Is AI a circular money machine? 3 reasons not to worry. – Crypto News

AI looks like a circular money machine. Microsoft owns a major stake in OpenAI, which in turn invests in AMD. Nvidia puts billions of dollars into OpenAI and holds equity in CoreWeave, one of Microsoft’s cloud suppliers. The same dollars are simply bouncing between balance sheets, rather than creating new economic activity. Equity markets are soaring, paying little mind to this web of financing.

So alarmists say. But the evidence suggests something different, a powerful technological transformation that remains grounded in fundamentals. There is little reason to panic about an AI bubble—at least not yet.

A bubble requires euphoric overvaluation and a disconnect between price and performance. Today’s AI boom shows neither. The market is concentrated in highly profitable incumbents with real earnings, strong cash flows, and global-scale innovation. There is plenty of excitement, but that is different from irrational exuberance.

It is true that AI-related companies, specifically those in the Magnificent Seven, account for roughly three-quarters of S&P 500 returns, 80% of earnings growth, and 90% of capital-spending growth since late 2022. No wonder more than half of fund managers surveyed by Bank of America called AI stocks a bubble and ranked an AI-driven equity bubble as the biggest risk facing markets. This seems doubly alarming given the financial links between these firms.

Yet these links are more likely just signs of evolving corporate strategies and risk management. AI is still sorting out its business models. Partial equity stakes allow companies to share risks and profits as they figure out how to leverage layers of the AI stack—chips, data, software, or applications—to deliver the most value. Antitrust regulators likely would have blocked full mergers, so looser forms of vertical integration may be the only practical way for these firms to innovate across corporate boundaries.

Global spending on AI is expected to top $375 billion this year. Spending on data centers alone could hit $3-4 trillion annually by 2030. Some see echoes of the late-1990s telecom and internet boom, when fiber-optic firms were valued by the number of cities they promised to serve, not the profits they actually earned.

AI companies are different. Most AI-related capital spending is financed through free cash flow, not debt. The Mag 7 have strong balance sheets, post average returns on equity near 46%, and trade at roughly half the earnings multiples of dot-com era darlings.

James Pethokoukis, my colleague at AEI, has noted that today’s valuations reflect solid profits, “fortress-like balance sheets,” and disciplined forecasts, not blind exuberance. The AI capital expenditure boom looks more like a race than a bubble—one fueled by genuine technological advances and intense global competition, particularly with China.

Unlike the internet bubble, where clicks stood in for cash, AI is already producing measurable economic gains. Small businesses using AI see productivity gains approaching 40%. AI systems have long powered Amazon’s logistics, Netflix’s content suggestions, and Meta’s advertising efficiency. Generative tools are cutting costs in design and coding. And large-language-model users are beginning to bypass traditional search engines entirely, creating new advertising markets for AI companies.

Still, investors and policymakers should keep an eye on a few pressure points. First, profitability. Some firms experimenting with AI are falling short. Many by now are familiar with the MIT study that found 95% of organizations report earning zero return on investment on their AI projects, largely from their failure to make meaningful change.

Power is another problem. Data centers are already straining U.S. electric grids. Their power needs are expected to triple by 2028. Yet some utilities hesitate to expand capacity without support of federal regulators, who are worried about impacts on rates.

Third, regulation: Europe is trying to stymie U.S. AI companies with a regime of fines, and China is a politically risky market.

The real test for AI will be in how firms respond to disruption—whether by shrinking from it, as the MIT study found, or by embracing it, as Amazon and Goldman Sachs have done.

For now though, what we are witnessing doesn’t look like a bubble. It looks, to me, more like the early stage of a decades-long technological supercycle. The challenge isn’t unrealistic dreams, but too little electricity to power them and too much regulation that squelches them.

Guest commentaries like this one are written by authors outside the Barron’s newsroom. They reflect the perspective and opinions of the authors. Submit feedback and commentary pitches to ideas@barrons.com.