Anthropic Forecasts Extraordinary AI-Driven Growth

Anthropic is laying out one of the most extraordinary economic scenarios ever seriously modeled by a major technology company.

In its most aggressive case, the AI lab estimates that artificial intelligence could push U.S. GDP to around $44.4 trillion by 2030, roughly one-third higher than it would otherwise be and equivalent to growth rates almost unheard of in a developed economy.

Anthropic co-founder Jack Clark brought that argument directly to policymakers at the IMF and World Bank meetings in Bangkok this week, saying rapidly improving AI could transform productivity, accelerate scientific discovery and vastly increase economic output.

The numbers are difficult to overstate.

Anthropic’s economic model examines several possible futures depending on how capable AI becomes and how quickly businesses adopt it.

In the modest scenario, U.S. GDP reaches about $34.1 trillion by 2030.

In the substantial scenario, it reaches $36.3 trillion.

In the extreme scenario, it reaches $44.4 trillion, measured in 2025 dollars.

That final scenario would represent something approaching 15% annual economic growth.

For comparison, advanced economies normally consider 2% to 3% annual growth healthy.

Growth at the pace Anthropic is modeling would look less like an ordinary business cycle and more like an industrial revolution compressed into a few years.

The mechanism is productivity.

Every economy ultimately depends on how much useful work people and machines can perform.

AI is beginning to automate or accelerate tasks that previously required highly educated human workers: coding, research, legal analysis, financial modeling, writing, customer support, data analysis and increasingly scientific work.

If those systems continue improving, companies would theoretically be able to produce much more output using the same amount of labor and capital.

Anthropic already sees that happening inside its own company.

The firm says AI agents were performing more total work hours than Anthropic’s human employees by March 2026. It has also reduced the proportion of junior positions in its hiring as AI systems take on more entry-level technical work.

That matters because Anthropic may provide an early glimpse of what highly AI-integrated companies eventually look like.

Its employees work with some of the most advanced models available, including systems that have not yet reached the broader public.

Their expectations for the future are therefore dramatically more aggressive than those of ordinary American workers.

Anthropic staff surveyed by the company expect AI capabilities and adoption to increase quickly enough that GDP could effectively double roughly every four and a half years by 2030.

That would represent one of the fastest periods of sustained economic expansion in modern history.

But there is an important distinction.

Anthropic is not predicting that this extreme scenario definitely happens.

Its model explores what the economy could look like if AI reaches certain levels of capability and adoption.

That makes the $44.4 trillion figure a scenario, not a forecast in the traditional sense.

Still, the fact that such numbers are now being seriously modeled tells us how quickly expectations around AI are changing.

Only a few years ago, the dominant debate was whether generative AI could write useful emails or generate acceptable computer code.

Now economists and technology companies are asking whether it could materially change the growth rate of entire countries.

The IMF is already seeing early signs of that effect.

It recently raised its global growth forecast partly because enormous AI-related investment in data centers, semiconductors, electricity and computing infrastructure is helping support the U.S. and Asian economies despite war and energy disruption elsewhere.

AI investment has become large enough to influence global GDP before many of the productivity gains from actually using AI have even arrived.

That is crucial.

Today’s AI boom is largely an infrastructure boom.

Companies are spending hundreds of billions of dollars on chips, data centers and power generation.

The much larger economic transformation would come if the resulting models begin performing significant portions of the world’s knowledge work.

That is the future Anthropic is trying to model.

Imagine a pharmaceutical company that can run years of research in months.

A software company that can produce ten times as much code with the same workforce.

A law firm where AI handles most research and document preparation.

An engineering company where models can design, simulate and test thousands of alternatives almost instantly.

A scientist working alongside AI researchers that never sleep.

Those productivity gains would compound.

Cheaper software makes other companies more productive.

Faster scientific research creates new products.

Automation lowers costs.

Lower costs create additional demand.

New industries appear around capabilities that previously did not exist.

That is how transformative technologies generate economic growth far beyond the industry that originally created them.

Electricity did not merely create electric companies.

It transformed factories, transportation, communications and households.

Computers did not simply create Microsoft and Apple.

They changed nearly every industry.

Anthropic is essentially arguing that sufficiently advanced AI could produce the same effect at much greater speed.

But there is an uncomfortable side to that scenario.

The faster the growth, the more disruptive the labor-market transition becomes.

Anthropic’s model finds that in its most extreme scenarios, knowledge workers face significant displacement and wage pressure even while society becomes dramatically richer overall.

The company’s own employees appear acutely aware of that risk.

Anthropic’s latest research found that people working closest to frontier AI expect far more automation of their own jobs than the general public does.

That creates perhaps the strangest possibility of the AI revolution.

The economy could become extraordinarily wealthy while large numbers of individual workers simultaneously feel economically threatened.

GDP could surge.

Corporate profits could explode.

Scientific progress could accelerate.

Products could become cheaper.

Yet many white-collar workers could discover that the skills they spent years developing are suddenly worth much less.

That is why the real economic question surrounding AI may eventually shift.

Today, everyone asks whether the technology will generate enough value to justify the trillions being invested in it.

If Anthropic’s most aggressive scenarios are even partially correct, the future question may be the opposite:

What happens if AI creates too much economic value too quickly for society to adapt?

For now, the extreme scenarios remain hypothetical.

But the direction is becoming harder to ignore.

AI is already driving semiconductor demand, data-center construction, electricity investment and corporate capital spending.

The IMF says it is already supporting global growth.

Anthropic says the next stage could be dramatically larger.

The world has spent the past two years debating whether the AI boom is a bubble.

If Anthropic is right, we may eventually look back and realize that the extraordinary part was not how much money was invested in AI.

It was how much economic output the technology ultimately created.