or Gabriel Leydon, the Super Bowl was a showdown. But it wasn’t between the Seahawks and the Patriots. It was between the game he developed, Game of War: Fire Age, and its better-known competitor, Clash of Clans. Consistently in the top five apps worldwide by revenue, Game of War has been grossing more than $1 million a day, according to Think Gaming, double what it was bringing in just last spring.

But even at that rate, Game of War has persistently trailed Supercell’s Clash of Clans. And so on that Sunday in early February, the two games did battle via media buys. Both were armed. Clash of Clans brought a rapier: Liam Neeson, with a witty, irony-laden riff on a dude addicted to his fantasy duels. Leydon brought a nuclear bomb: Kate Upton. She emerged from some kind of medieval hot tub, and, donning body armor in slow motion, plunged into a melee of monsters and knights.

Critics mocked Upton’s performance, naturally, but in the real world, Game of War trounced its enemy. In the hours after the Super Bowl, the mobile game, for a few hours at least, passed its rival, as measured by rankings from Apple’s App Store. Which is exactly what Leydon had hoped for when the company he founded and runs, Machine Zone, spent $40 million on a four-month campaign with Upton as its spokeswoman.

Together, with a third game, UCool’s Heroes Charge, which also bought air time, the Super Bowl was a coming-out party for multiplayer games for iOS and Android phones. Unlike the games played on consoles such as Sony’s PlayStation or Microsoft’s Xbox, mobile games can seem pretty downmarket, with cheap-looking graphics and no immersive storyline. They are, however, beginning to dominate the entertainment market.

In Clash of Clans and virtually every other multiplayer mobile game, the contests and battles are one-on-one. But in Game of War, the combat is worldwide and simultaneous. The goddess Athena cheers on from the sidelines as players build cities, amass armies, and strike up alliances with other players, all of whom interact on a large map of kingdoms and battles.

The competition drives the purchase of virtual currency inside the game, giving players the ability to build things faster, making them more attractive candidates for alliances. And it’s those alliances—and the ability to play not just with dozens but thousands of people all over the world in real time—that give Game of War devotees a special little rush.

Such purchases are the nub of the “free-to-play” business, in which games are given away in the hopes that a few devotees who’ll pay to win will subsidize the whole enterprise. Like high rollers at casinos, these players are known as “whales.” In one case, Het Nieuwsblad in Belgium reported last year, a 15-year-old Game of War player from Antwerp spent €37,000 ($41,400) on in-app purchases on his mother’s credit card.

Those in-app purchases explain the revenue of $1 million or more each day. Last summer, the Wall Street Journal wrote that Machine Zone was expecting to bring in more than $600 million in 2014, a number Leydon won’t confirm or deny. The same article stated that he was in talks with JPMorgan Chase about fundraising at a valuation of more than $3 billion, which would match the $3 billion valuation Supercell reached with funding from Japan’s SoftBank.

“Culturally, it’s shocking to people,” Leydon says. “They say it’s just an app game. They don’t understand the scale of what’s happening here in the industry, because it’s kind of a sea change.” Indeed, those who play primarily on their phones make up 29 percent of the overall gaming audience, more than three times the size of the group that are considered hard-core console gamers, according to a marketing study released last spring by the NPD Group. Console and software sales, meanwhile, have dropped every year since 2008, says Michael Pachter, a gaming analyst at Wedbush Securities. In Leydon’s view, Upton vs. Neeson was just the first bout. “Next Super Bowl,” he says, “you might see 10 ads.”

Machine Zone has 300 employees, most of them in Palo Alto. Seated at a long conference table, Leydon, 35, looks cheerful if a little haggard, riding on three hours of sleep, working a heavy layer of stubble, and wearing the usual Zuckerberg-style hoodie. The office is the size of a warehouse, and in one dimly lit alcove, Leydon has stashed a relic from his past. It’s a coin-operated single-shooter arcade game from 2007 called America’s Army. Leydon grew up in Silicon Valley, spending whatever time he wasn’t on a skateboard at video arcades. After a detour through sound engineering school, he got his start designing games for Atari, Tsunami Visual Technologies, and others.

America’s Army was a stripped-down shoot’em-up adaptation of software the U.S. Army had been marketing as a recruitment tool. Leydon spent a year and a half developing it for arcade-game maker Global VR. Shortly before he was done, YouTube’s founders sold their company for $1.65 billion. “It was amazing,” he says, shaking his head. “Two young guys. They worked on it for, like, a year and a half. I was like, ‘That’s engineering, and what I’m doing is engineering. What’s the difference between what I’m doing and what they’re doing?’”

Leydon abandoned the arcade world, which was dying anyway, while holding on to one arcade-related skill set: the ability to design games that persuade players to spend money to keep on playing.

He was one of the first to see how the smartphone had the potential to be a virtual supermarket in everyone’s pocket and purse, but he had to fail a few more times before really nailing it. At the time—around 2007 and 2008—Facebook apps such as RockYou’s Zombies and a Stanford team’s KissMe were surging.

With his partners Mike Sherrill and Halbert Nakagawa, Leydon started a company called Addmired. “We were kind of afraid to compete on Facebook,” he says. “Zynga was 300 people. We were just three people.” So the company’s first apps were on Myspace, among them a photo-sharing precursor to Instagram called Addable. It was popular but unprofitable; Leydon noticed how the photos were secondary to the social aspect, and the most viral part of the app was how “it turned into a chat room.” But the partners were barely surviving. “We were working out of a 300-square-foot studio apartment. My co-founder’s bed was right next to me, and I had a $15 chair that permanently wrecked my back.”

In 2008, Addmired used a few thousand dollars of seed money from startup funder Y Combinator to begin work on what’s commonly recognized as the first free-to-play app for the iPhone, a largely text-based role-playing game called iMob, which they built in three months. There was no such thing as an “in-app purchase.” Instead, Leydon gave away a version, hoping enough players would get hooked and buy a paid version that unlocked virtual currency. “We finished it a month before we released it,” he says, “because I thought no one would want to play it.” When it was finally launched, iMob worked a little too well: Hundreds of thousands of people downloaded it. “The first day, we couldn’t believe what was happening. The second day, it was shocking. The third day, nothing worked.”

The system crashed. “We weren’t sleeping,” Leydon says. “We did everything possible to keep that game going for two or three months.” All the money Addmired made went out the door, spent on contractors to prevent crashes. “We did so many things wrong, just out of ignorance for how online games work, that it kind of became a mission in the company to solve that problem, to make sure we didn’t go through that again.”

Leydon recognized then that the future of mobile games wasn’t with fancy console-style graphics. “People were trying to put console games on the iPhone,” he says. “But it’s different. If your game’s just about graphics, it’s not going to work.” Smartphones, he realized, commandeer short bursts of attention; some people check their phones a hundred times a day. “They’re looking for something to do. They’re trying to fill time. And they’re jumping from app to app.” A successful game needed to require players to make quick decisions and then not be able to resist coming back umpteen times to check on the consequences of those decisions.

The popularity of iMob and a few other apps—and the blast of revenue Addmired enjoyed after Apple’s App Store opened in 2008—got Leydon noticed. The company, renamed Machine Zone in 2012, raised $13 million from Menlo Ventures, Anthos Capital, Baseline Ventures, and other sources to start developing what became Game of War: Fire Age, according to Dow Jones VentureSource. “What we’re doing,” he said then, “is creating a more social experience and then wrapping a game around it.” It took a team of 80 people 19 months to build the app. Still stinging from what happened with iMob, Leydon wouldn’t release the game until it underwent nine months of testing and five months of live beta.

To allow for the building of alliances within the game, the app needed a sophisticated chat feature. In the middle of development, Leydon decided it made sense for all players, no matter what their native tongue, to speak the same language. So he built a messaging platform that included a translation engine based not only on Google Translate but also on crowdsourcing. Within Game of War, text messages, comments, and push notifications are translated with about 70 percent accuracy. Log in to the game in Polish, and everything you read from other players is in Polish, too. The game’s players correct any errors they notice in exchange for virtual rewards within the game. From the start, the average player corrected 10 misspellings and slang expressions per day.

Leydon says the ability to translate slang in chat is nothing compared with the game’s other innovation. Game of War accommodates about 3 million users in simultaneous play, with what the company clocked as a 0.2-second response time. The closest thing remotely like it would be Eve Online, the PC-based virtual-reality environment that has its hundreds of thousands of subscribers cordoned off into separate servers called shards, which are able to keep tabs on what every player is doing at once. Because Eve maxes out at about 65,000 players per shard, Machine Zone abandoned that model altogether. “Game of War is built around a cloud kind of system,” Leydon says, “where the game doesn’t need to know where everyone is. It can process messages very fast, allowing synchronization to happen at very low latency speed.”

That means practically no lag time, language barrier, or limit on the number of people playing together. “It’s closer to a social network than it is a video game,” Leydon says. “Facebook has pokes and messages and things like that. In Game of War, you have attacks and friends and chats.” Leydon argues that Game of War’s social interplay is far more complex; among other things, Facebook interactions across language borders are limited.

“People look at ads and top grosses, and they don’t look at what Game of War really is,” Leydon says. “This is the largest real-time concurrent interactive application ever built. There’s nothing even close to it.”

Game of War is expanding along with the gaming audience that’s growing the most, what the NPD Group’s study calls “avid omni gamers,” who play on both their tablets and their phones and are most likely to spend money on in-app purchases. The segment is up 6 percent, to a 22 percent portion of the overall audience, according to the study. From the start, Machine Zone was recording long user times in Game of War. The average player, Leydon says, plays for two hours per day, in 12-minute sessions, 10 a day.

Even some of the people who don’t like Game of War—more than one reviewer finds its rules incomprehensible and its commercial appeals tacky in the extreme—appreciate how richly textured it is for the small percentage of users willing to immerse themselves in the experience. “The interesting thing about Machine Zone is in the mobile game space, they’re a very niche product compared to a Candy Crush Saga or even Clash of Clans,” says Patrick Walker of EEDAR, a market research company. “Their TV campaign was as broad as can be, with a supermodel and these fun battles. But the game itself is these deep relationships with alliances and resource management. It’s not like playing a shoot’em-up game or a sports game. The people interested in the game are willing to invest more time to understand these deep, complex rules.”

“I look at Game of War as a scale niche product,” says Leydon. “I don’t need to have everybody on the earth like it. I just need to make it for the people that do like it. That’s just a free-to-play rule, because only 3 percent of the people actually pay to play, and just quadruple that even stay around.” That means that of the tens of thousands of new downloads Game of War got on Super Bowl Sunday, Leydon only needs a few thousand players to become paying customers. “Arcade games were good training,” Leydon says. “You had to make something that made people want to keep going. Not even to pay, but to keep going again and again and again. That’s the challenge, and not many people are good at it.”

More empire-building games are entering the free-to-play market. Zynga, creator of FarmVille, has a new game called Dawn of Titans coming out this year. Leydon, meanwhile, intends to focus on what his new networking technology can accomplish outside the gaming world. He says dozens of companies have asked to license Machine Zone’s translation engine. Its applications, he says, span beyond gaming and into finance, logistics, social networking, and data analysis.

“We’re a technology company,” he says. “We’re not really a game company. What we accomplished here is actually where we’re going next. Getting so many devices to participate in the same experience at the same time—that’s going to be the most important part of the business.”

Robert Kolker is a Bloomberg Businessweek contributor. Follow him on Twitter @bobkolker.

Correction: The original version of this story incorrectly stated that Game of War has ‘no lag time.’ In fact, information cannot propagate faster than the speed of light, and simultaneity is always relative. The story has been amended.

---