Latest

Every published item, in order · 7187 items
Nov 5
IndustryUS vs ChinaSelected10:25
DeepSeek Tokens Split Across 20 Inference Providers

Open-weight models decouple the model from the provider, so serving capacity becomes a commodity market where dozens of hosts compete on price and latency. Teams deploying DeepSeek can shop across providers rather than depend on one vendor. The 17% self-hosted share shows even the model's creator holds only a minority of its own serving traffic.

AI Will@FinanceYF5Source
Tomorrow 08:00, the daily lands in your inboxTelegram channel or email. Same screening as this page, no ads.
IndustryBuildersSelected02:49
Karpathy maps the inference chain from 1970s sci-fi to LLMs

The post reframes today's AI as the end point of a contingent chain rather than an obvious trajectory, which sharpens how readers judge current forecasts. Practitioners and analysts studying AI's historical drivers get a compact map linking computing, networking, data, Moore's law and scaling economics. It also clarifies why capital and capability now concentrate in a handful of labs that train and deploy copies of a few models.

Official accountAndrej Karpathy@karpathySource
IndustryBuildersSelected01:30
David Pawlan on Assistant Benchmark and 147 agents

Agent trust and privacy trade-offs emerge from real user behaviour rather than surveys: power users sacrifice privacy for perks like free pizza, which reshapes how consumer agents should be designed. Builders of agent products can use Pawlan's head-to-head testing method as a template for measuring whether their agents can hold competing preferences like price versus quality.

a16z@a16zSource

Only the latest 2000 items are listed; for older ones see the daily archive