AI Now Absorbs 77% of Venture Dollars. The Herd May Be Wrong

By Daniel Okafor

AI captured 77 percent of global venture dollars in the first half of 2026, up from about 27 percent when ChatGPT launched in late 2022. That figure comes from the 2026 Global Innovation Index, released last month by the World Intellectual Property Organization (WIPO). It is the sharpest signal in the report, and it frames a hard question for every budget owner: if everyone funds the same bet, who funds everything else?

Where the money actually went

WIPO's index ranks 139 economies on 79 measures of innovation. The list runs from research spending to patents to school test scores. The report also tracks where innovation investment flows. The headline numbers are blunt.

Global venture capital rose 28 percent in 2025 to $510 billion. That is the biggest annual jump since 2021. But the number of deals fell for the fourth consecutive year.

So the pool of capital is growing while the number of bets shrinks. More money is going into fewer deals. And most of it, WIPO's data shows, is going into the same category.

Soren Kaplan, a WSJ bestselling author and leadership strategy adviser, laid out the pattern in an expert opinion column for Inc.com. "Venture funding is rising even as the number of deals falls," Kaplan wrote. "Leaders should examine what their crowded technology portfolios are leaving unfunded."

That is the operational question. Not whether AI deserves capital. It likely does. The question is what a portfolio stops funding when one category takes three quarters of every venture dollar.

Why smart money crowds together

Kaplan asks why smart investors and innovators pile into the same place at the same time. One answer is conviction: AI is going to change the game, so it makes sense to participate. "Disrupt or be disrupted as they say," he notes. "But what's the risk associated with all this?"

The deeper answer is psychological, and it has a paper trail. Economists David Scharfstein and Jeremy Stein explained the dynamic in a classic paper in the American Economic Review. Their economic model showed a mechanism that still runs today: managers who worry about their reputations copy other managers' investment decisions, even when their own information says they should do something different.

The logic is simple. Being wrong together spreads the blame. Being wrong alone ends careers. As Kaplan puts it: "Being Wrong Together Feels Safer Than Being Wrong Alone."

When the future is uncertain, decision-makers watch what everyone else does. Following the crowd provides comfort. Reputation risk, not analysis, drives the trade.

What should leaders check this quarter?

Kaplan's column ends with a direct challenge to the reader: "Is more of your budget going to AI than it was two years ago?"

Run the audit. Compare today's technology portfolio against the 2024 version. Count the line items that grew because a rival funded the same thing. Then count the line items that died for lack of a champion.

The WIPO numbers make the stakes concrete. A 28 percent rise in venture capital, to $510 billion, means the money exists. A fourth straight year of deal decline means fewer places it lands. A 77 percent AI share means the concentration is extreme by any historical standard.

The pattern is not new. Scharfstein and Stein modeled it decades before ChatGPT. What is new is the scale: an entire asset class now behaves like the reputation-driven managers in their paper. The leaders who outperform will be the ones who can show, with numbers, which unfunded bets their crowded portfolios left behind.