The Power of Open Source: Building Giants in the Open
Explore how open-source AI powers production today: candid lessons from European founders (n8n, Black Forest Labs), a pragmatic tour of the open infra stack (Snowflake, OpenAI, NVIDIA, Supabase/Postgres), and investor playbooks for OSS licensing and sustainable monetization—plus 2026 bets on agentic apps, fine-tuning, unified data platforms, world models, and robotics demos. ⸻ ## ⏰ Timestamps 0:00 – Welcome & Event Setup Opening remarks, why Hugging Face hosted this side event at Slush, and overview of the agenda (3 panels + demos). ⸻ Panel 1 – Open-Source AI Founders (n8n & Black Forest Labs) 3:10 – Founder intros & origin stories Backgrounds of Jan (n8n) and Robin (BFL), their transition from creative/academic work to founding open-source AI companies. 6:20 – Building global OSS companies from Europe YC experiences, German GmbH setups, and lessons learned growing OSS brands globally. 10:40 – Licensing, business models & the OSS tension Fair-code decisions, monetization transparency, and handling license shifts with community trust. 15:00 – Product philosophy & internal dogfooding How both teams use community contributions internally and balance openness with product direction. 17:36 – What’s next for AI tools & assistants Future vision on agents, automation systems, and world-model-inspired workflows. 18:53 – Panel 1 wrap-up ⸻ Panel 2 – Open Infra & Developer Ecosystem (Snowflake, OpenAI, NVIDIA, Supabase) 19:23 – Intro: The rising developer stack for AI apps Why 2025–2026 are the “agentic application” years and how infra is adapting. 21:51 – Snowflake: Open formats & lakehouse future Iceberg, Polaris metadata catalog, ingestion tooling, and how Snowflake integrates HF models through Cortex/containers. 25:29 – OpenAI: OSS models & developer tools GPT-OSS + GPT-5 coexistence, Safeguard policies, and OpenAI’s focus on startups building on top of ChatGPT. 28:52 – NVIDIA: Full-stack open tooling & robotics Nemotron recipes, Cosmos world models, Groot for robotics, and NVIDIA’s view of open synthetic-data pipelines. 31:59 – Supabase: Postgres as the AI backbone How developers use Postgres for agents, AI app generation patterns, and Supabase’s MCP integrations. 41:17 – Market shift: From POCs to production AI apps Evaluations, agentic workloads, enterprise dataset integration, and distribution via ChatGPT/Apps. 45:05 – 2026 predictions for data + AI infra Unifying analytical & transactional workloads, Postgres everywhere, and more open-source model ecosystems. 45:30 – Panel 2 wrap-up ⸻ Panel 3 – OSS Investors (General Catalyst & Accel) 45:54 – Venture perspective on open-source AI Introductions, backgrounds, and why VCs see OSS infra as a long-term bet. 49:01 – What “open source” really means in 2025 OSI definitions vs pragmatic definitions, and avoiding “bait-and-switch” licensing. 51:04 – Hybrid models & regulated markets What should remain OSS, what should be proprietary, and why enterprise features often aren’t open. 54:42 – Measuring traction in OSS startups Real usage vs. vanity metrics, contributors, connectors, and examples (Kafka, Nix, Temporal, etc.). 57:17 – The optimistic future of OSS AI Why OSS keeps proprietary labs honest, ensures accessibility, and accelerates innovation. 58:25 – Final thoughts ⸻ Closing & Demos 1:00:53 – Community thanks & event close Shout-out to attendees, instructions for demo area, robotics corner, and networking.
