An overview of macro tech trends for 2026, as “AI eats the world”: bubbles, the AI platform shift, Big Tech has FOMO, capex, Nvidia, US power backlogs, and more
What happens in a platform shift? The New Thing! All tech innovation, investment and company creation switches …
Context & Ripple Effects
The 2025 macro view had already framed generative AI around LLM scaling limits and a capex surge; this overview extends that arc from a model race into an AI-centered platform transition. It also anticipates the later focus on exploding AI capex and chip-demand bottlenecks.
The significance is that AI investment is no longer only a software or product question: the cited power constraints make physical capacity a potential gate on how quickly platforms and startups can deploy compute.
First-order effects
- Big Tech’s AI urgency redirects near-term capital and strategic attention toward compute capacity, while Nvidia benefits from the resulting hardware demand.
- Power backlogs can delay or constrain data-center expansion even where companies are willing to fund more AI infrastructure.
Second-order effects
- Scarce compute and electricity capacity increase the advantage of companies that can secure infrastructure early, while forcing rivals and AI startups to compete more directly for access.
- The investment cycle broadens from model developers to the infrastructure stack, reinforcing the subsequent emphasis on chip demand and supply-chain bottlenecks.
Third-order effects
- If AI remains the organizing platform shift, competitive advantage may increasingly combine models with infrastructure access, capital capacity, and deployment distribution rather than rest on model performance alone.
- Bubble risk rises when FOMO-driven spending is priced on anticipated AI demand before durable returns are established; capacity constraints may make that adjustment uneven rather than immediate.
The trend: AI is becoming an infrastructure-led platform shift in which capital, chips, and power shape the pace and distribution of innovation.