joe lonsdale

Foundation models for robotics are entering a GPT-3 era where pre-training on massive diverse physical interaction data yields emergent capabilities like ambidexterity and tool generalization, with reliability crossing 99% on one hour of task data — enabling commercial deployment in late 2020s not 2030s.

Current LLMs already enable economic revolution if development stopped today; the constraint is organizational ability to provide context, restructure workflows, and reform institutions (e.g., legal billing models) — a multi-year diffusion process creating sustained tailwinds for implementation-focused companies.

The AI investment landscape maps to a six-layer stack where value accrues differently across layers; 8VC focuses on Level 5 (apps) while acknowledging massive capex at Levels 0-2 (energy, chips, data centers) with historical parallels to industrial revolution investment rates.

Joe Lonsdale and Jeff Woo argue that bringing 'atoms not bits' manufacturing back to the US — exemplified by Elon Musk's lead and companies like Saronic — is critical for national security, job creation, and industrial resilience, and they are actively allocating capital to this theme.

Palantir's deployment at the SBA demonstrates how AI can systematically uncover fraud networks across federal programs by integrating data ontologies, recognizing patterns, and crawling connections — creating a replicable model for government-wide waste reduction that could drive significant demand for enterprise AI platforms in the public sector.

Palantir's AI platform ingests disparate data sets to map fraud networks across PPP and 8(a) programs, identifying patterns (overseas IPs, non-citizens, pass-through schemes) that human auditors missed; model will scale government-wide.

China has captured over one-third of global therapeutics development by stealing U.S. IP through patent office exploitation and talent transfer, while U.S. FDA delays (700 days vs 90 days for Phase 1 trials) hinder competitive response.

Valuation multiples for core AI beneficiaries remain reasonable (Nvidia 25x, Micron 9x forward PE) and earnings are growing rapidly; the limiting factor is enterprise adoption diffusion, not technology capability, implying sustained growth rather than speculative bubble.