What NFX Is Thinking
*Synthesized from 15 recent items.* ## AI & agents - AI captures labor value, not software value — TAM jumps from $1T software to $50–60T global labor. [What Comes Next Is Bigger Than SaaS Ever Was](/nfx-tracker/ai-bigger-than-saas) - A 12-person startup can operate at 1,200-person capacity today. Software is an employee, not a tool. [1,000 Simultaneous Experiments](/nfx-tracker/next-mental-state-for-founders) - Best AI products win by removing attention demands — ambient, screenless, acting without supervision. [The Screenless Startup](/nfx-tracker/screenless-startup) - AI drives marginal cost of humanness to zero. Therapy, hospice, education, digital clones are the new frontier. [Software with a Soul](/nfx-tracker/software-with-a-soul) - AI-native games window: 36 months. Incumbents won't win it. Start with AI, ask what engages people second. [AI Games Are Coming](/nfx-tracker/ai-games) - Token costs dropped ~90% in one year. Category-defining products haven't launched yet — most big opportunities undiscovered. [What Comes Next Is Bigger Than SaaS Ever Was](/nfx-tracker/ai-bigger-than-saas) - AI automates execution; pure expression becomes the only defensible moat — Bowie over defensiveness, always. [The Infinite Game](/nfx-tracker/infinite-game) ## Biotech / life sciences - Seed-stage biotech can now reach human trials for under $10M — same infra cost collapse as cloud. [In Vitro, In Vivo, In China](/nfx-tracker/vitro-vivo-china-bio) - China overtook the US in clinical trial volume in 2025; 30–40% lower costs, growing regulatory friction risk. [In Vitro, In Vivo, In China](/nfx-tracker/vitro-vivo-china-bio) - Capstan: dosed in China seed-stage, acquired by AbbVie for $2.1B mid-Phase I. The playbook works. [In Vitro, In Vivo, In China](/nfx-tracker/vitro-vivo-china-bio) - Centivax in Phase 1 for universal flu vaccine; platform extends to pan-herpes Alzheimer's, malaria, universal antivenom. [Centivax Universal Flu Vaccine](/nfx-tracker/centivax-universal-flu-vaccine-phase-1) - Women's health is the biggest undervalued market in healthcare; platform play targets $30B by 2030. [Cyclana Bio](/nfx-tracker/cyclana-bio) - Longevity crossed from fringe to mainstream in 2025. NFX frames aging itself as a startup-sized problem. [The Founder's Guide to JPM](/nfx-tracker/founders-guide-to-jpm) ## Founders & company building - Run 50–100 parallel experiments at seed — 100 experiments yield ~99% probability of finding what works. [1,000 Simultaneous Experiments](/nfx-tracker/next-mental-state-for-founders) - Conviction + vision are the moat now. Capital and talent access are fully commoditized. [What Comes Next Is Bigger Than SaaS Ever Was](/nfx-tracker/ai-bigger-than-saas) - Hire for high agency, multi-domain fluency, low ego + high self-worth. Unlearning beats domain depth. [1,000 Simultaneous Experiments](/nfx-tracker/next-mental-state-for-founders) - AI expands the territory — defensiveness is the losing move. Go further into the water than you feel capable. [The Infinite Game](/nfx-tracker/infinite-game) - Be the founder who builds the codec before building the game. Infrastructure fluency precedes breakthrough products. [AI Games Are Coming](/nfx-tracker/ai-games) ## Space infrastructure - Space launch costs follow containerization's 36x collapse exactly: $54,500/kg → $2,700 (Falcon 9) → <$100 (Starship). [Ports, Rails, Roads, and Orbits](/nfx-tracker/ports-roads-rails-orbits) - Space infra window mirrors AWS 2006 — upper stack layers (compute, navigation) still undefined, closing fast. [The Stack Above the Cloud](/nfx-tracker/stack-above-the-cloud) - U.S. has one to two decades before China erodes its launch and orbital compute lead. [Ports, Rails, Roads, and Orbits](/nfx-tracker/ports-roads-rails-orbits) - Portfolio bets: Stoke Space (24-hour turnaround), Earthtraq (precision tracking), Starcloud (orbital computing). [The Stack Above the Cloud](/nfx-tracker/stack-above-the-cloud) --- *Last updated: 2026-04-15 15:48 UTC*