After missing TikTok and AI, Zuckerberg is making smaller bets
Meta is abandoning billion-dollar moonshots in favor of a portfolio of experimental apps, starting with a prediction market built around virtual points.
In recent years, Mark Zuckerberg has watched other companies capture markets that could have belonged to Meta, one after another. TikTok became a formidable competitor by perfecting short-form video and the algorithmic feed. Companies such as OpenAI and Google launched successful AI chatbot services, even though Meta sits on a treasure trove of valuable user data. Zuckerberg, meanwhile, has often found himself reacting too late or defending against rivals, while also dealing with costly and ultimately unsuccessful bets such as the metaverse initiative, which never took off and has now effectively been abandoned.
Now, having learned from those setbacks, Zuckerberg is pushing Meta in a new direction that could allow it to move more quickly and flexibly into emerging markets. The strategy centers on developing standalone applications that capitalize on new online trends and user behaviors, helping Meta expand its user base and prevent, or at least mitigate, another TikTok-style disruption.
The first application expected to emerge from this effort is a prediction market designed to compete with platforms such as Polymarket and Kalshi, with one fundamental difference: users will not wager real money.
The previous attempt failed
Prediction and betting markets have attracted significant attention in recent months, particularly following cases in which users allegedly exploited insider information to profit from major events, including reports surrounding the capture of Venezuelan President Nicolás Maduro and the outbreak of the war with Iran.
Despite these controversies, the sector continues to grow rapidly. This week, reports emerged that Kalshi is preparing for an IPO, potentially in late 2027, after annual revenue surpassed $2 billion, driven in part by surging activity surrounding the NBA playoffs and the FIFA World Cup.
The success of these markets has attracted new competitors. Sports betting giants FanDuel and DraftKings have recently begun offering prediction markets of their own. In traditional sports betting, gamblers bet against the house, which sets the odds. Those odds are structured to favor the operator, which profits from bettors' losses. In prediction markets, by contrast, users bet against one another on binary outcomes, whether an event will or will not occur, and odds fluctuate dynamically based on market demand. Platform operators generate revenue by collecting commissions on trades, and the popularity of these markets has already pushed annual trading volumes into the billions of dollars.
According to a report by The New York Times, Zuckerberg now wants Meta to enter the market as well. Employees familiar with the matter told the newspaper that he recently assembled a small team to develop a smartphone app similar to Kalshi and Polymarket, but with a twist. Instead of wagering real money, users would bet with virtual points, although Meta has reportedly not ruled out introducing real-money betting in the future.
The app, currently being developed under the internal codename Arena, will operate separately from Facebook, Instagram, WhatsApp and Messenger. However, Meta intends to use its existing platforms to drive users toward the new service. Arena remains under development, and Meta could ultimately decide not to launch it. Internally, the project is considered experimental, but also a high priority.
The decision to rely on virtual points appears intended to avoid one of the biggest controversies surrounding platforms such as Kalshi and Polymarket. Because real money is at stake, these markets have repeatedly been linked to allegations of insider trading. In February, an Israeli citizen and an IDF reservist were accused of using insider information to place bets on military operations through Polymarket. In April, a U.S. soldier was accused of earning $400,000 by allegedly using confidential information to bet on the capture of Venezuelan President Maduro.
Removing the monetary element would largely eliminate the incentive for such behavior on Meta's platform. It would also reduce the likelihood of incidents such as the one involving journalist Emanuel Fabian (The Times of Israel), who reported being threatened by anonymous individuals demanding that he alter a report about the interception of an Iranian missile so they could win a bet on Polymarket.
Will regulators object?
Meta is apparently betting that the competitive instinct, and the psychological appeal of accumulating virtual points or currency, will be enough to preserve user engagement even without real-money gambling.
On the other hand, eliminating financial stakes could make the platform more vulnerable to manipulation. Today, major media outlets routinely cite prediction market odds from Polymarket and Kalshi as indicators of public sentiment or collective expectations, largely because participants have money at risk. Without financial consequences, and especially if Meta's platform attracts a massive user base, the incentive to manipulate predictions for political, ideological or reputational reasons could increase substantially.
There is, of course, no guarantee that Meta's prediction market will succeed. The company has tried something similar before. In 2020, it launched Forecast, a crowdsourced prediction platform where users forecast future events using a points system. The service never evolved beyond an interesting experiment and was shut down in 2022.
Now that prediction markets have become one of Silicon Valley's hottest trends, Meta appears hopeful that timing will make the difference. However, any attempt to enter the sector is likely to face scrutiny from regulators and lawmakers, who already view the company with deep skepticism.
Democratic Senator Richard Blumenthal has already made clear that Meta's latest initiative will not escape congressional attention.
“Meta copied slot machines to addict kids to Instagram. Now Zuckerberg is turning his company into a prediction market,” he wrote on X. “Meta’s business model is profiting from addiction—kids, gamblers, & more. Stop it through KOSA & my prediction markets bills.”
Meta's prediction market is part of a much broader strategic shift, reflecting the company's increasingly uncomfortable competitive position. It trails TikTok in short-form video and popularity among younger users, spent tens of billions of dollars on the ultimately unsuccessful metaverse strategy, and remains behind OpenAI, Google and Anthropic in generative AI, even as it commits tens of billions of dollars annually to AI infrastructure.
Meta's AI-powered smart glasses have been one notable success, selling millions of units and capturing roughly 80% of what remains a relatively small market. But with Google already unveiling competing products and Apple expected to enter the category by 2027 at the earliest, Zuckerberg knows he cannot rely on smart glasses alone to drive Meta's future growth.
Meanwhile, Meta's core platforms, Facebook, Instagram, WhatsApp and Messenger, already reach about 3.65 billion daily active users, leaving relatively little room for further expansion. The company needs new engines of growth, and quickly. This time, Zuckerberg appears unwilling to stake Meta's future on a single massive, expensive vision as he did with the metaverse.
Against this backdrop, The New York Times reported that Zuckerberg has instructed multiple teams inside Meta to develop several standalone applications across different categories. The idea is to experiment in adjacent markets without tying every initiative directly to Meta's existing platforms.
Alongside the prediction market, another project reportedly under development, internally known as Meta Photos, would use artificial intelligence to create new forms of digital media. Beyond its existence, however, little is currently known about the application or the company's broader portfolio of experimental apps.
These projects are bets as well. Unlike the metaverse, however, they are relatively small, grounded in existing market trends rather than an ambitious attempt to reshape the internet itself. Zuckerberg's strategy now appears to be diversification: launch multiple products, identify what gains traction, and double down on the winners.
Most of these experiments will probably fail. But after watching TikTok redefine social media and OpenAI reshape AI, Zuckerberg appears to believe that finding just one breakout success could make the entire strategy worthwhile.
