Google

Alphabet's search/AI giant — in this vault less a primary subject than a recurring third force alongside Anthropic and OpenAI. Frontier models (Gemini), DeepMind research (AlphaEvolve), and hyperscale infrastructure.

Recurring vault threads

• Governance position in the AI Licensing Regime (US) three-labs split: prefers an industry-funded body (finance's FINRA / power grid's NERC model) over Anthropic's government-veto and OpenAI's government-agency preferences. The Economist's Leader endorses a version of Google's model.
• The Apple deal: Apple's "new Siri" (June 8, 2026) ditched in-house models for Google's frontier models at ~$1bn/year — the highest-profile case of owning-the-relationship beating owning-the-model (Apple).
• Constitutional stance: on the consequentialist side of the industry's AI Constitutionalism split (with OpenAI, Waymo), vs Anthropic's deontology.
• Sovereign-scale infrastructure actor: 34+ subsea cables funded since 2008, 18 owned without partners — private firms doing foreign policy (Subsea Cables); plus the SpaceX-Google compute deal (The Way We Use AI is Changing) in the emerging compute market.
• AI-improving-AI: AlphaEvolve saved ~0.7% of Google's worldwide compute — a live Recursive Self-Improvement datapoint.

Cross-references

• AI Licensing Regime (US) · AI Constitutionalism · Subsea Cables · AlphaEvolve · Apple
• Anthropic · OpenAI — the other two of the three-labs governance split