The economy is about to get bigger. Your slice of it might get smaller. That’s the uncomfortable takeaway from a new model Anthropic’s economics team published this month, and it’s worth ten minutes of your time because the fix starts with your own budget, not Washington.
Here’s what the model actually says, where the scary headline number comes from, and five moves you can make this year.
What did Anthropic’s 2030 model actually find?
Anthropic, the company behind Claude, built a model of the US economy as millions of individual tasks and asked what happens as AI takes some of them over. It ran three scenarios for 2030:
Today, roughly 60 cents of every dollar the economy produces goes to workers and 40 cents goes to capital: the machines, software, and shareholders.
Every scenario grows the pie. What changes is who gets it.
Is the 45.2% number real, or is it clickbait?
It’s real, but it’s the extreme case. You’ll see “labor’s share drops to 45%” in headlines without that context, so let’s be precise.
The 45.2% figure comes from a scenario where AI is better than people at the vast majority of knowledge-work tasks and gets adopted fast. Anthropic itself says this would likely require AI that improves itself, and some of the economists who reviewed the model called it closer to a thought experiment.
The more grounded signal is the middle scenario. When Anthropic surveyed more than 10,000 Americans about what they expect AI to do, the typical answer landed close to “substantial.” About 10% of respondents expected something like the extreme case.
In the substantial world, here’s the part that should get your attention: knowledge-worker wages go flat, while other workers see raises. The model’s own example is coders and call-center agents moving into jobs like electrician and nurse.
⚠️ The realistic risk isn’t mass unemployment by 2030. It’s quieter: your raises stall while prices and everyone else’s pay keep moving. Keep reading, because move #3 is the one most people skip.
Who is most exposed?
The model splits work into tasks AI leaves alone, tasks it helps with, and tasks it fully automates. You’re most exposed if your day is mostly:
- Drafting and summarizing (reports, emails, briefs, documentation)
- Rule-based review (coding reviews, bookkeeping, contract checks, claims processing)
- Answering the same questions repeatedly (support queues, internal help desks)
You’re least exposed if your work needs hands, a physical location, or a human relationship: nursing, the trades, most in-person care, and jobs where someone has to be legally accountable.
Most people are a mix. That’s the point of the first move.
5 money moves to make now
Split your job into tasks and circle the automatable ones
Grab your last two weeks of calendar and to-do lists. Write down every recurring task, then mark each one: safe, AI-assisted, or AI could do this alone.
The US Department of Labor’s O*NET database lists the standard tasks for nearly every occupation, and it’s the same task list Anthropic’s model is built on. Look up your job title and compare.
If more than half your week sits in the last column, treat that as a financial risk, the same way you’d treat a job at a company with shaky earnings.
Become the person who runs the AI, not the person it replaces
In the substantial scenario, knowledge work doesn’t vanish. It gets done by fewer people who are much more productive with AI.
Pick one tool your industry is already adopting and get genuinely good at it. Not “I tried ChatGPT once” good. Good enough to show your manager a workflow that saves the team hours every week. That’s the skill that keeps you on the right side of a flat-wage market.
It doesn’t have to cost much. If you’re paying for an AI subscription to learn, make sure you’re not overpaying for it. Claude Pro annual vs monthly billing
Own a piece of the capital side
This is the move most people miss. If more of the economy’s growth flows to capital, the simplest hedge is to own some capital.
For most Americans, that means the boring stuff done consistently:
- Contributing at least enough to your 401(k) to get the full employer match, which is an instant return you shouldn’t leave on the table
- Investing through low-cost, broad-market index funds rather than trying to guess which AI company wins
- Automating a monthly contribution so it happens whether or not you’re paying attention
You don’t need to pick the winners. A broad fund owns a slice of the companies deploying AI, which is exactly the side of the ledger the model says is growing. (This is general education, not personal investment advice. Talk to a fiduciary advisor about your own situation.)
Build a bigger cushion than the old rule says
The model’s biggest risk isn’t the size of the economy. It’s the time it takes to switch jobs. In the more disruptive scenarios, displaced knowledge workers stay between jobs longer because moving into a new field takes retraining.
The classic three-month emergency fund assumes you’ll land something similar quickly. If your role is highly exposed, aim for six months of essential expenses in a high-yield savings account.
Quick way to find extra cash for that cushion: check whether your state is holding money that’s already yours. find unclaimed money in your state
Keep one off-ramp open
Nobody’s saying quit your office job to become an electrician. But the model is clear that demand, and pay, rise for work AI can’t touch.
An off-ramp can be small: a certification you could finish in a few months, a side gig in a hands-on field, or a skill that serves local customers directly. The goal is to have a Plan B that doesn’t depend on the same tasks AI is getting good at.
The bottom line
Anthropic’s model isn’t a prediction. The company says outright that the future depends on how fast AI improves, how quickly companies adopt it, and how the gains get shared. But the direction across all three scenarios is the same: more money flows to whoever owns the tools.
You can’t control the economy. You can control whether you’re only selling your hours, or also owning a slice of what’s replacing them.
You can plug in your own guesses about AI and see what 2030 looks like under them:
Your 15-minute checklist
- ✅ List your recurring tasks and mark which ones AI could do alone
- ✅ Look up your job on O*NET and compare the task list
- ✅ Pick one AI tool your field uses and schedule practice time this week
- ✅ Confirm your full 401(k) match is being captured
- ✅ Set up an automatic monthly investment, even a small one
- ✅ Check your emergency fund against six months of essential expenses
- ✅ Write down one realistic off-ramp skill or side income