lithography systems would likely force a scaling back of the U.S. Replicator programme's long-term deployment targets.

Share your assessment in the comments below.