Anthropic says AI could almost double US productivity growth

Anthropic’s latest research suggests that current-generation AI could lift U.S. labor productivity growth to 1.8% annually if the technology reaches broad adoption over the next decade, nearly doubling the rate seen since 2019. The findings are based on the company’s discussions with its AI model, Claude, and reflect how businesses are already using artificial intelligence for tasks such as reviewing documents, summarizing customer interactions, and providing quick insights for employees. When the projected labor productivity gains are applied across the entire economy, they translate into a 1.1% annual increase in total factor productivity, a measure of how efficiently both people and systems work together.

Anthropic arrived at that figure by assuming that workers continue to receive about 64% of national income, consistent with historical averages, and by estimating how much time employees spend on tasks that AI can support. Jobs where workers devote more hours to AI-compatible tasks and jobs with a larger share of total wages had a greater influence on the overall productivity boost. The company measured how long these tasks take with and without AI, using Claude to estimate time spent on various job functions. Those estimates were then matched with employment data from the U.S. Bureau of Labor Statistics, highlighting roles in software development, management, marketing, and customer service as the top contributors to potential gains.

If achieved, the improvement would mark a significant shift. Total factor productivity has grown at under 1% annually since the early 2000s and averaged 2.1% since 1947. Early adoption already shows momentum. A Menlo Ventures analysis found that organizations are increasingly using AI for code generation, chatbots, enterprise search, and document summarization, with meeting summarization reaching 24% adoption in surveyed companies. Healthcare, legal services, and finance are among the fastest adopters, using AI to streamline coding, intake steps, and document review.

Anthropic’s September Economic Index report found that about 77% of enterprise API usage reflects automation patterns, showing that companies are already assigning specific tasks to AI. However, the company noted several uncertainties. Its projections assume universal adoption and steady AI capabilities and worker skills.