All News (Page 375)

The rapper Afroman, famous for his 2000 hit “Because I Got High”, will be speaking at the Bitcoin 2026 conference next month in Las Vegas. The artist, whose given name is Joseph Foreman, is coming off what some are calling the ‘Lemon Pound Cake’ trial victory, in which an Ohio jury ruled earlier this month that he did not defame sheriff’s deputies who invaded his home in 2022. 

The artist went viral this month after prevailing in a legal ordeal that began four years ago when authorities raided his home to search for evidence for drug trafficking and kidnapping, but failed to find any. Afroman then released a song about the raid, with a music video including footage of the officers tearing apart his home and inspecting a lemon pound cake that was in his kitchen. Crypto boosters hailed his legal victory, identifying with the notion of feeling violated by government entities. 

“[Afroman’s] story of standing up to power and winning resonates deeply with Bitcoin’s ethos of sovereignty and financial freedom,” organizers of the conference wrote on LinkedIn. 

With the recent cancellation of Token2049 in Dubai, Bitcoin 2026 in Las Vegas will be one of the biggest crypto conferences this spring. Afroman will join a list of speakers including President Donald Trump’s son Eric Trump, executive chairman of Strategy Michael Saylor, and the chair of the Commodity Futures Trading Commission Mike Selig. 

The crypto industry is looking for reasons to celebrate after a brutal five month stretch. The price of Bitcoin is down about 47% since its all-time high in October to its current price of about $67,000, according to Binance. This tailspin comes despite a friendly regulatory environment under the second Trump administration. 

When announcing Afroman’s appearance at the Las Vegas conference, the event organizers posted, “Vibes are guaranteed to be high.”

This story was originally featured on Fortune.com

Many parents and kids alike are wondering whether college has the same return on investment it once did. And they have reason to worry: Hiring just hit a level not seen since the economy was “closed down literally” during the pandemic.

Going to college was once seen as a one-way ticket to a successful and lucrative career. Still, there are a growing number of six-figure jobs that don’t require a degree, while entry-level job opportunities for recent graduates remain sparse. 

Some parents are so anxious about today’s job market that they’re exploring alternatives to the four-year degree, with one in three open to the idea of their kids attending a trade school instead, according to late 2025 survey results from Britebound (formerly American Student Assistance), which surveyed more than 2,200 parents of middle and high school students about their attitudes, perceptions, and decision-making regarding their kids’ post-high-school plans. 

The fact that 35% of parents believe career and technical education is best suited for their children represents a major jump—from just 13% in 2019, according to Britebound. While parents still prefer traditional college for their kids, it’s much less so than in the past. The percentage of parents preferring it dropped to 58%—a 16 percentage point drop from 2019.

And another study from Britebound last summer shows it goes both ways: 70% of teens also report their parents are more supportive of forgoing a college education for something different, like trade school or an apprenticeship. 

“Parents are waking up. College doesn’t carry the same [return on investment] it once did, because the cost is outrageous, and the outcome is uncertain,” Trevor Houston, a career strategist at ClearPath Wealth Strategies, previously told Fortune. “Students now face the highest amount of debt ever recorded, but job security after graduation doesn’t really exist.”

The average cost of college in the U.S. is more than $38,000 (including tuition and room and board) per student per year, according to the Education Data Initiative, and the average cost of college has more than doubled this century. Private schools almost always cost more than the average. Meanwhile, more than 4 million Gen Zers are jobless and blame their “worthless” college degrees. 

Trade jobs that pay six figures without a degree

One of the primary reasons trade school is becoming a more popular option for students is its potentially strong ROI, especially as college becomes more expensive and fewer traditional entry-level jobs are available. And many can land recent high school grads six-figure salaries. 

According to the National Society of High School Scholars, some trade jobs that don’t require a college degree and pay six figures include:

• Aircraft mechanics ($135,628)
• Plumbers, pipe fitters, and steamfitters ($132,275)
• Construction managers ($130,000)
• Industrial electricians ($122,500)
• Energy technicians ($115,076)

What’s more, the need for these workers will continue to grow, especially as older generations who work in trades start to retire, Julie Lammers, president and CEO at Britebound, previously told Fortune. 

“An aging workforce in the trades and a surge in demand to meet infrastructure needs, ever-growing real estate demands, and changes to U.S. energy production mean that there are considerably more job openings than skilled workers to fill the need,” she said. 

How much does trade school cost vs. college?

Aside from trade school, students can also pursue apprenticeships, career-training programs, boot camps, industry certifications, and occupational licenses. Many of these are just pennies on the dollar compared with earning a college degree. A coding boot camp can cost as little as $7,000—and that’s just a one-time fee as compared with nearly $40,000 for one year of college. 

These career paths made possible by trade schools, apprenticeships, boot camps, and other training and certification programs were dubbed by IBM as “new-collar jobs.” In October 2017, IBM launched its apprenticeship program to train people for new-collar jobs that prioritize skills over degrees and focus on in-demand job functions like cybersecurity, design, data science, mobile development, cloud, artificial intelligence, and blockchains—all career paths that can also lead to six-figure salaries. 

The Trump administration in late 2025 also announced its Tech Force program, which does not require a college degree or work experience for technology professionals who are willing to serve two-year stints at federal agencies. If you’re accepted to the program, you can earn about $150,000 to $200,000, given the demand for tech professionals in today’s rapidly evolving tech landscape.

“This is a clarion call,” Scott Kupor, director of the U.S. Office of Personnel Management, said in a statement at the time. “If you want to help your country lead in the age of rapid technological advancement, we need you.”

In March, OPM also launched the Early Career Talent Network, a recruitment push for entry-level workers to work for the federal government.

“We’ve got close to half of our population that’s within 10 years of retirement age,” Kupor told Fortune‘s Sasha Rogelberg. “So if you just did nothing else, you’ve got this major demographic challenge of a large number of people who will likely either retire or certainly be retirement-eligible over the near term, without us actually replenishing the pipeline of early-career people coming in.”

A version of this story was originally published on Fortune.com on December 19, 2025.

More on trade school:

• I’ve owned and operated trade schools for 40 years and Gen Z is onto something: They recognize the value of vocational education
• AI boom is fueling demand for skilled trades—and demand for technicians, HVAC workers, and electricians is soaring, with six-figure salaries to match
• ‘College doesn’t carry the same ROI it once did’: 70% of teens say their parents support them going to trade school or getting an apprenticeship

This story was originally featured on Fortune.com

We’ve all been there: in a work meeting, trying to stop our eyes from glazing over as a colleague spews an endless monologue about “leveraging the company’s adaptive strategy to optimize our value and reinvigorate our operations.” 

That incomprehensible, buzzword-heavy language has a name: “corporate bulls–t.” That’s at least according to Shane Littrell, a cognitive psychologist and a postdoctoral researcher at Cornell University. He studies how people evaluate and share knowledge, and how misleading information shapes people’s beliefs, attitudes, and decision-making.  

As a self-proclaimed BS-hater himself, Littrell defines BS as “dubious information that is misleadingly impressive, important, informative, or otherwise engaging.” It’s easy to mistake BS for the necessary, everyday jargon used in professional settings, but its distinguishing factor is that while the language intends to sound smart or impressive, it fails to be accurate, meaningful, or if at all, helpful, he told Fortune. 

Over four studies with 1018 subjects, Littrell built the “Corporate Bulls–t Receptivity Scale,” a way to measure how attracted individuals are to this type of language and how business savvy they perceive different statements. People who find that buzzword-heavy corporate-speak profound and informative perform worse on measures of workplace leadership and decision-making, but it does not mean people who are more receptive to corporate-speak are bad at their jobs, just that they may not make the best leaders or decision-makers.

It’s not about intelligence or education, Littrell said, who noted the results were uniform between studies where more than 70% of the participants had a bachelor’s degree or higher and those with less education. 

“Part of that has to do with just the environment that you’re in. You have to use that language a little bit just to navigate the workspace,” he said. “Anybody can fall for bulls–t when it’s packaged up to appeal to your biases.” 

The dangers of meaningless corporate-speak 

The workplace is “fertile ground” for BS to fester, Littrell said, when you’re trying to impress your boss and compete with colleagues. 

“These organizational settings are saturated with these authority cues, like job titles, and these power hierarchy structures, and everybody [is] talking about their leadership vision,” he explained. “It makes it especially easy to pass that off as insight. There are always people that are trying to climb the corporate ladder, and in a lot of situations, this type of language is used in a way to try to impress everyone around them.”  

But corporate BS is more than just annoying, Littrell said. It can have a harmful effect on credibility and morale. This can be especially troubling when a leader uses it because it can undermine how employees understand goals, feedback, or decision-making.  

Corporate-speak can also lead to reputational damage and financial cost for companies, Littrell said. He gave the example of a snafu PepsiCo found itself in 2008 after an internal report explaining the company’s $1 million logo redesign leaked online. 

“The Pepsi DNA finds its origin in the dynamic of perimeter oscillations. This new identity manifests itself in an authentic geometry that is to become proprietary to the Pepsi culture,” the company’s design consultant, Peter Arnell Group, wrote in the internal report. “[The Pepsi Proposition is the] establishment of a gravitational pull to shift from a ‘transactional’ experience to an ‘invitational’ expression.” 

This proposal was not only confusing, but also created a lasting internet and media embarrassment for the company. Even the design firm’s founder admitted that “it was all bulls–t.” 

Establishing new norms can stop BS

It doesn’t have to be this way, Littrell said. A simple way companies can reverse course is by rewarding “anti-bulls–t” behavior by making clear communication the norm from the top down. This can stop a cycle where a leader uses convoluted language, and then employees feel like they have to speak that way, too.

He suggests establishing an environment that encourages people who aren’t the leaders to ask more questions, which can nip the impulse to appear like you know everything. “Sometimes people feel a social pressure where they don’t want to look stupid by answering like they think everybody else understands it, and they don’t want to raise their hand and ask a question, because they feel that that might make them look stupid,” he explained. 

Lastly, he encourages companies to reward behaviors like clear communication and asking questions in performance reviews, which he says are very critical for establishing expectations.

“One of the more important conversations is those performance reviews and the way leaders and employees communicate with each other that can cause the most problems, especially in their personal success and the organization’s success.”

This story was originally featured on Fortune.com

Hello and welcome to Eye on AI. In this edition…Anthropic suffers multiple sensitive data leaks…OpenAI ditches Sora, and loses its deal with Disney…Mistral raises money for AI data center drive…AI could reduce political polarization…and why countries that are late to adopt AI could be in even worse economic shape than you think.

The big news this week was my colleague Beatrice Nolan’s scoop from Friday that Anthropic has trained a new AI model, called “Mythos” (Capybara seems to be the internal code name for the same model), that the company says represents a “step change” in capabilities. Anthropic is particularly worried about the cybersecurity risks the model poses. Ironically, we found out about this new model because Anthropic inadvertently spilled the beans by leaving a draft blog post about it in an unsecured and publicly searchable database—along with other potentially sensitive documents about an upcoming CEO retreat and some internal documents that mentioned employees’ paternity leave.

Now, just today, it appears Anthropic has suffered another major security lapse, accidentally leaking the code of the agentic harness that sits around Claude Code. Bea has more on this latest, and potentially more consequential, data leak here. Meanwhile, Axios reports that the new cybersecurity capabilities of AI models are getting so concerning that Anthropic and OpenAI have both recently told the government about the new dangers of the models they are developing and provided government security experts with early access.

Intern, expert, or dog?

Ok, now, if you own a dog, as I do, there will be moments you’ll recognize that we fundamentally don’t understand how dogs perceive the world.

This week, while walking my dog, I spied a strikingly beautiful cat with an unusual coat. It looked like an orange tabby mixed with gray tabby, with a good deal of white fur thrown in the mix too. I noticed the cat right away, but it was moving across a yard that was elevated from sidewalk level, so my dog couldn’t see it. She could definitely smell it, however. She put her nose in the air and tugged at her leash, pulling her way up the steps that led to the yard.

By the time she got to the top step, the cat had mostly hidden itself behind a nearby flower pot. It stood behind the pot motionless, but with its white head popping above the pot’s edge. It stared intently at my dog and me. I could see the cat quite clearly. But, despite being just 15 feet away, my dog could not. She sniffed the air intently and pivoted first left and then right, but she could not see the cat, even when seemingly looking directly at it.

Eventually, I persuaded my dog to give up her hunt for the unseen, but well-smelled, cat, and continue our walk. But I couldn’t stop thinking about our differences in perception—and how this applies to AI. People often offer executives advice for how they should think about using AI by making analogies to our relationships with various categories of people. Treat AI agents like talented interns, was a popular one a few years ago, in the months following ChatGPT’s debut. A graduate student who is occasionally off their meds, was a colorful variant that Emad Mostaque, the cofounder and former CEO of Stability AI, liked to use. You should treat AI like PhD.-level researchers, was an analogy in vogue last year. (OpenAI CEO Sam Altman was among those talking about this idea.) More recently, people have started saying it is better to regard AI models like wise and experienced, but occasionally still fallible, colleagues. Certainly their performance on certain tough benchmarks of professional tasks, such as OpenAI’s GDPval, would lead one to endorse that idea. Middle managers is another analogy that comes up often.

But the more we learn about the large language models that underpin today’s AI agents, the more clear it becomes how inadequate all these analogies are. LLMs are nothing like people at all. They are far more like other species, like your dog. We can no more understand what and how these LLMs perceive and reach their outputs than we can truly understand the thoughts of our pets.

Actually, it’s worse than this, because unlike with our pets, you can ask an LLM to explain to you what it’s thinking and it will tell you. That sounds like a great thing, much better than the situation with our non-verbal dogs, cats, and turtles. The problem: Researchers have begun probing the activations of the artificial neurons in AI’s digital brains, and these experiments indicate that what an AI model tells you it is thinking—the model’s so-called “reasoning traces”—may or may not actually reflect what it is, in fact, thinking.

So interacting with an LLM is probably the closest thing we’ve had so far to interacting with an alien, one that has some capabilities that far exceed our own, but also has glaring weaknesses, and which can, at times, be just like us—deceptive, dishonest, or dissembling.

Multimodal models see ‘mirages’

This past week has brought yet more evidence of how weird these models are. A paper from researchers at Stanford University showed that multimodal AI models, those that can accept inputs in both text and images (and sometimes audio files too), suffer from a phenomenon they dubbed “mirage reasoning.”

The models will purport to analyze images a user has never actually uploaded to them. When prompted about medical images, but not actually supplied any images, the models will nonetheless offer diagnoses. Weirder still, these assessments are often correct. When the researchers tested the models on benchmarking tests for multimodal AI, the models obtained what the scientists said were “strikingly high scores”—about 70% to 80% of the scores they obtained when they did have access to images. Worryingly, the researchers found the models had a tendency to find evidence of pathologies in the phantom images, showing that the models may have a bias towards diagnosing disease that could lead to dangerous and expensive misdiagnoses if used in real-world medical settings.

The models’ sight is weak; their text pattern finding, unparalleled

The researchers have no clear understanding of exactly why the language models engage in mirage reasoning, or why they can score so highly on the benchmarks even when the images are not provided. But one experiment they conducted does suggest a possible explanation. The researchers fine-tuned a version of the open source AI model from Alibaba, Qwen-2.5, on a public training set for a popular benchmark that is designed to test how well AI models can answer questions about chest X-rays. But they trained it on this set with the accompanying images removed. They picked Qwen-2.5 in part because, at just 3 billion parameters, it is a relatively small model and therefore easy to fine-tune. But more importantly Qwen-2.5 was released a year before the chest X-ray benchmark they were using debuted, which the scientists hoped would minimize the chance that the set of questions actually used for the test itself would have ended up in Qwen-2.5’s initial pre-training data. (This kind of “data leakage” is a real problem for validity of AI benchmarks and a reason they need to be continually updated; otherwise models just memorize the answers as part of their pre-training.)

Nonetheless, this fine-tuned version of Qwen-2.5 outperformed every frontier AI model tested on the normal, image-included version of the X-ray challenge. It also beat the scores of human radiologists by 10%. Again, even though it did not have access to any of the images! The scientist found the model, despite never seeing any images, offered “reasoning traces comparable to, and in some cases indistinguishable from, those of the ground-truth or those generated by frontier multi-modal AI models.”

This implies, the scientists said, that there are hidden patterns in questions themselves, perhaps in their phrasing, or in the structure of how those questions appear in the benchmark test, that are too subtle for any human to detect, but that nonetheless are sufficient to allow the model to guess the answer. This, combined with the researchers other findings, seems to suggest that multimodal models barely use the visual inputs they are given at all and instead lean heavily on linguistic patterns even when being asked to analyze images. It also suggests, alarmingly, that most of the multimodal benchmarks may not provide a good measure of how these models will perform in real-world clinical settings. 

Again, this is totally bizarre and alien to the way humans work. This is like my dog, able to smell the cat, but not see it—while I relied on my sense of sight, but could smell nothing. Our tendency to wrongly anthropomorphize AI models may lead us to misdesign the systems we use to run and govern AI agents, with potentially bad consequences. It also speaks to the way AI systems continue to improve in capability but lag in reliability that I wrote about last week. We need to engineer our AI workflows for alien minds, not our own.

With that, here’s more AI news.

Jeremy Kahn
jeremy.kahn@fortune.com
@jeremyakahn

Before we get to the news, if you haven’t yet read my colleague Sharon Goldman’s magisterial feature story on how construction of Meta’s massive Hyperion data center is upending the lives of people who live in rural Richland Parish, Louisiana, drop whatever it is you are doing right now, and go and read it. Here’s the link. It’s a deeply reported and deeply nuanced portrait of what happens when a community suddenly finds itself living in ground zero for the biggest, most expensive infrastructure build-out in American history. 

This story was originally featured on Fortune.com

Americans aren’t getting laid off. And they’re not quitting. They’re simply just not getting hired, and the numbers haven’t been this bad since the pandemic closed the economy by force.

The Bureau of Labor Statistics reported Tuesday the hiring rate fell to 3.1% in February, with just 4.8 million hires, the lowest since April 2020. Job openings dropped to 6.9 million, down 358,000 from January. The quits rate held at a low 1.9%, while layoffs also stayed pinned at 1.1%, and retirements fell back near record lows. Everyone, it seems, is staying put, whether in their jobs or in unemployment.

“It’s a brutal job market,” Heather Long, chief economist at Navy Federal Credit Union, told Fortune. “To see that 3.1% hiring rate, the lowest since April 2020, when the economy was closed down literally during COVID—it just underscores how little hiring is going on.”

The comparison to 2020 is what makes this report so jarring. Back then, hiring collapsed because businesses were physically shuttered. Today, unemployment is around 4%, businesses are open, but employers are still barely bringing anyone on.

A ‘locked-out’ market for new hires

Nicole Bachaud, labor economist at ZipRecruiter, wrote in a note it’s a “locked-out market” for new entrants, driven by the combination of stalled hiring and delayed retirements blocking the natural pipeline. 

“Aside from the 2020 dip, the hires level has not been this low since 2014, when the labor market was still rebuilding after the Great Recession,” she wrote.

She also attributed part of the problem to another force majeure: bad weather. Construction and accomodation/food services were the two industries where hiring fell most, and those are the ones most sensitive to weather events. February marked a brutal month across the country, with blizzards and blackouts. 

Skanda Amarnath, executive director of Employ America, an economic strategy firm, said bad weather and health care strikes explain part of the February drop, but not all of it. 

“We can probably attribute 50 to 60% to just kind of the one-offs,” he told Fortune. “But there’s something fundamental at play too.” 

He pointed to reduced immigration as one factor quietly draining dynamism from the system: less population growth means less churn, fewer people switching jobs, and fewer new hires.

Long flagged a more immediate warning sign: hospitality and construction are typically where displaced workers land first, not the places that should be very sensitive to macroeconomic headwinds. 

“Most people, if they lose a job, think, okay, I could at least be a bartender or work at a restaurant,” she said. “And clearly there was a deceleration in that area.”

How the war will impact jobs in America

The JOLTS data is from February, before the U.S.-Israeli campaign against Iran upended global energy markets. With Brent crude hovering above $115 and the Strait of Hormuz effectively closed, the question is whether the labor market’s low-hire, low-fire equilibrium can survive an energy shock. Bachaud warned surging gas prices would hit transportation, manufacturing, retail, and consumer spending—”further pulling back hiring activity in the March data.”

Long said the war could be the final straw in the camel’s back for the labor market. 

“It is not inconceivable that companies go from no hiring to starting to fire in order to make their budgets work,” she said, adding the April jobs report, due in May, “could really be a first big warning sign.”

For the Fed, the report deepens the potential stagflation bind. Amarnath noted inflation has been running a full percentage point above the central bank’s core target and trending in the wrong direction, even before the war. 

“The Fed’s got to be on guard for risks that their policy is not actually tight enough,” he said.

The March jobs report, due Friday, will offer the next read on the labor market. Both economists cautioned against drawing too direct a line from JOLTS to payrolls, but the broader picture is getting harder to wave away. Long said if Friday delivers another weak number, “it’s looking more like some early demand issues are back in the picture. And that’s really nerve-wracking if you’re going to layer the war in Iran on top of that.”

This story was originally featured on Fortune.com

Billionaire investor Warren Buffett said he has not talked to his longtime friend, Microsoft cofounder Bill Gates, since he was engulfed in a scandal over his alleged ties to Jeffrey Epstein. 

The retired Berkshire Hathaway CEO said it’s been radio silence between the pair since Gates’ involvement with Epstein became clearer earlier this year, following the government’s release of millions of pages of related documents.

“I haven’t talked to him at all since the whole thing was unveiled,” Buffett told CNBC in an interview published Tuesday that included his first public comments on the Epstein files.

“I don’t want to be in the position where I know things,” Buffett added. “I could get called as a witness.” Buffett added he did not want to say much on the topic “until things are resolved.” 

Still, the 95-year-old said he was thankful he had never run in the same circles as the disgraced financier and had never met him.

“If I lived in New York at some party,” he may have run into him, said Buffett, who has lived in Omaha, Neb., for more than 65 years. 

Gates and the Epstein files

Among some of the accusations Gates has faced as a result of the documents released this year are that he allegedly had an affair with Mila Antonova, a Russian bridge player, during his marriage to ex-wife Melinda French Gates. He also allegedly gave Epstein permission to act as a fixer to help negotiate the exit of Boris Nikolic, the chief science advisor for the Gates Foundation and at Gates’ investment firm, then-called Bgc3. Nikolic received a $5 million exit package. Epstein also reportedly played a role in the exit negotiation for Microsoft Windows president Steven Sinofsky, who received $14 million from the company, for which he allegedly paid Epstein a $1 million fee.

“It’s astounding to me that anybody could be that successful as a con person,” Buffett said of Epstein on Tuesday.

In an interview with the Wall Street Journal last year, Gates said of Epstein, “In retrospect, I was foolish to spend any time with him.” More recently, Gates apologized to Gates Foundation staff in a town hall last month and acknowledged having two affairs with Russian women, the Journal reported. Epstein later found out about the affairs, Gates said during the meeting, but the affairs didn’t involve victims of Epstein’s sex trafficking operation. 

“I did nothing illicit. I saw nothing illicit,” Gates said during the town hall. 

The Gates Foundation, for its part, said in a previous statement that a small number of its employees interacted with Epstein to try to secure potential funding for its philanthropy but that “at no time were financial payments made by the foundation to Epstein, nor was he employed by the foundation at any time.” 

A spokesperson for Bill Gates said in a statement to Fortune that the Microsoft cofounder was committed to answering all questions and demonstrating he wasn’t part of Epstein’s criminal activity. Gates, along with seven others, was asked to testify before the House Oversight Committee earlier this month as part of its investigation into Epstein. 

“Gates has acknowledged it was a serious error in judgment to meet with Epstein,” the spokesperson said.

A spokesperson for Buffet did not immediately respond to Fortune’s request for comment. 

Billionaire philanthropy shake-up

The friendship between two of the world’s richest men began in 1991, when Gates’ mother, Mary, invited her son to join her and her friends for a gathering at her home, which Buffett had been invited to by late Washington Post Editor Meg Greenfield. Gates ultimately attended because Greenfield had invited the then-publisher of the Post, Katharine Graham, whom he wanted to meet.

Although Buffett and Gates later said they weren’t particularly excited to meet, they hit it off immediately, leading to a decades-long friendship and close collaboration on the Gates Foundation and the Giving Pledge, which the duo founded with French Gates. French Gates left the Gates Foundation in 2024 and now has her own philanthropic organization, Pivotal, which aims to “accelerate the pace of social progress for women and young people in the U.S. and around the world.”

Still, in recent years, the relationship between the two billionaires has cooled. Buffett stepped down from the Gates Foundation’s board in 2021, saying his “physical participation” was no longer needed for the Foundation to reach its goals, following Gates’s announcement of his divorce. Although Buffett has reportedly donated nearly half of the Gates Foundation’s funding, about $43 billion, he told The Wall Street Journal in 2024 that no more of his money would be donated to the foundation following his death. 

Although he didn’t elaborate as to why he made the decision to cut off the Gates Foundation after his death, Buffett said most of his remaining wealth after his death will go to a charitable trust overseen by his daughter and two sons, whom he “trusts completely,” according to the Journal. 

This story was originally featured on Fortune.com

There’s a common adage when it comes to sales: Go where the people are.

It seems that’s what Ulta Beauty’s doing after its March 17 launch on TikTok Shop, becoming the first specialty beauty retailer in the country to launch on the platform where scrolling and discovering new products is encouraged. Now, if you’ve ever scrolled through a TikTok video and wondered what foundation that person is using, you can scroll through Ulta in app and purchase it for yourself. 

Ulta’s move onto the platform comes as retailers reconsider what truly is the front door of retail. It also comes as TikTok’s commercial future in America was stabilized following a landmark deal with the Trump administration. 

The backdrop to Ulta’s launch is as much political as it is commercial. Earlier this month, the Trump administration finalized a deal allowing TikTok to continue operating in the U.S. in exchange for a reported $10 billion brokerage fee paid to the U.S. government, according to The New York Times. Investors including Oracle, Emirati investment firm MGX, and Silver Lake (which each own 15% of the company) will pay the U.S. government $10 billion for brokering the deal, $2.5 billion of which was already paid in January. The deal comes as a resolution to years of national security concerns over the app’s Chinese parent company, ByteDance, and effectively resolved the uncertainty that hung over TikTok since 2020, giving brands and retailers a clearer runway to invest in the platform’s commerce capabilities. Ulta announced its expanded TikTok integration just days after the deal’s terms were finalized.

Go where the audience is

TikTok isn’t a place brands are trying to build an audience—it’s where an audience of historic scale is already shopping. It’s also where new possibilities in the consumer space are emerging, including the ability to enhance shopping with AI.

“We are excited about the opportunities, both on social and AI-enhanced commerce platforms, to bring our undeniably Ulta Beauty experience and assortment to life,” Ulta Beauty CEO Kecia Steelman said during the company’s Q4 2025 earnings call. “We will initially launch with a thoughtfully curated assortment of only-at-Ulta brands, which will add another exciting tool to our brand-building playbook.”

It’s uncertain how the AI-enhanced commerce platforms will work. However, Ulta’s approach to TikTok is part of an overarching recognition by the beauty retailer that sees “firsthand how discovery is happening everywhere today–and social platforms play an increasingly influential role in how guests engage with brands,” Lauren Brindley, chief merchandising and digital officer at Ulta Beauty told Fortune in a statement.

“Partnering with TikTok Shop is a strategic and complementary extension of our discovery ecosystem,” she added. “It allows us to meet guests in the moments that inspire them, reduce friction between content and commerce, and drive incremental growth by welcoming new-to-Ulta Beauty shoppers into our community.”

Steelman made similar comments in the earnings call, adding it’s meeting users where they are. TikTok Shop is “where guests can purchase immediately as they engage with content from Ulta Beauty and our brands on the platform.”

The data backs her up. TikTok Shop logged more than 103 billion U.S. searches with e-commerce intent in 2025, and total transaction volume on the platform rose nearly 80% year-over-year, according to data shared directly from TikTok to ModernRetail. There are now 71.4 million active social shoppers on TikTok in the U.S. alone, up 24.5% from 2024, and 45.5% of all U.S. TikTok users made at least one social commerce 

The company has been expanding its use of generative AI, including agentic AI tools and an internal AI Center of Excellence, to personalize marketing across its 46 million loyalty members. The TikTok Shop launch, in that context, is the consumer-facing result of a back-end transformation years in the making.

It also arrives alongside one of Ulta’s strongest recent quarters: $3.9 billion in Q4 2025 sales, an 11.8% year-over-year gain, with comparable sales rising 5.8%. Steelman, who took the CEO role in January 2025 after 11 years at the company, has been clear about the turnaround.

“We had to get our swagger back,” she said. “I felt like we lost our swagger just a little bit, and I feel like we’ve got our swagger back.”

However, some are cautious to laud the partnership out of fear of what has happened time and again with self-conscious users (who are primarily underage) on social media platforms. Yale Medicine dermatologist Dr. Kathleen Suozzi, who has researched the skincare routines and purchases of kids and teens as influenced thanks to social media, questioned if “this strategy is capitalizing on the impulsivity” of a younger cohort, especially given her work while “looking at behavioral patterns in teens and tweens around skincare.”

“I think this really targets teens and tweens in a major way,” Suozzi told Fortune, adding that because TikTok Shop all occurs in app, it’s just that much easier for all users, regardless of age, to impulse buy.

“We see really everything related to social media and kids, it’s that chasing of this idea, this false idea of perfectionism, this representation of what skin should look like, and chasing these beauty ideals that are not realistic or appropriate,” she added. “That’s really what this is feeding into.”

Suozzi mentioned how less sophisticated users might not be able to discern between what is a sponsored ad as compared to a regular user on the app not touching up their appearance. This “comparing your skin to these influencers that have filters and lighting, and the pressure to follow multi-step routines—this is contributing to increased anxiety about your appearance and compulsive product use. And that’s also what this integrated platform is going to feed even more. You see something, you immediately want it, you buy it.”

Beauty’s broader bet on TikTok

Ulta is not alone in recognizing TikTok’s pull on the beauty industry. Although it does not have an official store on TikTok Shop, Sephora previously partnered with the platform to pioneer a creator program connecting emerging brands from its Accelerate incubator with content creators.

Ulta’s TikTok Shop launch takes that conviction further, turning discovery into a direct purchase moment rather than a brand-building exercise. TikTok Shop’s Head of Beauty, Ajay Salpekar, framed the partnership as additive rather than disruptive:

“TikTok is where culture, commerce, and discovery come together in a seamless way, so it makes sense for retailers and brands to be part of our ecosystem,” Salpekar told Fortune in a statement. “For Ulta Beauty, TikTok Shop offers the power of discovery, helping to reach new shoppers for the release of exclusive launches and to support the scale and growth of new-to-market brands.”

This story was originally featured on Fortune.com

Tiger Woods’ eyes were bloodshot and glassy, his pupils dilated and he had hydrocodone pills in his pocket when interviewed at the scene of his car crash last week in Florida, according to a sheriff’s office report released Tuesday.

Woods’ movements were slow and lethargic, he was sweating as he talked to deputies and told them he had taken prescription medication earlier in the morning, according to the incident report released by the Martin County Sheriff’s Office. Woods told deputies he had been looking at his phone and fiddling with the radio before he clipped a truck in front of him, the report said.

Deputies found two white pills, which were identified as the opioid hydrocodone used to treat pain, in his pocket, the report said.

When asked by a deputy if he took any prescription medications, Woods said, “I take a few.”

The golfer was traveling at high speeds on a beachside, residential road on Jupiter Island when his Land Rover clipped the truck and rolled onto its side, according to the sheriff’s office, which noted Woods showed signs of impairment.

The truck had $5,000 in damage, according to the sheriff’s report.

The truck driver and another person helped Woods out of his vehicle, with the golfer needing to climb out from the passenger side. Neither Woods nor the truck driver were injured.

During a field sobriety test, deputies noticed Woods limping and that he had a compression sock over his right knee. The golfer explained he had undergone seven back surgeries and over 20 leg operations and that his ankle seizes up while walking. Woods, who was hiccupping during the questioning, continuously moved his head during one of the sobriety tests and deputies had to instruct him several times to keep his head straight, the report said.

“Based on my observations of Woods, how he performed the exercises and based on my training, knowledge, and experience, I believed that Woods normal faculties were impaired, and he was unable to safely operate the motor vehicle,” the deputy wrote after the tests.

Woods, 50, is the most influential figure in golf and has become as recognizable as any athlete in the world. The first person of Black heritage to win the Masters in 1997, he has captivated golf fans with records likely never to be broken.

But his injuries kept him from accomplishing more, including those suffered in a 2021 car crash that damaged his right leg so badly he said doctors considered amputation.

At this latest crash, Woods agreed to a Breathalyzer test that showed no signs of alcohol, but he refused a urine test, authorities said. He was arrested and released on bail eight hours later.

Woods’ agent at Excel Sports, Mark Steinberg, has not responded to multiple messages seeking comment. No one from Woods’ camp or the PGA Tour — he is on the board and is chairman of the committee reshaping the competition model — have commented since his arrest.

Woods, who has been involved in many crashes over the years, is charged with driving under the influence, property damage and refusal to submit to a lawful test. He is scheduled for arraignment April 23. Online court records do not list an attorney for him.

Under a change to Florida law last year, refusing a law enforcement officer’s request to take a breath, blood or urine test became a misdemeanor, even for a first offense.

___

AP Golf Writer Doug Ferguson in Jacksonville, Florida, contributed to this report.

This story was originally featured on Fortune.com

Mark Zuckerberg texted Elon Musk asking if he could assist him with Department of Government Efficiency (DOGE) efforts last year, according to newly released court documents.

The newly unredacted filings are part of an ongoing legal battle between Musk and OpenAI that began in 2024, with the xAI CEO alleging that OpenAI and CEO Sam Altman violated the company’s original mission of developing AI to benefit humanity. In February 2025, Musk submitted an unsolicited $97.4 billion bid to acquire OpenAI and block its conversion into a for-profit entity.

“Looks like DOGE is making progress,” Zuckerberg texted Musk on Feb. 3, 2025, according to an unsealed exhibit. “I’ve got our teams on alert to take down content doxxing or threatening the people on your team. Let me know if there’s anything else I can do to help.”

Musk reacted with a heart to Zuckerberg’s message and responded, “Are you open to the idea of bidding on the OpenAI IP with me and some others?”

Zuckerberg offered to discuss the matter “live,” and Musk suggested he would call the Meta CEO the next day, the filings show.

The communication shown in the filings indicates a thawing relationship between the two entrepreneurs after a decade-long rivalry. In 2016, Meta contracted Musk’s SpaceX to launch a satellite that would have given internet access to individuals in sub-Saharan Africa, but the rocket exploded. Zuckerberg said he was “deeply disappointed” by the failure. In 2023, Musk offered to fight Zuckerberg in a cage match, which did not materialize.

During the time of the interaction in February 2025, Musk was spearheading DOGE, the special advisory created on President Donald Trump’s first day of his second term to eliminate headcount and contracts to shrink the federal budget. Meta donated $1 million to Trump’s inauguration fund in 2020, marking a positive shift in the relationship between the tech company and the administration. The White House last week appointed Zuckerberg to serve on a tech advisory council.

Musk appeared to have favored Meta’s AI models in some of his DOGE-related work. Wired reported in May 2025, citing internal materials, that DOGE used Meta’s Llama 2 to review and classify email responses from federal workers to the January 2025 “Fork in the Road” email offering deferred resignation to employees opposing the administration’s sweeping workforce changes.

Meta declined Fortune’s request for comment.

The ongoing legal dispute

Court filings from August 2025 indicated Musk approached Zuckerberg about assembling a cadre of investors to finance a takeover of OpenAI. According to the documents, neither Meta nor Zuckerberg signed a letter of intent or made a bid for OpenAI.

According to a statement in the filing, Meta was “spending heavily to develop its own Al capabilities” and has been “offering pay packages of $100 million or more to leading Al researchers and attempting to poach OpenAI employees.” 

Meta argued at the time that OpenAI’s request for additional documents was “overly burdensome.”

Altman’s company completed its transition into a more traditional for-profit corporation in October 2025, with Microsoft, its largest external shareholder, gaining a 27% stake in the company and retaining access to its technology through 2032.

In a separate unsealed filing released last week, Musk’s lawyers argued his communications with Zuckerberg should not be included in the litigation.

“Musk’s personal relationships and communications – including with other high-profile individuals – are also tangential and prejudicial,” the lawyers wrote. “Defendants included in their exhibit list for trial, for example, several private exchanges between Musk and Mark Zuckerberg discussing Musk’s political activity and this lawsuit. 

“Those recent communications have nothing to do with Musk’s claims and are nothing more than Defendants’ attempt to stoke negative sentiments toward Musk because of his association with Zuckerberg,” they concluded.

This story was originally featured on Fortune.com

Gen Zers may be turning their tassels, flying the nest, and securing their first full-time jobs—but many are still bankrolled by mom and dad to stay afloat. Now, it’s leaving both the young generation and their parents feeling the squeeze.

Around 64% of parents with Gen Z children, aged 18 to 28, said that their adult kids still rely on them for money, housing, or other financial support, according to a new survey from Wells Fargo. 

And their continued support has led to a money pinch, as 56% reported that assisting their grown-up offspring is straining their own finances. 

But despite assumptions that Gen Zers are living outside their means, most parents aren’t stepping up to finance the lavish lives of their adult children. 

Emily Irwin, head of private wealth planning at Wells Fargo, tells Fortune that they’re actually helping cover essential living expenses rather than extravagant getaways and shopping sprees. Gen Zers are battling a sluggish entry-level job market, stagnating wages, and high cost-of-living while wanting to financially prepare for the future. And parents don’t want to wait until they pass down wealth at the end of their lives to step in.

“[Adult Gen Z] kids who are receiving the financial support are really in this perfect storm,” Irwin says. “They’re feeling uncertain about their career, their profession, and the stability of receiving a paycheck. They’re combining that with a desire to want to save more than they have even in prior years.”

Why parents are giving Gen Z kids their future inheritance now

Irwin says she’s heard from Wells Fargo clients that they want “their dollars in action during their lifetime, versus simply at death,” and it’s fueling an earlier wealth transfer informed by those who were once in their children’s shoes. 

Parents want to turn the tide on the “big inheritance movement” they once benefited from (perhaps too late) in life, Irwin says.

“Having gone through that cycle themselves, receiving an inheritance in their 50s, 60s, sometimes even 70s, is less impactful,” Irwin explains. “They say to us, ‘We got this and we could have really used it when we were starting a family, buying a home, buying a business, paying down our debt, maybe making a career shift.’”

While fronting rent and loaning money to their kids undeniably puts a strain on their wallets, Irwin observes that the financial stress actually largely stems from “a complete lack of communication.” 

Parents and their adult Gen Z kids aren’t being open about this financial support: how much the children really need, when the assistance might end, and if money needs to be paid back. Without any transparency, financial troubles are bound to bubble to the surface. 

“What I really encourage parents to do is have direct conversations with their children,” Irwin says. “Discuss everything from: Is this a gift or a loan, or some sort of a hybrid? Is there an expectation of it being paid back, and if so, with or without interest?…How long do they plan to be able to give financial support?”

Gen Z’s financial and career predicament 

There’s little question that Gen Z is under immense economic and career challenges. 

Last year, around 58% of students who had recently finished college were still looking for their first job, according to a 2025 Kickresume report. Meanwhile, just 25% of graduates of previous generations—including millennial and Gen Xers—found it hard  to land work after college. 

On the financial front, young people are also struggling. Gen Z’s average FICO score slipped three points to 676—39 points lower than the national average of 715, according to a 2025 FICO report. Erin Stillwell, head of payments at Globant, told Fortune last year that “Gen Z is the first cohort facing high inflation, digital credit, and social-media-driven consumption pressure simultaneously.” 

The “perfect storm” of issues has become so intense it’s even keeping many young people up at night. Around seven in 10 Gen Zers said they couldn’t sleep because they’re so stressed about rising prices, rent, and job security, according to a 2025 report from Amerisleep. Contrary to the belief that the young generation loves to spend and expects a life of luxury, most are simply trying to hold on. 

“I wouldn’t say Gen Z is living outside their means,” Irwin explains. “Gen Z’s a little bit in [a] unique position…The last few years, we’ve had higher inflation—that’s a reality. We’ve had higher interest rates—that’s a reality…[Many] of them feel like there’s instability in their job.”

This story was originally featured on Fortune.com

As Americans are barely getting by because of inflation, tariffs, and a cost-of-living crisis, saving for retirement can feel like the priority lowest on the totem pole. 

But multimillionaire serial investor and entrepreneur Kevin O’Leary says saving is more important than ever before. 

“What piece of advice do I give my kids over and over and over again about money?” the Shark Tank star questioned in a recent Instagram video. “Don’t spend it. Save it. Invest it. Let it compound. That’s the gift the market gives you.

O’Leary’s golden rule of investing is straightforward. He says to take 15% of every dollar you earn, whether it’s from your paychecks, side hustles, or birthday money from grandma, and put it directly into the market.

“Just let it compound,” he said. 

View this post on Instagram

A post shared by Kevin O’Leary (@kevinolearytv)

For the average American worker who makes $68,000 per year, that simple rule will pay off in the end, he argued.

“If you make $68,000 a year, the average salary, and you do this your entire life, just 15% of your paycheck, you’ll end up a millionaire at retirement at 65,” O’Leary said.

Does Kevin O’Leary’s math check out?

Most national estimates place the average American salary at roughly $66,000 to $69,000 per year. Assuming O’Leary’s estimate of $68,000, the numbers would break down as follows. 

Assuming O’Leary’s 15% rule, an American making $68,000 per year would save about $10,200 per year, or $850 per month. Invested consistently over a 40-year career, say from age 25 to 65, and assuming the S&P 500’s historical average return of roughly 10%, that $850 monthly contribution would grow to approximately $5.3 million by retirement.

Even using a more conservative average return of 7% would still put the average American in millionaire status, with a final portfolio worth around $2.2 million.

While the math works on paper, it’s becoming more unrealistic for average Americans to save that much money each month. 

For workers in the $50,000–$79,999 income bracket, 55% report feeling behind on retirement savings, and this group is among the most likely to lack adequate preparation. The overall personal saving rate as of mid-2025 sits at just 4.4% of disposable income, according to the Bureau of Labor Statistics — meaning someone earning $68,000 saves roughly $3,000/year toward retirement, on average. Among 401(k) participants specifically, Vanguard data show the median total contribution rate (employee + employer) is about 11.5%, though this applies primarily to those with 401(k) access.

For a household earning $68,000 before taxes, take-home pay is about $52,000 to $54,000 after federal and state taxes, leaving just $3,600 per month for other expenses. 

According to RentCafe, the average rent in the U.S. is $1,740 per month, leaving just about $1,860. Then tack on groceries, which Bureau of Labor Statistics data shows can be as high as $400 per month for a single person. (Now we’re down to about $1,460). 

Then there’s student loan payments (averaging $434 per month) and utilities (about $300 per month). That only leaves $726 — not enough to meet the 15% of earnings saved that O’Leary suggested. 

Even if we assume 15% of the average American’s take-home pay of about $52,000, that means they’d have to invest $650 per month (still making them a millionaire by age 65), but that leaves just $150 per month in discretionary pay.

O’Leary argues, though, that younger generations need to stop spending on unnecessary items.

“The best piece of advice I can give anybody: don’t buy stuff you don’t need,” he insisted. “Invest it instead.”

What other investors say 

O’Leary’s advice largely mirrors the advice of index fund investing long shared by Warren Buffett, who has repeatedly said the average investor is best served by putting money in a low-cost S&P 500 index fund and leaving it alone. 

“Put 10% of the cash in short-term government bonds and 90% in a very low-cost S&P 500 index fund. (I suggest Vanguard’s),” Buffett wrote in a 2013 shareholders’ letter. “I believe the trust’s long-term results from this policy will be superior to those attained by most investors—whether pension funds, institutions, or individuals—who employ high-fee managers.”

Suze Orman, financial advisor, author, and podcast host, has also said Americans need to prioritize saving or investing at least 10% of their earnings each year—particularly given longer life expectancies and rising health care costs in retirement. She’s even argued that 70 should be the new retirement age because Americans aren’t financially prepared enough.

“You likely have plenty saved up to breeze through 15 years or so of retirement. But, people, if you stop working in your 60s, your retirement stash might need to support you for 30 years, not 15,” she wrote in 2017.

This story was originally featured on Fortune.com

The promise of AI-driven productivity has many employees fearing for their heads. But to Marc Andreessen, co-founder and general partner at Andreessen Horowitz, the technology is more of a bogeyman, masking a long-standing business fluke that has quietly lingered in boardrooms for years.

In an interview on the 20VC show with venture capitalist and host Harry Stebbings, the billionaire said AI was the scapegoat for layoffs that are actually a result of overhiring in the wake of the COVID pandemic.

“Essentially, every large company is overstaffed,” he said. “It’s at least overstaffed by 25%. I think most large companies are overstaffed by 50%. I think a lot of them are overstaffed by 75%.” He added, “now they all have the silver bullet excuse: Ah, it’s AI.”

Andreessen’s comments are nothing new for an industry that is pushing back against the “silver bullet excuse” of AI, which some tech leaders including OpenAI’s Sam Altman have coined as “AI washing,” or blaming otherwise normal layoffs on the increased use of AI. 

A long list of business leaders and AI experts have said the labor market is due for a massive upheaval due to AI. Some have already carried out layoffs and attributed them to the tech. Block CEO Jack Dorsey laid off 40% of his workforce in February, saying he thinks “most companies are late” to the AI layoff trend. Australian-American firm Atlassian made a similar move. Meta is also reportedly planning sweeping layoffs thanks to greater efficiency brought about by AI-assisted workers. 

The post-pandemic hiring blitz

Tech companies embarked on a hiring spree in the wake of the COVID pandemic. Following the initial employment shock during the onset of the pandemic, hiring shot up to 8.3 million by May 2020, according to the Bureau of Labor Statistics. The dawn of remote work and the shift to digital opened up a pool of labor that spanned the globe. By June 2022, nonfarm payrolls surpassed pre-pandemic levels. Firms acted like the metaphorical kid in the candy store with talent, grabbing every shiny new candidate that crossed their applicant tracking system, with some, like Amazon, even doubling their headcount between 2019 and 2021. 

But many tech firms have cleaned house following the hiring craze. Amazon has cut nearly 30,000 workers over the past year to reduce layers and remove bureaucracy. In 2023, Google parent company Alphabet cut 12,000 jobs after a pandemic hiring spree. Even Dorsey conceded that some of the Block cuts were thanks to overhiring.

“This entire labor displacement thing is 100% incorrect,” Andreessen said. “It’s classic zero-sum economics.” He said that most coders, for example, are employing AI, which is taking over much of the workload. But that’s not a flashing red light that layoffs are on the way. Instead, it just means more work for those workers as AI boosts productivity rather than cutting labor costs.

The venture capitalist argues that fears of AI-driven mass layoffs stem from the “lump of labor” fallacy, the belief that there is a fixed amount of work in the economy at any given time. “It’s always been wrong, it’s going to be wrong again,” he said.

But recent studies on the impact of AI complicate Andreessen’s assessment. An Anthropic study released earlier this month demonstrated that AI is already theoretically capable of performing the majority of tasks associated with engineering, law, finance, and business. And a study from professional services firm Cognizant mapped out the projected magnitude of AI layoffs this year, finding AI-related job cuts could total more than nine times what they were last year, surpassing 500,000. But that number is still a far cry from the sweeping projections leaders like Anthropic CEO Dario Amodei have made about an AI-related white-collar job apocalypse.

Still, Andreessen thinks AI is a smoke screen for layoffs. He doesn’t believe the technology is sophisticated enough yet to replace human workers.

“AI literally until December was not actually good enough to do any of the jobs that they’re actually cutting,” he said. “It just can’t have been AI.”

This story was originally featured on Fortune.com

Silicon Valley’s startup culture has long sold itself on alluring perks: cold brew on tap, nap pods tucked between standing desks, and even free slippers for their “no-shoes” offices. The pitch was simple: work hard, but live well while building the next big thing.

But as the race for top AI talent accelerates, startups are increasingly leaning on a far more direct incentive: eye-popping paychecks.

Software engineers at venture-backed startups are receiving median base-salary offers of $200,000—a 25% increase from 2022—according to Levels.fyi. In some cases, newly-minted computer science graduates are fielding offers upward of $300,000 annually, sky-high wages once reserved for seasoned engineers at Big Tech giants, said Chris Vasquez, CEO of startup recruiting firm Quantum. 

“Prior to this, I’d probably never seen anyone over $300,000 on base salaries at seed companies,” Vasquez recently told The Wall Street Journal. Now, “They’re able to take home FAANG [Facebook, Amazon, Apple, Netflix, Google]-level cash comp.”

AI itself is helping fuel the frenzy. New tools are making it easier and faster than ever to build and scale companies, lowering the barrier to entry for budding professionals and intensifying competition for a small pool of elite talent. 

At least in the short term, that’s good news for young engineers entering the workforce—despite broader concerns that AI could eventually significantly shrink the number of traditional tech roles. If salaries alone are any indication, demand for the best of the best talent has never been higher.

The battle for AI talent is raging —so companies are dishing out 7-figure paychecks

After the world’s best AI talent spends a few years fine-tuning their skills, their compensation could even stretch into the seven figures. And as industry insiders note, financial equity can be an even bigger draw than base salary for companies with sky-high ambitions.

Employee stock grants alone can range from $2 million to $4 million at a Series D startup, according to Tim Tully, a partner at venture capital firm Menlo Ventures.

“That was unfathomable when I was hiring research scientists four years ago,” Tully, told Fortune last year, noting that those working on foundational AI and theoretical breakthroughs hold the golden tickets to top-tier companies.

At Big Tech companies, the offers are even more eye-watering as firms pour billions into AI, igniting a nonstop tug-of-war for talent among companies like OpenAI, Meta, Google, Microsoft, and Anthropic.

The most intense battle centers over a small pool of fewer than 1,000 AI research scientists who can build today’s most advanced large language models. OpenAI CEO Sam Altman even said last year that the competition intensified to the point where Meta offered signing bonuses as high as $100 million to lure top talent. The ChatGPT-maker’s average stock-based compensation hit a whopping $1.5 million among its roughly 4,000 employees in 2025—the highest of any tech startup in history—the WSJ reported.

Even with sky-high salary promises, uncertainty clouds the AI job market

The boom comes with a familiar caveat: the odds of survival remain slim. 

For every success story that begins in a garage or dorm room, countless companies stall out—even after making a name for themselves. 

Moreover, not every tech worker is cashing in at the top of the market. While a select group of candidates can command eye-popping offers, most new graduates are still landing more modest—but still sizable—paydays. 

The average starting salary for computer science majors is expected to be around $81,500 for the class of 2026, according to the National Association of Colleges and Employers, up 7% from the previous year.

Taken all together, the numbers point to a job market defined by opportunity and imbalance: companies are paying a premium for the very best talent, even as layoffs remain omnipresent and the future demand for tech workers remains uncertain.

This story was originally featured on Fortune.com