Tech Trader: AI Developers for $100 Million? The Next AI Race Is Here. — Barron's
Dow Jones Newswires ·
By Adam Levine
As Big Tech ramps up spending on the AI race, much of the focus is around Nvidia chips, data centers, and the energy grid. But artificial intelligence is quickly raising the costs for human capital, as well, at least for the most prominent AI developers.
Meta Platforms, in particular, is reshaping the AI landscape by raising the cost for top researchers, and only the well-heeled may be able to keep up. For Meta CEO Mark Zuckerberg, the hiring is part of his existential worries. His antennae are always up, looking for threats to his company.
Zuckerberg has taken a page from Intel's third — and arguably most influential — CEO, Andy Grove. In a 1996 book titled Only the Paranoid Survive, Grove offered lessons for how leaders should avoid disruption.
"When it comes to business, I believe in the value of paranoia," Grove wrote. "Business success contains the seeds of its own destruction. The more successful you are, the more people want a chunk of your business and then another chunk and then another until there is nothing left. I believe that the prime responsibility of a manager is to guard constantly against other people's attacks."
When Instagram and WhatsApp looked like a threat to Facebook's social-media dominance, Zuckerberg bought them at high prices. "It is better to buy than to compete," he wrote in a 2008 email that was revealed in the FTC v. Meta antitrust trial.
Fast forward to today, and Zuckerberg is using Meta's hefty operational cash flows — $91 billion in 2024 — to navigate the rapidly changing AI world, another inflection point that could disrupt the company. In 2025, Meta projects spending some $70 billion building AI data centers. Amazon.com, Microsoft, and Google parent Alphabet are spending even more than that. But Meta is unique in that it isn't renting out these AI servers in the cloud, but rather using all of them for its own purposes. In that respect, it is outspending the others, and each of its researchers has more computing power at their fingertips.
But Meta still has work to do. While its Llama models were initially well-received, the fourth version faced issues when it was released earlier this year. More than three months after announcing its most advanced version of Llama 4, called Behemoth, it still hasn't been released.
Enter Zuckerberg's paranoia and a new AI race: spending on talent.
First, in a structure that avoids regulatory scrutiny, Meta bought 49% of start-up Scale AI for $14 billion. Scale AI provides data-labeling services to Meta and other companies making AI models. The deal bought access to Scale's CEO and founder, Alexandr Wang, who subsequently was named the head of a new "superintelligence" unit within Meta. Superintelligence is a machine that can do any intellectual task at least as well as the best human, and it is the ultimate goal of AI model makers.
Shortly after the deal, Scale AI announced significant layoffs, boosting the idea that Zuckerberg was more interested in Wang than any bigger company asset.
"We're streamlining our data business to help us move faster and deliver even better data solutions to our GenAI customers," Natalia Montalvo, a spokesperson for Scale AI, told Barron's. "We also plan to make significant investments and hiring across our enterprise and government AI businesses."
In 2024, the average full-time U.S. software developer in the private sector made $183,000 a year, according to the U.S. Census Bureau's Annual Social and Economic Supplement. The highest developer salary in the survey was $1.1 million. That compares to a 2024 average of $65,000 for all U.S. workers.
For the best AI developers, that's just a start.
After essentially spending $14 billion on one employee, Zuckerberg went after the top AI researchers in the world. The Wall Street Journal reported that Wang would have a team of 12 other new hires. Five others confirmed they were hired by Meta through LinkedIn and X posts. The salaries are a closely held secret, but poaching top talent from Google and well-funded OpenAI surely didn't come cheap.
On the Uncapped podcast in June, OpenAI CEO Sam Altman said that Meta was offering $100 million signing bonuses to his employees. Meta has denied reports of massive pay packages. Whatever the cost, it's a small price to pay for Zuckerberg if he believes the future of his company is at stake.
"The market is setting a rate here for a level of talent, which is really incredible and kind of unprecedented in my 20-year career as a technology executive," Meta Chief Technology Officer Andrew Bosworth said in a June 20 CNBC interview. "But that is the great credit to the individuals who, five or six years ago, put their head down and decided to spend their time on a then-unproven technology, which they pioneered, and have established themselves as a relatively small pool of people who can command an incredible market premium."
In the same interview, Bosworth said that OpenAI is matching some of these offers — and that gets to the root of the issue. The barrier to entry was already high for AI research, but now it may have gotten higher. Google and Microsoft may be able to match Meta's AI investment, and so can OpenAI, to a more limited extent. It's a win for AI scientists, but good luck to the start-ups trying to play in the field.
Write to Adam Levine at [email protected]
To subscribe to Barron's, visit http://www.barrons.com/subscribe
Source:
Dow Jones Newswires