Levy describes agents autonomously fixing bugs, responding to security incidents, and reading entire contract corpuses — tasks never previously staffed — generating net new revenue and expa…

Pre-training and post-training breakthroughs plus exponential compute/data growth mean model capabilities keep doubling every few months; five US labs (Anthropic, OpenAI, Google, Meta, xAI)…

Continued compute availability and exponential high-quality data growth keep pre-training and post-training scaling laws intact, with no wall in sight across all major labs.

Kimi K3 ranks third globally on composite benchmarks, near-clone of top US models, and wins some subjective tests; China's open-weight ecosystem has closed the gap from 1-2 years to mere mo…

Kimi K3 ranks third globally on composite benchmarks, near-parity with top US models on some tasks, collapsing previous assumptions of a 1-2 year China lag; open-weight ecosystem in China i…

Levy notes the consensus two years ago placed China 6-12 months behind; Kimi K3's near-parity performance suggests export controls on chips have not prevented algorithmic catch-up, implying…

Consumer AI is near saturation but enterprise needs massive intelligence for complex workflows (M&A due diligence, life sciences, manufacturing); real-world diffusion limited by permits, cl…

Levy argues the rate limiter for enterprise AI value is diffusion — humans must integrate model intelligence with proprietary data, navigate permits/regulations, and close real-world feedba…

Enterprise adoption is constrained by the speed at which humans can integrate model intelligence with proprietary data, navigate real-world permits/regulations, and redesign end-to-end work…

Levy and Lonsdale agree that AI agents automating regulatory paperwork, environmental reviews, and permitting could compress multi-year physical-world timelines to weeks, making 'atoms' the…

Agents perform net-new work humans never did (reading all contracts for upsell signals, generating 5,000 personalized ads vs 5), expanding total economic output rather than replacing labor;…

Negative public sentiment driven by industry's own fear-mongering is translating into concrete policy threats: NY data center moratorium, red-state ban proposals, potential new tax regimes…

AI creators publicly expressing existential fears (Hinton, Anthropic leaders) creates cognitive dissonance: 'you're building it and scared but not stopping' — driving populist opposition, d…

Levy blames AI creators' public anxiety (Hinton, etc.) for scaring voters and politicians, predicting AI becomes a top 2028 election issue with populist restrictions on compute, data center…

AI agents make engineering cheaper, so every company (law firms, banks, manufacturers) now hires engineers to deploy agents; new job families emerge (AI automation engineers) and total engi…

Levy observes law firms, banks, and manufacturers now hiring engineers to build custom agents/models, arguing lower engineering cost unlocks latent demand in non-tech sectors — a structural…

Every 15-20 years a platform shift enables new startup waves; AI requires a new application layer translating model capabilities into domain-specific workflows (legal, finance, marketing, H…

Models improve exponentially but need application layer to connect to real workflows; this 'bridge layer' (agent-first companies, infrastructure, post-training, data) is where massive value…

Levy identifies five stack layers (agent-first apps, horizontal agent platforms, agent infrastructure, post-training/model tuning, data/infra) where startups will capture value by translati…