Technical giants pay millions of dollars

Meta CEO Mark Zuckerberg suggested $ 100 million to sign bonuses in Top Openai.

David Paul Morris | Bloomberg | Gets the image

Artificial intelligence arms racing is heated, and when the technological giants go to the top, they hang millions of dollars in front of a small pool of specialist talents in what has become known as AI talent.

It sees large technology firms like Meta. Microsoftand Google Company for the best AI researchers, trying to strengthen your artificial intelligence units and dominate the multi -billion dollar market.

Recently, Meta CEO Mark Zuckerberg has started expensive hiring to improve new AI Superintelligence laboratories. Here were poachers The scale of the co -founder of AI Alexander van As part of a $ 14 billion startup investment.

Meanwhile, Openai Sam Altman Executive Director said recently Bonuses for $ 100 million signs And even higher compensation packages.

If I’m going to spend a billion dollars on construction (AI) models, $ 10 million for engineer – a relatively low investment.

Alexandra voica

Head of Corporate Affairs and Policy in Synthesia

Google is also a player in talent, tempted Varun MohanCo -founder and CEO of Artificial Intelligence Start Windsurf to join Google Deepmind in a $ 2.4 billion agreement. Meanwhile, Microsoft AI quietly hired two dozen Google Deepmind staff.

“In the software engineering space, intensive competition for talents was held even 15 years ago, but since artificial intelligence became more capable, researchers and engineers specializing in the field remained relatively stable,” said Alexander Voika, the head of the CNBC AI Video Platform.

“You have this situation with the demand and supply when the demand is now flying, but the supply was relatively constant, and as a result, there is inflation (wages),” added Voica, a former meta employee, and now a counselor at Mohammed Bin’s artificial intelligence university.

Voica said multimillion-dollar compensation packages are the phenomenon that the industry has never seen before. “

That’s what AI’s Talent Talent is:

Construction of AI models costs billions

“Companies that create products pay for the use of these existing models and build them on top, so the capital costs are lower, and not much pressure on the burning of money,” the vocation said. “A space where everything is very hot from the point of view is companies that build models.”

AI experts are in demand

The average salary for machine training in the US is $ 175,000 in 2025 according to the data.

Pixelonestocker | Moment | Gets the image

Machine training engineers are experts from the II who can build and train these large language models – and the demand for them is high on both sides of the Atlantic.

“There is, of course, a great increase in demand for AI-focused AI analytics, and in machine learning, in particular, so people who work with large linguistic models, and people who place more advanced or backed by GPT, or more advanced AI-made technologies.”

This includes the “slender talent pool” of experienced professionals who have worked for many years in the field, as well as scientists on II research, who have completed the PhD in the top five universities in the world and are subjected to technical giants after graduation.

Reportedly this leads to Mega Pay packages Offering 250 million dollars A 24-year-old AI Matt Daitke genius, which came out of the PhD, at the University of Washington.

Meta sent CNBC to Zuckerberg’s comments to information where Facebook founder said there was an “absolute prize” for the highest talent.

“A lot of specifics that have been reported are not accurate in themselves. But it is a very hot market. I mean how you know, and there are a small number of researchers who are the best in demand in all different laboratories,” Zuckerberg said.

“The amount spent on a set of people is actually still small compared to shared investments, and all when you talk about Super Intelligence.”

Litvinoff estimated that in the London market engineers and major engineers are currently earning a six -digit salary within 140,000 to 300,000 pounds for older roles.

In the US, the average salary for the Mashing Learning Engineer is $ 175,000, reaching almost $ 300,000 at a higher end, by -real.

Startups and traditional branches remain behind

Because the technological giants continue to breed the best minds in the II bait mammoth salaries, there is a risk that startups remain behind.

“Some of these startups trying to compete in this space of construction models are difficult to see for them a way forward because they are stuck in space: models in construction are very expensive, but companies that buy these models, I do not know if they can afford to pay the prices that cover the construction of the model,” Voica noted.

Mark Miller, founder and CEO of Resurevision.ai, Recently told the journal Startups The fact that this talent war also created a “massive gap” in traditional fields.

“Whole industries, such as insurance, healthcare and logistics, cannot compete for salaries. They need innovation but cannot access talents,” Miller said. “The current situation is absolutely impossible. You can’t have one branch that retains all talent and the other.”

Voica said AI experts will have to make a choice. While some will take higher wages and Big Tech bureaucracy, others will lean toward startups where wages are lower, but employees have more ownership and influence.

“In a big company, you are, in fact, a cog in the car, whereas in the start you can have a great influence. You can have a great influence through your work and you feel it is an impact,” the warrior said.

Until the price of construction of AI models decrease, however, there may be high salaries for AI talents.

“While companies have to spend billions of dollars to build a model, they will spend tens of millions or hundreds of millions to hire engineers to create these models,” Voica added.

“If tomorrow is suddenly, the cost of building these models is reduced by 10 times, the salaries I might expect will also decrease.”

Source link