20VC: Who Wins the Inference Market | Has Open-Source LLM Inference Peaked? | What Happens to the Cost of Compute | The Real Limits of Post-Training and RL with Erik Bernhardsson, Modal CEO

Erik Bernhardsson is the Co-Founder and CEO of Modal, the AI infrastructure platform for training models, running inference and powering AI agents. Modal has raised over $1.2 billion, including a recent $750 million Series D led by Accel at a $15.75BN valuation. The company has surpassed $1BN in ARR, growing more than 15x year-on-year. Previously, Erik built Spotify's original music recommendation system. AGENDA:  05:30 - Why wait until your mid-30s to start a company? 06:40 - Is founding a startup really like "eating glass"? 08:00 - How to Raise $1BN with No Pitch Deck 08:45 - The danger of raising too much venture capital 09:40 - How much cash is Modal burning on GPUs? 10:30 - Do early-stage startups actually have moats? 13:30 - Who wins the inference market in the next 5 years? 15:30 - Are 90% software gross margins dead in AI? 17:45 - The "open-air drug market" for GPUs 21:30 - Will GPU hardware prices ever collapse? 22:15 - Are we about to overbuild data centers? 23:45 - Can custom silicon dethrone NVIDIA's monopoly? 24:30 - Is CUDA really the moat everyone claims it is? 25:20 - Are Neo Cloud providers partners or mortal enemies? 26:35 - Nebius vs. CoreWeave: who built the better business? 27:25 - Open-source vs. frontier models: which takes the lead? 28:20 - Can Europe ever build a competitive frontier model? 30:20 - The bull case for a $100B verticalized model company 37:15 - Modal's biggest resource allocation mistake 38:00 - Has open-source LLM inference already peaked? 41:30 - The real limits of post-training and RL 44:20 - How much does Modal spend every month on coding agents? 45:50 - The massive gold rush in selling data environments to labs 47:20 - Should enterprises own their own intelligence? 49:50 - Is model routing a defensible standalone business? 53:20 - $20M+ compensation packages for exceptional new grads 54:20 - The #1 portfolio company Erik would bet on 55:10 - What Erik changed his mind on most this year  

GPT-4正在为你翻译摘要中......

翻译全文