The U.S. Treasury is wrapping up two days of debt auctions, with Wednesday seeing a $39 billion sale of 10-year bonds and Thursday capping off with a $22 billion sale of 30-year bonds. The government regularly holds such auctions to fund its operations, and typically only wonks and bond traders mind the numbers. But lately, more and more people are tuning in.

Behind the growing attention is a particular worry. What if buyers stop stepping forward? What if demand falls – and falls hard?

Treasury auctions happen online. Investors say how much they want to buy and the minimum yield they’ll accept. The government fills the lowest-rate bids first, working up until all the bonds are sold. Everyone whose bid is accepted gets the same final rate — the highest yield the government is willing to accept.

For now, bidders are still showing up in force. Wednesday’s $39 billion sale of 10-year notes was well received, with yields only slightly below expectations and signs of healthy demand. Though one auction did spark volatility across the broader market, April and May’s auctions, held in the wake of Trump’s tariff-heavy “Liberation Day” announcement, also saw investor appetite below panic-button levels.

Fears of market turmoil

Still, fear persists that one weak auction could send bond yields spiking and markets tumbling.

That’s especially true for Thursday's 30-year auction, where yields hover just below the psychologically significant 5% line. A bad sale could push them over the edge again. A strong one might calm nerves.

But it’s not just the usual market mechanics in play. The rising scrutiny of these once-sleepy, even boring events reflects widespread unease with the state of U.S. debt. With deficits ballooning, interest costs rising, and President Donald Trump proposing a second round of tax cuts, auctions like Thursday's are becoming tests for global faith in American fiscal discipline. That is, solvency.

“The durability of the ‘sell U.S. assets’ trade may be overstated,” BMO Capital Markets' Ian Lyngen said. Still, the fact that anyone outside of fixed-income desks is pondering that trade at all marks a significant shift.

For now, demand for U.S. debt is holding up. But the question of “how long can this last?” is no longer just for bond geeks. It’s becoming a national one.

Japan’s bond troubles may be a warning sign

One reason investors and commentators are watching closely because they’ve seen what can happen when demand falters — and seen it recently. In Japan, a series of poorly received auctions has shaken the country’s historically stable bond market. Yields on 30- and 40-year Japanese government bonds have soared to multi-decade highs, and insurers who hold them are facing tens of billions in paper losses.

Now the Japanese government is considering direct intervention. According to Reuters, Japan may begin buying back super-long bonds issued at lower interest rates in an attempt to absorb some of the over-abundant supply. The plan is still under consideration, with the hope being that such buybacks would calm a market growing increasingly anxious about Japan’s rising debt load.

Despite assurances from Tokyo, demand for ultra-long bonds continues to deteriorate. What’s more, because Japan’s debt-to-GDP ratio is about 260% (and the Bank of Japan is already sitting on more than half the market), the country has few good options.

The fear is that the U.S. could one day face similar pressure: too much supply, not enough demand, and a loss of faith from investors who’ve long treated U.S. government debt as one of the world’s safest assets.

Boeing stock was down some 6% Thursday morning after a fatal crash involving one of its 787-8 Dreamliner aircraft.

The plane, operated by Air India, tragically crashed just minutes after takeoff from Ahmedabad, killing many of the people onboard and causing additional casualties on the ground, according to Indian officials. It's not yet clear if there are survivors from the flight itself.

The plane was bound for London’s Gatwick Airport when it went down in a densely populated area, hitting residential buildings, a medical college, and offices near the airport. The cause of the crash has not yet been determined.

The event marks the first fatal crash involving a 787 Dreamliner since the aircraft entered service in 2011. In addition to the human toll, it delivers a devastating blow to what’s been thought of as one of Boeing’s most modern jets. While the Dreamliner faced some battery-related groundings early in its history, it has, until now, avoided the kind of tragedies that plagued Boeing’s 737 MAX program, which saw two deadly crashes in 2018 and 2019, and additional issues in 2024.

Spokespeople for Boeing have said the company is in contact with Air India and assisting investigators.

Safety questions surge back into focus

Just days ago, a raft of articles speculated about whether or not Boeing's turnaround might finally be turning. Now the crash revives old concerns and surfaces new ones just as recently-installed CEO Kelly Orthberg was attempting to restore public confidence.

Shares of key Boeing suppliers, including Spirit AeroSystems and GE Aerospace, also dropped approximately 2% each. Some analysts described the stock moves as knee-jerk reactions as fears about Boeing’s safety issues take hold once more.

As emergency responders remain on the scene in India, aviation specialists now turn to black box data for signs of whether this is an isolated incident or the start of another painful chapter.

Boeing’s response will be closely watched, too.

The future of robotics may be humanoid — but Tesla says this move was all too human. The EV giant is suing a former engineer for allegedly stealing trade secrets from its Optimus program and founding a rival AI robotics firm days later.

As Bloomberg reported, Tesla has filed suit against Zhongjie “Jay” Li, a former engineer on its Optimus humanoid robot program, accusing him of stealing proprietary designs and launching a rival startup in the heart of Silicon Valley. According to Tesla, Li didn’t just walk away with company secrets — he allegedly copied some of the robot’s most complex components and recreated them under a new name: Proception Inc.

The suit, filed Wednesday in federal court in San Francisco, paints a picture of quiet defection and stealth data siphoning. 

Tesla claims Li downloaded sensitive files — specifically those related to robotic hand sensors — onto personal devices in the weeks leading up to his September 2024 departure. Just six days later, Proception was incorporated. Within five months, Proception was demoing a five-fingered “ProHand” that Tesla calls a dead ringer for its design — or, “hands that bear a striking resemblance to the designs Li worked on at Tesla,” according to the complaint.

Now, Tesla wants damages and an injunction blocking Li from using what it has called the most sophisticated hand ever made. Tesla’s Optimus program, still in early development, is aimed at creating general-purpose robots capable of factory labor, household tasks, and even child care. CEO Elon Musk said in May that the robot will be “the biggest product of all time.”

Robot wars

The humanoid robot industry, long stuck in the uncanny valley, is heating up as AI and sensor tech finally catch up to sci-fi ambition. 

Several companies, including Figure AI (which just landed investment from Amazon and OpenAI), 1X Technologies, and Apptronik, are racing to deliver general-purpose bots. But making a truly dexterous robotic hand — one capable of gripping, pinching, and manipulating small objects — is one of the field’s holy grails.

That’s the very capability Li allegedly reproduced at Proception.

He was “entrusted with some of the most sensitive technical data in the program,” Tesla’s lawyers said in the filing, and Li's “conduct is not only unlawful trade misappropriation — it also constitutes a calculated effort to exploit Tesla’s investments, insights, and intellectual property for their own commercial gain.”

If Tesla’s allegations hold, the event would reprise a familiar plotline. The company has sued at least three former employees since 2020 for hauling data off to Zoox, XPeng, and most recently, Rivian; all those cases were settled before trial.

On the company’s earnings call in April, Musk said that he expects the scale-up of the Optimus robots to be faster than any of the company’s other products — a million units per year in less than five years. Musk claimed that there would be thousands of the robots working in Tesla factories by the end of the year and promised that the Optimus robots would help produce a time of “sustainable abundance for all.”

The timing is awkward for Musk, who has been pounding the table about Optimus for months, predicting “thousands of Optimus robots working in Tesla factories by the end of this year” and, eventually, “millions per year.” Last week, Optimus’ head of engineering, Milan Kovac, abruptly quit “to spend more time with family abroad,” forcing Autopilot chief Ashok Elluswamy to take the reins.

Losing a top lieutenant while fending off alleged intellectual-property leaks underscores how critical — and fragile — Tesla’s robotics push has become.

Meta is suing a company that reportedly ran thousands of ads on its network to promote a so-called "nudify" app. The lawsuit was filed against Joy Timeline HK Ltd., the Hong Kong-based entity behind the app CrushAI, which enables users to generate fake and non-consensual sexually explicit images of other people.

The legal action follows multiple attempts by Joy Timeline to bypass Meta’s ad-review process, after the developer's ads were repeatedly removed for breaking the rules, Facebook's owner said in a statement Thursday. 

The lawsuit also follows a letter sent by Sen. Dick Durbin, D-Ill., in February to Mark Zuckerberg, urging the CEO to address his company’s role in letting Joy Timeline run ads that violate Meta’s policies. 

Durbin cited a report by tech news outlet 404 Media and research by Cornell Tech’s Alexios Mantzarlis that found that at least 8,010 CrushAI-related ads ran on Meta’s apps during the first two weeks of January.

Meta updated its policies regarding non-consensual intimate imagery over a year ago, to make it “even clearer” that promoting nudify apps is prohibited. 

“We remove ads, Facebook Pages and Instagram accounts promoting these services when we become aware of them, block links to websites hosting them so they can’t be accessed from Meta platforms,” the company said in Thursday’s statement. It also restricts search terms like "nudify," "undress," and "delete clothing" on Facebook and Instagram.

The problem is larger than Meta and CrushAI, however.

A Bellingcat investigation published last year uncovered how sites like CrushAI promote themselves on mainstream platforms.

The investigation described a loosely affiliated network of similar platforms, which include Clothoff, Nudify, Undress, and DrawNudes. The apps have “manipulated financial and online service providers” by disguising their activities to evade crackdowns, Bellingcat reported.

For instance, it found that accounts on G2A, one of the world’s largest online video game marketplaces, were being used to collect payments for Clothoff. The site disguised the sales as if they were for downloadable gaming content.

The network has also tried to exploit payment services and marketplaces such as Coinbase, Patreon, PayPal, Shopify, Steam, and Stripe, in order to receive payments.

X, the social media platform formerly known as Twitter, is leaning into a coercive new advertising strategy under Elon Musk and CEO Linda Yaccarino: pressure, threats, and lawsuits.

That’s the picture painted in a bombshell Wall Street Journal report published Wednesday. The report describes how Musk and Yaccarino led an aggressive campaign to rescue X’s declining ad business by threatening to sue companies that refused to spend money on the platform.

Desperate attempt to shore up ailing ad business

Verizon, which hadn’t advertised on X since 2022, pledged at least $10 million this year after receiving a legal threat. Ralph Lauren also agreed to return after being threatened with a lawsuit. In all, at least six companies rejoined the platform after similar pressure tactics, according to the report. Some deals included binding ad-spend commitments, while others included soft or non-binding targets struck under duress.

Behind the scenes, X lawyers reportedly warned advertisers they could be added to an ongoing lawsuit alleging an illegal ad boycott. X first filed that suit last August, targeting the World Federation of Advertisers and several major brands. Unilever, Amazon’s Twitch unit, and Lego were all among them. Some companies, including Unilever, were later dropped from the suit after signing new ad deals.

The campaign marks an extraordinary and punitive effort to revive the platform’s revenue. After Musk’s $44 billion takeover in late 2022, user engagement is reported to have sharply declined. Advertisers fled in droves amid chaos inside the company and loosened content moderation standards. Yaccarino, a former NBCUniversal ad executive, has been attempting to lure those advertisers back with brand-safety tools and new content deals, including NFL and NBA highlights. But the threats suggest serious desperation.

At one point, a lawyer for X demanded that Pinterest commit to its pre-Musk ad spend for two years — or be sued. Pinterest declined, having found that its ads were more effective when run in other venues. X then followed through on its legal threat, naming Pinterest in the complaint.

“Advertisers are understandably concerned about the litigious streak of Musk’s X interpreted by some as ‘invest with us, or else,’” Ruben Schreurs, CEO of ad consultancy Ebiquity, told the Journal.

Contempt for advertisers presaged sharp decline in ad revenue

Musk has made his contempt for advertiser pressure clear in public, too. In a November 2023 interview at the DealBook Summit, when asked about brands leaving the platform, he said, “If somebody’s going to try to blackmail me with advertising, blackmail me with money… go f—k yourself.”

Addressing Disney CEO Bob Iger directly, who was reportedly in the audience, Musk added: “Hey, Bob, go f—k yourself.”

X’s ad sales fell from $4.6 billion in 2022 to $2.6 billion last year, a 44% decrease, even as the larger ad industry grew by double digits over the same period. For their part, Meta, Google, YouTube, and Amazon all saw vast gains in ad spend over that period. Pinterest and TikTok also saw ad spending grow significantly.

Even Snap, with its choppy track record, saw ad revenue climb between 2022 and 2025.

Though it’s recently ticked up, X’s ad revenue remains well below pre-Musk levels, per the Journal. Emarketer projects 2025 will bring the first annual ad growth since the takeover, though it’s also expected to remain below pre-Musk-takeover levels.

Meanwhile the backlash may just be getting started. Some ad agencies are now demanding legal immunity from future lawsuits before agreeing to spend on X.

X has refused.

Investors are ditching U.S. equities in favor of overseas assets as concerns grow over fiscal policy, rising debt, and a possible recession. 

Exchanged-traded funds (ETFs) and equity mutual funds based in the U.S. saw outflows of $24.7 billion in May, according to LSEG data cited by Reuters. That’s the largest monthly exodus in a year. Meanwhile, last month European funds recorded inflows of $21 billion, taking year-to-date inflows to $82.5 billion, the highest in four years.

Data for 292 emerging market equity ETFs revealed inflows of $3.6 billion last month, bringing this year’s total to $11.1 billion, according to Reuters.

The S&P 500 is up just 3% this year. By comparison, it had risen 14% by this time last year. 

President Donald Trump’s reciprocal tariffs on imported goods, and his fiscal policy, have a lot to do with the flight, as it has eroded U.S. assets’ safe-haven reputation.

Since the unveiling of the levies on April 2, the U.S. dollar has shed about 5% of its value against a basket of foreign currencies. Concurrently, in the week or so following "Liberation Day," a U.S. government bond sell-off ensued, sending 10-year Treasury yields soaring more than 12%. Yields remain high, hovering around 4.45%, up from 4.16% the day before the tariff unveiling.

Adding to the bond market’s tariff-related jitters, last week Trump unveiled his “Big Beautiful Bill," which could add $2.4 trillion to the deficit over the next decade, according to the nonpartisan Congressional Budget Office. 

Typically, the dollar and yields have moved in step with each other, as higher yields signal to investors that economic growth is to come, thus attracting inflows of foreign capital and strengthening the currency.

That relationship appears to be breaking down, as the correlation reached its lowest level in nearly three years in June, according to the Financial Times' analysis of LSEG data. It's thought that higher borrowing costs could reflect higher risk, as the government takes on more debt while tariffs threaten growth.

When Vineet Jain, the CEO of cloud management company Egnyte, decided to spend four days without using Google Search, he was “blown away” by how clean and simple AI interfaces were: no clutter at the top of his screen, no ads, just the answers. Shortly after, his 15-year-old son told him, “We don’t use Google anymore. We do everything with ChatGPT.”

That shift in user behavior comes as the U.S. Department of Justice gears up for a defining ruling in its long-running antitrust case against Google. In a decision expected by August, Judge Amit Mehta will decide whether Google must unwind its lucrative default-search deals and spin off its Chrome browser. 

“If you remember the Samson and Delilah story where Samson’s power was in his hair, [Google’s] ‘hair’ is going to be search,” Jain said. And that search business may very well be on the chopping block — even if the DOJ doesn’t win its case.

The financial engine behind Google’s empire

In the first quarter of 2025, Google’s advertising revenue reached $66.9 billion, with $50.7 billion coming from Google Search — thanks largely to its iron-clad position as the default on browsers and devices. That default status is the quiet engine behind its dominance: In 2022, Google paid Apple over $20 billion to be the preset search engine on Safari, locking in the company’s share of mobile eyeballs. Chrome, Google’s browser, adds even more gravitational pull.

For years, regulators have circled Google over its dominance in search — and the advertising empire built on it. But what if the real existential risk isn’t legal at all? What if it’s behavioral? 

Search is no longer analogous to Google. ChatGPT and other AI-native tools are starting to pull traffic and attention away from traditional search engines by answering questions directly, skipping the blue links, and ignoring the ad slots altogether. According to a recent report from Mary Meeker, ChatGPT hit a billion searches a day in less than two years; Google needed 11 years to reach that mark.

Jain said Mehta’s ruling could be a “watershed moment” for Google “in terms of how they regulate their platform power. Because they have become a behemoth — in search, display, advertising, everything related to it.”

As AI search starts to capture a larger share of the search market, Jain said the argument Google could make is, “‘Hey, we are not a monopoly anymore. We’re not as dominant as you’re making us out to be.’”

But whether the company wins or loses the antitrust case, the deeper problem remains: Google is built around an advertising model that depends on search volume, attention, and high-intent queries. Unlike pure AI chatbots, which can deliver answers free of sponsorship, Google can’t simply declutter its interface without cutting off the revenue stream that fuels its empire.

And while Google has begun integrating AI into its products — from AI overviews to Gemini — it’s trying to have it both ways: serve direct answers while keeping users scrolling and clicking long enough to monetize their attention. That’s a far harder trick than it sounds, especially as users come to expect stripped-down, ad-free experiences from tools such as ChatGPT, Perplexity, or Claude.

And the scale of the business makes any change feel like trying to turn an ocean liner. “Google Search, in terms of its contribution to the overall Google pie, is so disproportionately large it’s almost like a massive, massive, ten-times-the-size-of-the-Titanic that you’re trying to move,” Jain said. “So it’ll take time.”

How Google adjusts to the changing market is the “$10-plus-billion question,” he said.

Defaults are powerful — until they aren’t

Google’s dominance in search was never just about having the best engine. It’s also about placement — and payments. That strategy is what’s on trial. Mehta is weighing whether these default arrangements violate antitrust law by unfairly shutting out competition. A ruling that forces Google to unwind its “default” approach could dramatically shake up how users access search — and open the door for competitors to gain ground. 

“Over time, given the size and heft [Google has], there will be a gradual atrophy in the search traffic going to them,” Jain said.

He said he expects some regulatory response regardless of the outcome. “There’ll be some action. I can predict. There’ll be some action,” he said. “I don’t know what it’s gonna be, and I hope the Chrome divestiture, if it happens, it’s done in a favorable way” — where it’s not bought by a competitor, such as OpenAI, which has expressed interest.

Any erosion to Google’s core business may not grab headlines like a courtroom ruling, but it could be as consequential. Just as Microsoft missed the mobile revolution while defending its Windows monopoly, Google risks clinging to its current business model while the next paradigm quietly passes it by. This isn’t the first time Google has faced a potential disruption — mobile search, voice assistants, and vertical platforms like Amazon and TikTok have all chipped away at its dominance. 

But generative AI is different. It doesn’t just offer an alternative to search — it threatens the idea of search itself.

The irony is that Google has long known this moment was coming. In 2011, then-CEO Eric Schmidt warned that “the next Google” wouldn’t look like a search engine — “we’re trying to move from answers that are link-based to answers that are algorithmically based, where we can actually compute the right answer,” he said. That future is here, but Google’s own business incentives are pulling it in the opposite direction. The company can’t become a pure-answer engine without undercutting the ad clicks that make up the bulk of its revenue.

And that tension will only grow more painful as AI-native tools accelerate. Jain predicts search will remain Google’s biggest revenue contributor “for at least the next five to seven years,” but beyond that, the fundamentals begin to wobble.

Still, he’s not betting against the company. “People underestimate Google and the intellectual horsepower they bring to the table. And they are no slouches,” Jain said. “I would not bet against Google at this point.”

But even the smartest players can miss the moment if they’re too busy protecting the past. So while Google isn’t going anywhere — traditional search might be.

Google wants to further cut headcount in order to free up cash for artificial intelligence. 

The tech giant has extended voluntary buyout offers to U.S. employees, The Information reported Tuesday. It’s part of a larger cost-cutting plan to help fund the tens of billions of dollars it’s spending on developing AI models and systems.

The offers were extended to staff across multiple divisions, including search and advertising, research, and engineering.

“Earlier this year, some of our teams introduced a voluntary exit program with severance for U.S.-based Googlers, and several more are now offering the program to support our important work ahead,” a Google spokesperson told Quartz. 

But the invitations may not come as a surprise, as Google has been trimming workers for years in order to build a war chest of cash for AI. It completed its largest-ever round of layoffs in 2023, cutting around 12,000 employees, or 6% of its workforce. It has cut jobs in smaller amounts since then. For instance, employees who work on Android, Chrome and Pixel products were offered voluntary buyouts in January, in addition to those in finance, legal and human resources.

Google is beefing up its AI spending as it competes with the likes of ChatGPT, Perplexity and Anthropic, which are refining their sophisticated large language models that threaten Google’s search engine dominance. 

Last month, it was revealed that Apple is exploring how to incorporate AI-powered search engines on its devices. AI tools are “so much better [than search engines] that people will switch,” said Eddy Cue, Apple’s senior vice president of services, during his testimony in the U.S. Justice Department’s lawsuit against Google parent company Alphabet. The news caused Alphabet’s share price to fall 8%.

In order to keep up with the post-search era — a term for users’ increasing preference for retrieving answers via chatbots instead of curated links — Google launched AI Overviews last year. The feature summarizes search results, replacing the need to sift through websites. It took this a step further last month, overhauling its search engine with the U.S. introduction of “AI Mode,” which answers search queries in a chatbot-style conversation with fewer links.

Wall Street isn't impressed with Apple this week. As the tech giant holds its annual Worldwide Developers Conference (WWDC), one analyst called the event "a yawner". Shares fell more than 1% Monday before inching back a bit Tuesday as investors were disappointed by AI announcements another analyst called "incremental at best."

Apple's annual developer showcase failed to deliver the splashy announcements that have defined previous WWDCs — no Vision Pro-style hardware debut, no Apple Intelligence reveal.

Instead, Apple spent most of its keynote unveiling "Liquid Glass," a new design language that makes iOS look, well, more glass-like. As with many Apple announcements, reactions spanned from really liking it to calling it a “bad” copy of the Windows Vista aesthetic from 2007.

But buried beneath the surface-level disappointment was a quieter announcement that could prove far more consequential for Apple's AI ambitions: the Foundation Models framework.

For the first time, Apple is giving third-party developers direct access to the on-device large language models that power Apple Intelligence — with as little as three lines of code. This represents a fundamental shift for a company that has historically guarded its core technologies like state secrets.

The framework enables developers to build AI features that run entirely offline, protecting user privacy while eliminating the cloud API costs that can make AI prohibitively expensive for smaller developers. Examples showcased included the quiz platform Kahoot creating personalized study questions from user notes and the hiking app AllTrails suggesting routes based on natural language descriptions — all without an internet connection.

What makes this significant isn't the current capabilities of Apple's 3 billion parameter model, which pales compared to frontier models like GPT-4 which is rumored to be more than a trillion. It's the strategic positioning.

Apple has never been first to market with transformative technologies. The company didn't invent smartphones, tablets, or smartwatches — it perfected them. The Foundation Models framework suggests Apple is applying this same playbook to AI.

Consider the math: Apple has over 1 billion active iPhone users and millions of developers in its App Store ecosystem. Even if Apple's current AI models are inferior to OpenAI's or Google's, enabling millions of developers to experiment with free, on-device AI could generate the breakthrough applications that Apple hasn't been able to create internally.

By enabling millions of developers to build on Apple's AI foundation, the company could leverage its ecosystem advantages rather than trying to out-innovate OpenAI or Google directly. As tech analyst Ben Thompson noted, Apple's initial success with the Apple II was because of third-party developers, and developers were critical to making the iPhone a sustainable success with the App Store. 

"Trusting developers and relying on partners may be a retreat from Apple's increasing insistence on doing everything itself, but it is very much a welcome one," Thompson wrote.

While competitors like Google and OpenAI race to build ever-larger cloud-based models, Apple is trying a different approach: making decent AI universally accessible across its device ecosystem. What's more, it's doubling down on local processing — something other companies have largely abandoned in their rush to the cloud.

This strategy aligns with Apple's broader positioning around privacy. In an era when AI companies are scraping the internet for training data and users are increasingly concerned about data privacy, Apple's on-device approach could prove prescient.

None of this guarantees success. Apple delayed its promised "more personal" Siri features in March and provided no update on timing during Monday's keynote. The company's AI credibility remains damaged from over-promising and under-delivering on Apple Intelligence.

But if Monday's start to WWDC was indeed a "retreat," as Thompson characterized it, it might be a strategic one. Rather than continuing to overextend in a battle it is unlikely to win, Apple can return to focusing on its unique advantages while making a bet that empowering developers and partners will pay off like it has before.

The Foundation Models framework might not generate headlines like ChatGPT or Gemini. But in Apple's typical fashion, it could prove more important than the flashier announcements from its competitors. The question is whether developers can build something compelling enough with Apple's tools to keep the company in the AI race.

Mark Zuckerberg doesn’t just want Meta (META) to be smart — he wants it to be super smart. The CEO is reportedly creating an AI lab to pursue “superintelligence” as Meta jockeys for position at the forefront of an ever-crowded AI field.

Zuckerberg is largely focused on achieving artificial general intelligence (AGI) — a currently theoretical humanlike level of intelligence where a machine can perform any task a person can. Meta could then weave that AGI across its product suite — chatbots, augmented reality glasses, and more.

According to Bloomberg, Zuckerberg has been personally recruiting top AI talent and has been prioritizing the staffing and development of this new team, which is being referred to internally as a “superintelligence group,” according to people familiar with the matter. The Meta CEO is reportedly looking to hire 50 people for the team, which he hopes will make his company a leader in AI.

Meta is reportedly in talks for a $10 billion-plus investment in Scale AI — a data platform company that helps companies train models — and the company will likely play a big role in Zuckerberg’s plans. Alexandr Wang, the company’s founder and CEO (and former housemate of OpenAI’s Sam Altman), has been tapped to join Meta’s venture, and the investment deal is poised to bring other Scale AI employees to Meta. Other researchers from Meta’s rivals have reportedly been offered seven- to nine-figure deals to jump ship.

Sources told The New York Times that the superintelligence lab is part of a larger reorganization effort and comes at a time when Zuckerberg has been frustrated by failing product releases, such as Llama 4, which has been questioned both internally and externally by developers. Despite Zuckerberg’s claims that Meta’s Llama model worked as well as (or better than) models from industry leaders, researchers largely decided that Meta’s model was designed to look more advanced than it actually was.

As a a result, the company delayed plans to release its next iteration, “Behemoth,” with the Wall Street Journal reporting that Meta’s brass was worried the model didn’t sufficiently improve on previous models. Sources told Bloomberg that these missteps have led Zuckerberg to get more involved in the company’s AI dealings. The CEO has reportedly created a WhatsApp group with top execs called “Recruiting Party” to find talent it will try to hire away to help with Meta’s efforts.

While Meta has invested billions in AI, it still lags behind others in the field. OpenAI’s ChatGPT continues to rake in money — and has increased its paid user base by 50% in three months. Microsoft (MSFT) has invested more than $13 billion in OpenAI, and its stock has soared as investors see that the company’s aggressive moves are paying off. Amazon (AMZN) has poured more than $8 billion into startup Anthropic — and is continuing to make big investments to expand its data services. And Google (GOOGL) has continued to develop its AI research lab, DeepMind, which Google parent Alphabet bought in 2014.

So Zuckerberg needs Meta’s AI lab to be super — in more ways than one.