Since 2000, asset values have risen much faster than GDP; that’s a boon for those who have assets to begin with but doesn’t do much for those who are just getting started and need broad-based income gains. Only about a quarter of the wealth generated came from investment, the remainder was mostly on paper. Moreover, debt has soared, with every dollar in investment generating $1.90 in debt. Wealth inequality is entrenched, with the top 1% in major economies accounting for at least 20% of wealth. And, finally, international financial imbalances are growing, contributing to today’s touchy trade and political environment.  

The size and shape of the imbalance differ from place to place. But there are two commonalities. First, rapid productivity growth is the most effective counterweight to today’s tilted profile. And second, artificial intelligence (AI) can help—up to a point. For AI to fulfill its potential, countries must not only position themselves to benefit from its capabilities in technological and business terms, but also macro-economically. Otherwise, it would be like eating a heaping bowl of carbohydrates to fuel a workout—and then skipping the gym. The results won’t be pretty.

Consider the United States. It is at the forefront of AI-related innovation, investment, and adoption. To keep up the positive momentum, however, it needs to save more (i.e., borrow less).  The national debt is, almost 120% of GDP, more than double what it was in 2000. If annual budget deficits keep growing, potentially higher inflation, interest rates, and long-term uncertainty could destabilize the economy and threaten the investment needed for a continued AI boom. Wealth could erode by almost $100,000 per capita in real terms by 2033.

While AI-based growth might boost fiscal revenues, there are also ways that AI might exacerbate US fiscal challenges. If labor market disruptions are significant, that could drive up related costs, such as unemployment insurance. Higher productivity, even if concentrated in a handful of sectors, pushes up wages generally; but that would translate into higher public-sector labor costs. Plus, many social benefits are tied to income. The bottom line: more AI without a healthier fiscal picture could simply add to the stress.

In China, the challenge is different; the economy needs to save less and consume more. In the wake of the prolonged downturn in the property market, Chinese households have bulked up their deposit savings—on the order of nearly seven percentage points of GDP compared to the their average levels in the 2010s. Deflation has ensued. Meanwhile, private corporate investment has slowed sharply, down to 1% of GDP a year recently, compared to 7% from 2017-21. While investment by state-owned enterprises (SOEs) has risen, these entities are much less productive. Overall, 23% of China’s industrial enterprises are losing money, the highest figure in more than two decades. Economic growth is therefore largely being driven by net exports, which is tricky given mounting international pressure to reduce imbalances in trade and investment. The most promising alternative is for consumers to spend more of their income.

China is an acknowledged leader in AI. Translating that into growth requires new business models and reform to unlock new household demand.

In Europe, meanwhile, the threat of prolonged economic stagnation is real, with an economy characterized by debt-cutting households, fiscal constraints, sluggish investment, weak productivity and falling interest rates. It needs to step up corporate competitiveness. So it needs to invest more, particularly in AI-driven innovation and infrastructure. That could mean becoming home to leading AI companies or nurturing the development of standout firms that can use it most dynamically. Research shows that the greatest economic impact from AI will likely come from a few firms going “all in, “rather than many of them making small bets.

At the moment, though, Europe is trailing far behind both China and the United States. According to a recent McKinsey-World Economic Forum report, it is globally competitive in only four out of 14 critical technologies, and accounts for just four of the world’s top 50 tech companies. The McKinsey Global Institute has estimated that large European companies face an investment gap of $700 billion a year in R&D and capital expenditures compared to their US counterparts. Corporate investment in Europe has declined relative to GDP since 2019, and also delivered about 25% lower returns than in the U.S.

What does this mean for companies? CEOs need to understand the swing factors—less borrowing in the U.S., more investment in Europe, more consumption in China, each at a scale of more than 3% of GDP—that can move the long-term trajectory of wealth and economic growth, and plan for each. But they are not bystanders in driving productive outcomes; they are the engines for both growth and widespread AI adoption. And the actions of just one CEO can make a big difference: only a few dozen firms drove the majority of productivity growth in economies including the United States, United Kingdom, and Germany over the past 15 years.

AI could be the disruption of the century, and a positive one, delivering broad-based growth through greater productivity. But that is far from inevitable. If AI is to be a platform for prosperity, countries and companies must ensure that the foundations are strong.  

The opinions expressed in Fortune.com commentary pieces are solely the views of their authors and do not necessarily reflect the opinions and beliefs of Fortune.