And then broadband Internet service began creeping across the nation. Game publishers began experimenting with a different model, allowing people to download titles directly to their gaming systems. There were issues. Servers were overloaded when a big game would drop. Gamers became angry when publishers revoked licenses to the games the players thought they owned. But both sides adapted, Toys “R” Us vanished, and GameStop started searching for a buyer. This was the second age of video games, and it, too, was (mostly) good.

And then Daniel Ek and Reed Hastings said, Let there be streaming. And gaming companies, keen on the success of Spotify and Netflix and their peers, began exploring how to use cloud-computing technology to deliver games instantly and on demand to players. This is the third age of video games, and some believe it could be the best of them all.

“The greatest disruption of entertainment is the combination of streaming and subscription,” says Andrew Wilson, CEO of Electronic Arts. “More people are engaging, with less friction, through cloud-driven services.”

Gaming companies are stepping carefully into the future. Consider two prominent players: Electronic Arts and Sony. In May, EA snapped up the cloud-streaming technology of GameFly, a game-rental service. The next month, EA showed off a prototype of a streaming service to journalists at E3, the video game industry’s annual trade show. Sony, meanwhile, offers PlayStation Now, a four-year-old service that gives players instant access to a library of 650 games—mostly older, catalog titles—for $99 a year. In 2012, Sony acquired Gaikai, an early game-streaming service; three years later, it bought Gaikai rival OnLive.

Such maneuvers make possible an all you-can-eat pricing scheme for video games that, at first blush, seems ill-advised—a $100 annual subscription inherently devalues individual titles that retail for $60 each. But in the long term it transforms the business model into one built on incrementally rising fixed costs yet exponentially rising revenue and customer numbers. “It’s not unreasonable for us to believe that, for what is essentially the same investment in the creation of the content, we might entertain an additional 100 million players through subscription than we would in the traditional model,” EA’s Wilson says.

Ubisoft CEO Yves Guillemot agrees. He too believes that streamed video games will become a significant part of his company’s earnings within five years. But he disagrees with Wilson on the specifics. Guillemot argues that a Netflix-like platform business model won’t work for games, preferring instead a “publisher channel” model that allows each game publisher to maintain a direct relationship with its players. “I believe there will be lots of services with lots of different types of games,” Guillemot tells Fortune.

Some publishers, mindful of early technical challenges that made for poor streaming game play, aren’t as bullish. “It’s exciting, but we really don’t think it’s a game changer,” says Strauss Zelnick, CEO of Take-Two Interactive. Nintendo feels similarly. “In the end, the consumer wants a great experience,” says Reggie Fils-Aime, president of Nintendo of America. “We don’t think it’s quite there yet.”

There is risk in a wait-and-see approach. Some of technology’s giants are investing in ways that could put them in the catbird seat for a $36 billion industry. Amazon, Microsoft, and Google maintain the world’s top cloudcomputing platforms; their ownership of Twitch, Xbox Live, and YouTube, respectively, gives them enormous audiences to leverage. And don’t count out Tencent, the Chinese conglomerate that owns stakes in Activision-Blizzard, Epic Games, Ubisoft, Riot Games, and two Chinese live-streaming platforms similar to Twitch: Douyu and Huya.

In other words, this game is far from over—music to the ears of hundreds of millions of potential gamers around the globe.

A version of this article appears in the September 1, 2018 issue of Fortune with the headline “The Great Game In The Sky.”