Anthropic CEO Claims 100 Million Word Context Window Is Feasible
A resurfaced clip has people arguing over whether giant context windows are near-term engineering or just another AI boast.
50537818145.4K
“From a machine learning persepective from an AI perspective, there is no reason why we cant make a context-window of 100m. Let's see what you've got, Anthropic.”
KimmonismusTECH#1364Story Brief
A resurfaced X post from Rohan Paul is recirculating an 11-month-old claim attributed to Anthropic CEO Dario Amodei: that a 100 million word context window is already possible "from a machine learning perspective," and that inference support is the real bottleneck. The clip, which Paul says came from the Alex Kantrowitz YouTube channel, also frames that scale as roughly equivalent to what a human hears in a lifetime. Another widely shared post on X turned the line into a challenge: fine, now show it.
“100m context would break so many agentic workflows wide open—excited to see how they handle memory management at that scale.”
Henryhndev
