6 Questions Defining AI's Trajectory
AI's path depends on minimal job displacement (0.4% cuts), data center politics, governance fights, energy-vulnerable infra funding, compounding enterprise adoption, and agents fueling entrepreneurship.
Job Displacement: Minimal Cuts, New Opportunities Emerge
Expect limited AI-driven layoffs—44% of 750 US CFOs surveyed plan cuts, but totaling just 0.4% of roles, nine times last year's but far from doomsday. Contrarian predictions like 30%+ college grad unemployment or 50% entry-level white-collar losses overlook nuances: AI exposure can boost hiring/wages if automated tasks complement non-automated ones, demand elasticity rises, or jobs have high task dimensionality. Tech hiring trends confirm this—product manager openings at 3-year highs, software engineer demand steady, AI roles exploding, overall tech jobs growing. Goldman Sachs projects 25% of US work hours automatable, displacing 6-7% of workers, but creating new categories: US needs 500,000 power workers by Dec 2025; data center construction added 216,000 jobs since Oct 2022. AI natives like OpenAI plan to double to 8,000 staff; ECB data shows they hire more than fire. Infrastructure buildout demands massive labor, shifting focus from losses to where new jobs arise.
Political Potency: Data Centers and Jobs Trump X-Risk
AI politicization hinges on jobs/data centers over existential risks, which lack resonance despite capability jumps. No clear partisan lines—Dems like AOC push AI money pledges and data center moratoriums; countered by Warner/Fetterman as 'dumb/China-first.' Republicans split (Trump/Hawley/Bannon/Santis). Data centers drive community concerns (energy bills), but White House 'ratepayer protection pledge' secures AI firm commitments to no bill hikes. Potency spikes if visualizing 10-15% unemployment; otherwise, deals mitigate. X-risk revives via Sanders but stays marginal.
Governance and Funding Risks: Power Struggles and Shocks
Who sets AI limits? Public Anthropic-Pentagon battles signal discomfort with private control, sparking calls for constitutional conventions (Stanford's Andy Hall). Nationalization whispers loom. Infrastructure funding shifted from hyperscaler balance sheets to private credit markets, risking credit crunches rippling to propped-up public markets. Geopolitical shocks amplify: Iran war spikes energy (WTO warns full-year highs crimp boom), disrupts Gulf $300B AI investments (UAE/Saudi via drone strikes), raises component/shipping costs. Time notes AI data centers drove 39% of US GDP growth (Q1-Q3 last year); disruptions hit economy-wide.
Adoption Compounding and Agent Agency: Leaders Pull Away, Entrepreneurs Thrive
Differentiated enterprise adoption creates winners: 20% fast adopters wildly outperform laggards via reinvesting AI gains into more AI, R&D, sales—not buybacks. Challenges are organizational (real org charts, data provisioning), not tech. Startups reinvent via agentic teams (e.g., OpenClaw), upending orgs; tiny teams hit millions in revenue. Agents don't fix output for cost savings—they expand leverage, giving users more work/output. Even if white-collar transitions displace some, agents enable pods of 4 knowledge workers/grads to launch businesses/consulting, sparking entrepreneurship boom over moping unemployment. Policy should support adaptability, not assume fixed entrepreneurial supply.