When Moonshots Fall Back to Earth

Alphabet is in a pretty cushy place these days: Its revenue is growing, it has $78 billion in cash, and its $500 billion-plus market cap ranks among the top five of all publicly traded companies. The biggest complaint at the company’s latest shareholder meeting was the lack of free swag. To investors, Google’s parent can do no wrong. And even if it did screw up, investors could do little: Alphabet’s dual-class voting structure prevents shareholder meddling.

Such dual-class shares are meant to free companies to experiment—and fail—by handing control to insiders. Companies can invest in long-term innovation instead of obsessing about short-term gains. At the heart of Alphabet’s long-term, fail-friendly strategy are its “moonshots,” or wildly ambitious and maybe impossible research projects. Since 2014, those moonshots, from Google (GOOGL) Glass to self-driving cars, have gotten more attention than the company’s boring old ad business, responsible for 90% of overall revenue.

Alphabet executives always made clear that many moonshots would fail. They never hinted that projects could be killed from within.

Last year, when Google restructured as Alphabet, the corporate rejiggering felt meaningless. A year later, with the new structure revealing big losses on Google’s “Other Bets” (the non-ad businesses), another narrative has emerged: The carefree attitude about experimentation has shifted, and that shift is being felt by money-losing projects.

Reports link the new cost-conscious regime of chief financial officer Ruth Porat, a former Morgan Stanley (MS) CFO, with the departures of executives like Google Ventures CEO Bill Maris and Jeff Huber, who helped create lab X (see “The Wall Street Veteran Who’s Helping Google Get Disciplined”).

Meanwhile, the com­pany’s most viable moonshot projects are seeking joint ventures to share the cost of turning research into real businesses. In January, Verily, Alphabet’s life-sciences moonshot, partnered with GlaxoSmithKline (GSK), while its Calico “life-extension” project allied with AbbVie (ABBV).

For more on Google, watch this Fortune video:

Alphabet’s most promising bet lately isn’t a moonshot—it’s in the cloud, where the company is playing catch-up to Amazon (AMZN). Alphabet is willing to spend freely on cloud computing, including $725 mil­lion for two new software acquisitions. The deals are investments in the future, just different from what we’re used to reading about.

Alphabet’s moonshots are returning to earth. For now, the bean counters have won.

A version of this article appears in the October 1, 2016 issue of Fortune.