How AI might shape society and what Superintelligence might look like
Geo Chen
· Fidenza Macro
· May 15, 2026 at 04:21
· ⏱ 13 min read
| Read on Substack ↗
Summary
The article argues that AI progress is accelerating toward superintelligence within five years via recursive self-improvement and unhobbling, leading to extreme concentration of power in frontier AI labs and corporations. For markets, Google stands out as the public company best positioned to capture value from this trend, while the broader thesis implies a bifurcation between AI haves and have-nots and increasing dual-use risks.
•Aschenbrenner estimates ~0.5 OOMs/yr compute efficiency gains from GPT-4, reaching ~2 OOMs by 2027.
•Current model intelligence is at PhD level in some areas but not yet an automated AI researcher/engineer.
•AI is being unhobbled via tools like Perplexity Computer, Claude, and Manus that access emails, files, and software.
•Recursive self-improvement could compress a human-decade of algorithmic progress into under a year, yielding 5+ OOM jumps.
•Superintelligence will likely be licensed only to a narrow set of governments and corporations, creating extreme pricing power for frontier labs.
•Google is explicitly identified as the company with AI expertise, compute, and capital to stay at the frontier and leverage superintelligence for enormous value.
Read time13 min
Length13,143 chars
Categorymacro
Trade Ideas
Geo ChenGlobal macro trader; ex-head of FX trading, Credit Suisse
Author states: 'Google stands out as the one company that has the AI expertise and compute to stay at the frontier and the capital to leverage Superintelligence to create enormous value.' This positio
Author states: 'Google stands out as the one company that has the AI expertise and compute to stay at the frontier and the capital to leverage Superintelligence to create enormous value.' This positions Google as a primary beneficiary of the superintelligence thesis.
Risk: Regulatory scrutiny, potential misalignment or security incidents, and competition from other frontier labs (OpenAI, Anthropic) could limit upside. Concentration of power also attracts political risk.