a16z Podcast · 4 min instead of 59 min · Oct 14, 2026
Scaling laws argue against a fast takeoff, but cyber risk is real
Nathan Lambert
Co-founder & executive director, Trillium Labs
Led post-training for Ai2's open Olmo and Tülu models, wrote the RLHF Book and writes the Interconnects newsletter.
Erik TorenbergHost · General partner, Andreessen Horowitz
Helped build Product Hunt; founded On Deck, Village Global and Turpentine.
About this segment
Lambert says exponential compute for linear gains makes self-building models unlikely to take off quickly. His bigger worry is that public cybersecurity is unprepared for capabilities that are already spreading.
From “Can Open Models Keep Up With OpenAI and Anthropic? | Nathan Lambert”, Oct 14, 2026. Watch the full episode
“All the scaling laws are exponential increase in compute for linear increase in performance.”
Nathan Lambertat 55:31 · a16z Podcast55:31Takeaways
- Labs spend exponentially more on inference to sustain the current pace of progress
- Math may be unusually suited to today's models, so math wins may not generalize
- Not fearing fast takeoff is not the same as ignoring present and spreading risks
In this episode
Full episode- 8:53Why distillation explains less of China's AI progress than assumed4:44
- 24:51Frontier gains outside math and code come from buying data4:40
54:55Scaling laws argue against a fast takeoff, but cyber risk is real3:41