Anthropic officials are urging companies not to cut AI usage to control costs, calling that the wrong move and pushing for more efficient use instead. Claude Platform product head Angela Jiang and platform-engineering head Katelyn Lesse said on Sequoia’s “Training Data” podcast that capping AI spend is riskier than getting smarter, like skipping an all-night Opus run when a cheaper approach works. Anthropic is open to a model router “within the Claude space” to steer tasks to cheaper options, as rivals like Vercel push cross-model routing amid high token costs. Cost-skeptical enterprises are questioning AI ROI ahead of Anthropic’s anticipated IPO, with OpenAI’s Sam Altman touting GPT-5.6 Sol as half the price of Anthropic’s Fable.
