What happens to a trillion-dollar AI industry when the intelligence it sells becomes something you can own outright, running on your own laptop, impossible to throttle, degrade, or switch off? And if a model one hundred times smaller than the frontier systems in the cloud can already match them on most everyday work, why are you still paying a subscription?
In this episode of Inside the Silicon Mind, we sit down with Jack O’Brien, founder of Subconscious, to dismantle the assumption that frontier labs will monopolize intelligence. Jack is a former Googler who spent years on the company’s search team, and he brings deep experience in machine learning infrastructure, post-training, and efficient inference. At Subconscious, he retrains open source language models to handle long-running agentic workloads, compressing frontier-level capability into models small enough to live on laptops, workstations, and eventually phones.
You’ll hear why Jack believes 95% of economically valuable AI work can already be done by small open source models, how the release of Alibaba’s Qwen models (including a 27 billion parameter version that rivals cloud systems a hundred times its size) arrived months ahead of his own trend line predictions, and why he thinks the “PhD in your pocket” (an owned, offline, personalized intelligence that no one can turn off) is closer than most people realize. He also ranks the big three labs, explains why Amazon has quietly become the Switzerland of AI, and makes the case that Apple’s bet on on-device intelligence could make it one of the year’s clearest AI winners.
Who this episode is for:
- Founders and CEOs deciding whether to build on frontier APIs or open source models
- Engineers and technical leaders working on inference, post-training, and agentic systems
- Investors evaluating frontier lab valuations and the economics of the model layer
- Knowledge workers and professionals (teachers, lawyers, doctors) wondering how near-free AI reshapes expensive expertise
- Executives at hardware and cloud companies watching the shift toward on-device AI
What you’ll learn:
- Why small open source models may already cover 95% of economically valuable AI work
- How intelligence is being compressed faster than expected, with laptop-scale models matching cloud systems a hundred times larger
- Why frontier lab valuations may still be justified even as the two-player market narrative breaks down
- Why Jack believes slow-moving Apple could emerge as one of the clearest AI winners through owned, on-device intelligence
- How AI could compress the cost of expensive professions while human trust keeps lawyers, doctors, and architects in the loop
Stay Tuned For Our Latest Episodes
Subscribe now and never miss an episode of Inside the Silicon Mind.
Start Listening Today!
Dive into game-changing conversations with the brightest minds in tech and business. Explore all episodes and start unlocking the secrets to success today.