Anthropic’s Three AI Economy Scenarios, and What Your Business Should Do in Every One

TEG is an official member of the Anthropic Claude Partner Network. We build Claude Skills, Projects, and MCP integrations for B2B ops teams. See our full Claude services.

Anthropic published its Economic Scenario Explorer in September 2026. It is an interactive model of how AI could change US jobs, growth and unemployment by 2030. It was developed by Anton Korinek, Chad Jones, Szymon Sacher, Tess Cotter and Peter McCrory. We read it as a business owner would: what, if anything, should change on Monday?

How does the model work?

The model thinks in tasks, not jobs. Each occupation is treated as a bundle of tasks, and each task can stay the same, be helped by AI, be automated, or be replaced by new AI-related work. Visitors answer five questions:

1. What tasks can AI do?
2. How much do people use AI?
3. How much does AI do by itself?
4. How much more productive does AI make people?
5. How long does it take people to find a new job?

The model then shows the economy those answers imply.

What are the three scenarios?

The figures below are Anthropic's, showing the model's 2030 US GDP and its change.

Which scenario do people expect?

Anthropic surveyed people and found that the typical respondent's answers imply something close to the substantial scenario: GDP about 10% higher by 2030 than it would be without AI, and unemployment around 5%. About 10% of respondents hold views in line with the extreme scenario.

Anthropic is careful about what this means. Its page says, "We don’t know yet how AI will reshape the economy," and "The future is not predetermined." It also notes the model leaves out policy responses, business cycles, aggregate demand effects from the data center buildout, financial market disruptions and catastrophic risks.

What should a business do while nobody knows?

This part is our view, not Anthropic's. The page does not give guidance for individual companies. But the task-based approach maps well to how a business can prepare. These moves pay off in the modest, substantial and extreme scenarios alike:

1. Look at tasks, not job titles. This week, list the repetitive work in each team: updating the CRM, drafting proposals, summarizing meetings, reconciling invoices. That list is your starting point. At Offerpad, it began with one task, reviewing vendor photos by hand. Claude now checks them, and the renovations team saves 1 to 2 hours a day. Read the case study.
2. Build the workflows, and connect them. A good prompt helps one person. A workflow connected to your systems helps the whole team and keeps working after the novelty fades.
3. Train your people. The five inputs include how much people use AI. Adoption is a skill your team can build, and it grows with practice.
4. Make someone responsible. Workflows need an owner to maintain them, fix them when something changes, and improve them as Claude changes. Most businesses do not have a person for this. That is the gap Managed Claude fills.
5. Review on a schedule. The model's inputs, such as what AI can do, will keep moving. A monthly look at what is working and what to build next keeps you current without guessing.

That is how we work at TEG: assess, build, manage. We find the tasks worth automating with a free AI Value Assessment, build the solutions, and stay on as your Managed Claude team to keep them improving.

What does the model not tell you?

It is a national model, not a forecast for your company. It does not follow individual workers, and Anthropic describes version 1.0 as a simplification. Use it to think about ranges rather than to predict a number.

Frequently asked questions

What is Anthropic's Economic Scenario Explorer?

It is an interactive model from Anthropic that lets you enter your own predictions about AI capabilities, adoption, autonomy, productivity and job-transition time, then see what that implies for the US economy by 2030. Version 1.0 is dated September 2026.

Is Anthropic predicting one outcome?

No. Anthropic says, "We don’t know yet how AI will reshape the economy," and that "The future is not predetermined." The page presents three scenarios (modest, substantial and extreme) and lets visitors compare their own views.

What does this mean for a small or mid-sized business?

The page does not give guidance for individual businesses. In our view, the practical lesson is to act in ways that pay off whichever scenario arrives: find the tasks Claude can take over, build and connect them, train your people and keep improving them.

Should we wait to see which scenario happens?

We would not. The steps that help in every scenario, such as finding repetitive work, building reliable workflows and training your team, are useful today and get easier with practice. Waiting mostly means starting later.

Does the model account for everything?

No. Anthropic lists what v1.0 leaves out, including policy responses, business cycles, aggregate demand effects from the data center buildout, financial market disruptions and catastrophic risks. Treat it as a way to think, not a forecast.

The bottom line

Nobody, including Anthropic, knows which scenario will play out. You do not need to. Businesses that spend the next year finding the work AI can take over, building it well and training their people will be better placed in all three.

The assessment takes about 3 minutes and you see your results without entering an email.

Source: Anthropic, Scenarios for our Economic Future, Economic Scenario Explorer v1.0, September 2026. Figures and quotes are from that page. The recommendations are TEG's own.