peter diamandis

Peter Diamandis holds SpaceX as his single largest investment and advocates a 'huddle' (long-term hold) strategy similar to Bitcoin maximalism.

Palantir is building an enterprise AI stack combining Nvidia compute, frontier open-weight models, and Palantir's application layer — allowing enterprises to control their own AI. This positions PLTR as the orchestration layer for sovereign/regulated AI deployments.

Starship 13's successful soft landing and Raptor relight demonstrate rapid iteration toward full reusability, targeting launch costs of $10-100/kg — a 500-5,000x reduction from Space Shuttle's $54,000/kg and 20-300x below Falcon 9's $2-3,000/kg. This cost collapse enables orbital manufacturing, Starlink V3 (0.5-1 Gbps globally), and massive space economy expansion.

OpenEXO (Peter Diamandis' venture) is launching a program for 30 companies to rewrite organizations as AI-native using local models like Kimi K3. Thesis: AI-native reorganization yields ~100x performance by eliminating token cost bloat (projected to exceed payroll) and tuning streamlined models to specific workflows. Speed and local control become critical as token costs explode.

Meta (and Google, X) monetize AI at the application layer via massive user bases (3.5B for Meta, 2B for Google, 1B for OpenAI). As model intelligence commoditizes (Netscape/Mozilla analogy), value shifts to distribution and product integration. Meta's Llama open strategy accelerates this commoditization to undermine closed-model competitors.

Science Corp (founded by ex-Neuralink president Max Hodak) received European CE mark for Prima, a retinal implant that restores central vision in macular degeneration via infrared projection from glasses to subretinal chip. Patients showed 5-line improvement on eye charts at 12 months. Prima serves as revenue-generating Stage Zero for Science's deeper BCI approach: growing neurons into the brain rather than destroying them with electrodes.

Lila Sciences constructing large-scale autonomous labs for AI-driven scientific discovery; federal labs to retrofit this capability, starting with narrow domains.

Laya Biosciences combines a scientific superintelligence trained on all accessible scientific knowledge with 1M sq ft of robotic labs; AI generates hypotheses, runs experiments overnight, updates theory, achieving 1000x faster iteration than human grad students.

Life Biosciences has dosed 18 patients with ER100 (three Yamanaka factors via viral vector) targeting retinal disease; primate studies showed reversal of glaucoma and NION disease; if successful, the platform should work across all organ systems.

Chinese lab Moonshot released Kimi K3, a 2.8T parameter multimodal open-weight model that jumped to #3 on the cost-performance frontier despite US chip embargo, proving transformer architecture still has massive headroom and frontier intelligence is now a perishable asset with weeks-long shelf life.