Wellington quantifies AI's macro footprint: 1.7% of GDP in capex, 25% of US growth, investment spending matching consumer spending; 19% of 2026 layoffs from AI but aggregate low; IMF flags AI debt risk; energy as binding constraint; job creation study
Key Questions
What is Wellington's estimate of AI's macro footprint on the US economy?
AI capex represents 1.7% of US GDP and contributes 25% of economic growth, with investment spending now matching consumer spending while creating inflation pressure from compute demand.
How significant is AI-related debt as a risk according to the IMF?
The IMF views AI debt as a larger stability risk than valuation bubbles, noting hyperscalers issued $159B in bonds in five months alongside Nvidia's $25B issuance, highlighting maturity mismatches.
What does recent research show about AI's impact on job creation versus layoffs?
High AI spenders grew headcount 10.2% versus flat growth for low adopters, with entry-level roles up 12%, while 19% of 2026 layoffs were AI-related but aggregate unemployment remains low in an augmentation phase.
Why is energy considered the binding constraint in the AI capex cycle?
Power and grid infrastructure are shifting focus from semiconductors, with 60% of data center capacity delayed and utilities preparing record spending as the cycle has not yet peaked.
How does AI adoption affect hiring patterns at different firm types?
AI-native startups are hiring fewer entry-level workers per Harvard research, yet broader high-adoption firms show net job growth, challenging narratives of widespread displacement.
Wellington Management analysis: AI capex now 1.7% of US GDP, contributing 25% of US economic growth; investment spending on AI has reached parity with consumer spending. Labor impact remains in augmentation phase—19% of 2026 layoffs are AI-related but aggregate unemployment low. Compute demand is creating inflation pressure. New: IMF flags AI debt as bigger risk than valuation bubble—hyperscalers issued $159B in bonds in 5 months, Nvidia $25B, maturity mismatch. J.P. Morgan notes 60% of data center capacity delayed. New: AI capex cycle not peaked, energy as binding constraint shifting focus from semiconductors to power/grid infrastructure. New: Ramp/Revelio Labs study: high AI spenders grew headcount 10.2% vs flat for low adopters, entry-level up 12%—challenges layoff narrative and suggests AI-driven job creation in high-adoption firms.