Ex-Google/Alphabet CEO · Chair & CEO, Relativity Space · AI investor · Expert Watch · monthly scan
Why we watch him. He sits at the intersection of frontier-AI capital, US national-security policy and live battlefield deployment, so he front-runs where defence, energy and compute money is about to flow. His "San Francisco consensus" and MAIM framings shape elite investor and policy expectations on AI timelines.
Current headline view: AI is accelerating faster than institutions can absorb; the US must win the race to superintelligence against China, but must avoid a reckless "Manhattan Project" that triggers sabotage and escalation.
Latest 1-page summary (as at 9 Jul 2026)
The "San Francisco consensus". The Valley now broadly believes AI transforms every domain within roughly three to six years (RAISE Summit, Jul 2025).
Superintelligence within a decade. Recursive self-improvement is the threshold; warned AI could "think for itself" within about four years.
The US must win the AI race with China. After DeepSeek, called for US investment in open-source AI to counter China's adoption lead.
MAIM, not a "Manhattan Project". A unilateral dominance bid would draw preventive sabotage; deterrence beats an accelerationist race.
Energy, not chips, is the binding constraint. The US needs roughly 92 more gigawatts for AI; "our energy system will become the Achilles heel of our AI ambitions."
Big Tech should power its own AI. Hyperscalers should build private generation, not push AI's power demand onto public grids and bills.
Drones are the biggest revolution in warfare's history. A $5,000 drone kills a $5M tank; Ukraine "changed my mind about almost everything."
Humans "on the loop", not "in the loop". Future autonomous systems supervised and audited, not authorising each shot.
Notable reactions (operators & researchers)
Up to three recent, high-signal reactions from operators or researchers (not commentators) to a specific new argument of his. Challenges first. As at 9 Jul 2026.
No high-signal reactions this cycle. First scan pending.