Anthropic Model: US GDP Up 32% by 2030, But Workers May Lose Ground

Anthropic's Economics team has released a scenario model exploring how varying levels of AI capability and adoption could reshape the US economy by 2030. Under the most extreme assumptions, GDP reaches $44.4 trillion (at 2025 price levels), 32.4% above a no-AI baseline, with annual growth potentially hitting 15%. The catch: knowledge workers could see wages fall more than 10%, unemployment could exceed typical recession levels, and the share of economic output going to labour could drop from roughly 60% to 45%.

What happened

Scenario	US GDP by 2030	GDP lift vs. no-AI baseline	Labour share of GDP
Modest	~$34.1 trillion	+1.6%	59.4%
Substantial	~$36.3 trillion	+8.3%	56.1%
Extreme	~$44.4 trillion	+32.4%	45.2%

All figures use 2025 price levels. Anthropic’s Economics team published the model in September 2026, framing it explicitly as a simulation rather than a forecast. The point is to show how different assumptions about AI capability, adoption speed, autonomy and productivity each produce a different economic outcome.

The three scenarios are separated by how much AI can do and how fast it spreads. The modest scenario compares AI’s impact to that of the internet: real but gradual, within the historical range for major technology shifts. The substantial scenario has AI performing roughly half of all knowledge-work tasks by 2030, with most of that technically autonomous, though adoption is still incomplete. The extreme scenario assumes AI outperforms humans across nearly all knowledge-work tasks, performs them autonomously, creates few new tasks for people, and spreads very fast. Anthropic notes the extreme outcome would likely require recursively self-improving AI (systems that improve their own capabilities without human intervention).

Why knowledge workers face the biggest disruption

Anthropic’s model treats jobs as bundles of individual tasks rather than all-or-nothing occupations. AI can leave a task alone, assist with it, fully automate it, or create new tasks around it. A nurse might use AI for patient monitoring, documentation and scheduling while keeping human-intensive tasks. That same nurse might also gain new responsibilities, such as reviewing AI-generated care plans.

In the more disruptive scenarios, some roles lose a large number of tasks to AI. Coders and customer-service workers are cited as examples of occupations that could face pressure to move into less-automated fields. That transition is not easy: it requires retraining and finding a genuinely different job.

Wage outcomes split sharply by scenario. In the substantial scenario, knowledge worker wages are essentially flat. In the extreme scenario, they fall by more than 10% by 2030. Workers in occupations less exposed to automation do better: productivity gains could stimulate demand in sectors like construction, where faster AI-assisted design and permitting might boost activity and lift wages for construction workers.

Who actually gets the gains?

The most striking finding is about the split between labour and capital. The model starts from a baseline where labour receives roughly 60% of economic output and capital receives 40%. As AI automates more tasks, capital’s share grows.

By the extreme scenario, capital captures 54.8% of GDP and labour takes 45.2%. That is a significant shift. A much richer economy does not automatically mean better pay or greater job security for the people doing the work, especially for those whose tasks AI replaces directly.

Anthropic draws a distinction here that is easy to miss in GDP-centric coverage: aggregate growth and individual worker outcomes can move in opposite directions at the same time.

Why it matters

Business owners and operators who rely on knowledge workers (developers, analysts, writers, customer-support teams) should take the wage and task-displacement findings seriously. Even in the moderate substantial scenario, which is more plausible near-term than the extreme one, growth roughly doubles its normal rate while knowledge worker wages stagnate. That suggests productivity gains flow to the business or to capital, not to the employee.

For anyone running AI integration projects right now, the task-level framing is practically useful. The question is not “will AI replace my team?” but “which specific tasks will AI handle first, and what does my team do instead?” That is a much more actionable question to answer today.

The capital-versus-labour finding also has implications for hiring strategy. If AI-assisted businesses capture more of the productivity upside, the competitive gap between companies that adopt early and those that do not will widen. We covered a related dynamic in our look at Cognition’s $2B raise at a $48B valuation, where the market is pricing in exactly this kind of productivity premium.

Our take

Anthropic is not predicting the future here, and the report is careful to say so. What the model actually does is force a cleaner conversation: stop asking “will AI help the economy?” (almost certainly yes) and start asking “who captures the benefit?” The answer, according to this model, is mostly capital owners, not workers.

For a business owner, that cuts both ways. If you own the systems and workflows powered by AI, you are on the capital side of that ledger. If your livelihood depends on billing hours for knowledge work that AI can replicate, the substantial scenario’s flat wages and the extreme scenario’s 10% wage drop are worth taking seriously now, not in 2028.

The task-level framing is the most practically useful part of this research. Map out which tasks in your business are already being touched by AI tools, which are next, and where human judgment still adds real value. That exercise alone is worth more than any GDP projection.

What to do about it

1. Audit your team’s work at the task level, not the job title level, and identify which tasks AI can already handle reliably today.
2. Redirect displaced task time toward higher-judgment work: client relationships, strategy, quality review of AI outputs.
3. Run small automation pilots (a workflow, a support queue, a reporting process) before committing to large structural changes. See our workflow automation service for how we approach this.
4. Track the labour-versus-capital split in your own business: are productivity gains showing up in profit margin or in team compensation? That ratio tells you how your business is positioned as AI adoption accelerates.
5. Revisit your hiring plans against these scenarios annually. The substantial scenario is already plausible on current timelines.

A bigger GDP number is not the same as a better outcome for your workforce. Plan accordingly.

Source: Bing News · Anthropic