HN Signal Picks

1. Apple Sues OpenAI Over Trade Secret Theft | 292 pts 133 comments https://news.ycombinator.com/item?id=48865019 Ex-employees allegedly took proprietary tech to OpenAI. This escalates the AI talent war between tech giants, potentially impacting AAPL valuation and broader AI sector sentiment.

2. GPT-5.6 Produces Mathematical Conjecture Proof | 300 pts 253 comments https://news.ycombinator.com/item?id=48863490 GPT-5.6 Sol Ultra generated a complete proof of the Cycle Double Cover Conjecture. A milestone in pure mathematical reasoning that validates continued rapid capability improvements at the frontier — bullish for AI infrastructure investment thesis.

3. NYC to Ban Deceptive Subscription Practices | 356 pts 192 comments https://news.ycombinator.com/item?id=48863464 Legislation targeting “hard-to-cancel” subscription models. If this trend spreads to other states/countries, it could materially impact SaaS retention rates and customer acquisition costs. Watch ADBE, CRM and other subscription-heavy names.

4. AI 2040: Plan A — Long-term AI Roadmap | 110 pts 77 comments https://news.ycombinator.com/item?id=48848425 Systematic exploration of AI’s 15-year development trajectory. Useful as a long-term investment framework reference for gauging where we are in the current AI cycle.

5. Boko Haram Using Frontier AI | 161 pts 137 comments https://news.ycombinator.com/item?id=48863707 Report on terrorist groups leveraging frontier AI sparks widespread discussion. Another case study for AI safety governance that may accelerate regulatory legislation, posing compliance cost and business model risks for AI companies.

6. SpaceX Plans 100K More Starlink Satellites | 35 pts 118 comments https://news.ycombinator.com/item?id=48863064 Ambition for 100x bandwidth. Low vote count but active comment section. LEO satellite competitive landscape worth tracking. Monitor ASTS, GSAT as comparable plays.

7. Xiaomi MiMo v2.5 Inference Optimization | 26 pts 6 comments https://news.ycombinator.com/item?id=48814170 Pushing hybrid SWA efficiency to the limit. Reflects sustained Chinese AI investment in inference optimization. AI inference cost reduction is a secular trend.

1. ai-berkshire — AI-Era Value Investing Framework ⭐ 12,639 | 📈 +3,596/week https://github.com/xbtlin/ai-berkshire Blends Buffett, Munger, Duan Yongping, and Li Lu methodologies with multi-agent parallel research. The marriage of AI and value investing is being productized, lowering retail research barriers.

2. caveman — 65% Token Optimization ⭐ 87,761 | 📈 +5,348/week https://github.com/JuliusBrussee/caveman Compresses LLM prompts by 65% using “caveman speak.” Directly reduces AI application operating costs — the LLM inference cost decline trend is accelerating at the tooling layer.

3. agent-skills — AI Coding Agent Skill Library ⭐ 76,807 | 📈 +7,944/week https://github.com/addyosmani/agent-skills Production-grade engineering skills for AI coding agents, maintained by a Google Chrome engineer. Nearly 8K weekly stars signals sustained developer enthusiasm for AI coding agents.

4. system_prompts_leaks — AI System Prompt Leak Collection ⭐ 55,814 | 📈 +7,765/week https://github.com/asgeirtj/system_prompts_leaks Aggregates system prompts from Claude, GPT-5.6, Gemini, Grok, and others. AI transparency and security discourse continues to heat up, also reflecting intensifying model competition.

5. RuView — WiFi-to-Spatial Intelligence ⭐ 79,881 | 📈 +3,657/week https://github.com/ruvnet/RuView Camera-free spatial awareness and vital sign monitoring. Privacy-compliant sensing tech could open new IoT and security markets.

Signal Analysis

Signal 1: AI Safety Governance Pressure Entering Upward Channel

Chain of logic: Apple sues OpenAI → tech giant AI competition shifts from cooperation to confrontation → Boko Haram frontier AI report raises security concerns → AI regulation expectations heat up → near-term headwind for AI companies (compliance costs), tailwind for cybersecurity (CRWD, PANW, S).

Signal 2: AI Coding Agent Ecosystem in Breakout Phase, Developer Toolchain Restructuring Accelerates

Chain of logic: agent-skills +8K stars/week + DesktopCommanderMCP + codex-plugin-cc rapid growth → AI coding agents moving from experiment to production → traditional developer tools (Atlassian TEAM, GitLab GTLB) face disruption risk, Microsoft/GitHub Copilot ecosystem (MSFT) benefits.

Signal 3: LLM Inference Cost Decline Trend Solidifies, AI Application Layer Benefits

Chain of logic: caveman 87K stars (65% cost reduction) + OmniRoute free gateway + MiMo inference optimization → sustained token consumption compression → AI application operating costs declining → bullish for AI application layer (CRM, NOW), but compresses infrastructure layer (NVDA) valuation premium thesis.