The stock market is making an unusually large bet on artificial intelligence, but the debate has become too focused on the wrong question. Investors are not simply deciding whether AI is real. They are deciding whether the profits created by AI will arrive quickly enough, and broadly enough, to justify the enormous amount of capital already flowing into the technology.

The market isn’t simply betting on artificial intelligence. It’s betting that one of the largest corporate investment cycles in modern history will produce returns commensurate with its cost. The largest technology companies alone are expected to spend roughly $715 billion in capital expenditures this fiscal year, much of it tied directly or indirectly to expanding AI capacity.

Every dollar of that spending is someone else’s current revenue. The investment question is how much ultimately becomes someone’s permanent earnings power. So far, the market has rewarded the companies building the infrastructure. Semiconductor companies, cloud providers and data center operators have become the primary beneficiaries of an investment cycle unlike anything in modern corporate history.

The important distinction from the late 1990s is that today’s leaders are not selling promises without profits. Many of the largest AI beneficiaries are among the most profitable companies in the world, generating hundreds of billions of dollars in annual revenue, producing enormous free cash flow and funding much of this expansion themselves.

That difference matters. The S&P 500 is currently trading a little past 20X forward earnings, above its long-term average, but the valuation story is not simply one of investors paying higher multiples for the same earnings. The market has earned part of that premium because expected profits have risen alongside stock prices. Unlike the late 1990s, when valuations expanded dramatically ahead of earnings, today’s market is being supported by a forecast for sustained corporate profitability.

The harder question is what happens after the infrastructure buildout. Capital investment creates revenue for suppliers, but it does not automatically create durable economic value for shareholders. Every dollar a hyperscaler spends on AI infrastructure becomes revenue for chipmakers, equipment suppliers and data center operators. The critical question is whether those investments ultimately produce higher productivity, stronger margins and better returns on capital for the companies making the investment, or whether competition gradually transfers much of that value to customers.

Every major investment cycle passes through a similar transition. The first phase rewards the companies building the infrastructure. The second rewards the companies that prove they can convert that infrastructure into durable economic value. Markets rarely struggle to identify transformative technologies. They often struggle to identify where the profits ultimately settle.

That is where today’s expectations become demanding. The S&P 500 is currently trading near 20 times forward earnings, above its long-term average, but investors are not paying that premium for stagnant profits. Analysts currently expect S&P 500 earnings to grow roughly 17% next year, a rate that would rank among the stronger earnings environments of the past several decades outside of early-cycle recoveries.

Those expectations may ultimately prove justified. AI could deliver meaningful productivity gains, lower operating costs and entirely new sources of revenue across the economy. But the evidence investors need to watch is whether today’s extraordinary capital spending begins translating into broader improvements in corporate margins and returns on capital, rather than simply moving revenue from one technology company to another.

The market is not repeating 1999. The companies leading this cycle have real profits, real customers and real demand. The risk is different. Investors may not be financing imaginary businesses. They may be paying today’s prices for tomorrow’s economic gains before the evidence is fully visible. Over the next several quarters, the question is not whether AI headlines continue. It is whether today’s extraordinary capital spending begins showing up in broader productivity, stronger margins and higher returns on capital. That is where this investment thesis will ultimately be confirmed—or challenged.