Want to read Slashdot from your mobile device? Point it at m.slashdot.org and keep reading!

 



Forgot your password?
typodupeerror
AI Businesses Technology

AI Arms Race Drives Engineer Pay To More Than $10 Million (ft.com) 32

Tech companies are paying AI engineers unprecedented salaries as competition for talent intensifies, with some top engineers earning more than $10 million annually and typical packages ranging from $3 million to $7 million. OpenAI told staff this week it is seeking "creative ways to recognize and reward top talent" after losing key employees to rivals, despite offering salaries near the top of the market.

The move followed OpenAI CEO Sam Altman's claim that Meta had promised $100 million sign-on bonuses to the company's most high-profile AI engineers. Mark Chen, OpenAI's chief research officer, sent an internal memo saying he felt "as if someone has broken into our home and stolen something" after recent departures.

AI engineer salaries have risen approximately 50% since 2022, with mid-to-senior level research scientists now earning $500,000 to $2 million at major tech companies, compared to $180,000 to $220,000 for senior software engineers without AI experience.

AI Arms Race Drives Engineer Pay To More Than $10 Million

Comments Filter:
  • None? This is badly.

    • I definitely know people making over $2M/yr, most of it in RSUs. $10M seems like director/VP kind of money for these companies, but it doesn't seem unreasonable.

      It's worth so much more in investor capital for companies to pay big bucks for the *apperance* of a technical acumen in AI that they're willing to have a very small number of high profile experts making the big dollars just to keep the money flowing in. Even if said experts are doing jack shit, are just talkers, or working their own agenda. Ultimate

  • That's a lot of money, considering we don't have anything even remotely close to AI.

    • In the marketplace, nobody cares about pedantic definitions. It's a product that is attracting a lot of eyeballs and followers these days, whatever you call it.

    • That's a lot of money, considering we don't have anything even remotely close to AI.

      Paying this crazy money only makes sense before the technology becomes mature. Once we have something that is obviously working, the compensation decreases because it's just engineering and no longer high-risk research.

      As for the statement that we don't have anything remotely close to AI. That may be arguably true for chatbots, but that's a small part of the AI push. There are many AI projects that are already very promising in the near term. Some involve generative AI, but many do not. Viewed in the v

    • by gweihir ( 88907 )

      It is insanity, greed, hype and FOMO. The money invested and the ages paid are totally decoupled from the value of the product (low as that is) now. Usually that is a sign that the end of the mass-hysterics is near.

    • by allo ( 1728082 )

      You don't need what you mean by "AI", thinking of sci-fi movies, to have AI that is worth the money.
      You want an "I, robot". Users want a bot that finds the content in the search results page they were looking for without having to click 20 clickbait links themselves.

      • While all that may be true, your description of what users want is AI, it's just a search engine.

        It's kind of the difference between telling the computer "summarize the causes of the last five major wars" and "tell me what the next major war will be, when it will happen, who the belligerents are and the most likely outcome and duration", and being able to explain that answer.

  • by dgatwood ( 11270 ) on Tuesday July 01, 2025 @01:52PM (#65489248) Homepage Journal

    How many years do they have to work there before they get the bonus? Because $10 million is more than enough to retire, even in the Silicon Valley. So they can probably assume that most of these engineers will work there until the bonus pay date, and then retire and do whatever they want to do instead of what someone else tells them to do.

    After all, most engineers are driven less by money and more by wanting to do cool stuff. If they have enough money to be able to only do cool stuff and never have to worry about money again, why would they want more money? Why would they choose to do what other people want them to do, when they can do the even cooler stuff that they want to do?

    Bonuses that big tend to be counterproductive.

  • It's not about top talent, it's about top level access to the training recipes and architectures. That's what Meta is buying. Even if you try to compartmentalise, a lot of people need a high level overview to do architecture exploration. Once the exploration is finished, their knowledge of the results is worth a lot more than their talent.

    Due to OpenAI and Anthropic wanting to portray a goody two shoes image, their means to keep that knowledge locked up as trade secrets is very limited.

    • Top talent is likely to have originally developed the training recipes and the inference technology, or at least worked closely with it. Hire those guys and they can write an improved version 2 for Meta.

    • The recipes are not secret. Most AI researchers know the full recipe. DeepSeek has published detailed papers on all their techniques. Meta and DeepSeek have published their full weights and models, and there are hundreds of models from them and many other companies published on Hugging Face. The math involved is simple Linear Algebra and Calculus - first year undergrad math at every university. DeepSeek is staffed by new grads with 1-2 years experience.

      That said, there is probably a few minor properiet

      • After a certain modest amount (less than you think), money can't buy happiness.

        Only people who have never had money say that.

        Money predictably and steadily reduces your stress the more you get, and once you've got enough to start having rich people problems, those problems are still less stressful than what the 99% go through daily.

        My salary has steadily increased at a well-above-average rate. I'm an over-90% earner, and I can't say the happiness I've been able to buy has slowed down, or shows any signs of slowing down.

        Overall, I rate your claim false.

      • Yeah, I always hear that money can't buy happiness bullshit and maybe that's true, but it can buy everything else, which would lead to happiness. So therefore, yes, money can buy happiness. That's why we need all of us to be poor, so those "better" people can be happy!

  • So, I'm having a hard time seeing how this works in reality. So you hire 10 engineers for 1 billion in expenses to the company. How do you measure value each eng brings to the company? At review time can you ask for $10M if you hit your targets?

    In short, where does the money come from to pay them? How is this justifiable or sustainable? Especially if you have little to no revenue?

    To me, this is just megalomania.
    • by GoTeam ( 5042081 )

      So, I'm having a hard time seeing how this works in reality. So you hire 10 engineers for 1 billion in expenses to the company. How do you measure value each eng brings to the company? At review time can you ask for $10M if you hit your targets?

      Wouldn't that be $100,000,000.00 per year if 10 engineers split 1 billion $?

      • Based on this: "Meta had promised $100 million sign-on bonuses to the company's most high-profile AI engineers."
        So, yes.
        How about the Jony Ive OpenAI deal? 6.5B all stock = 120million / employee (55 employees)
        You have to make the money somewhere to pay it out.
        For companies without products these valuations seem nonsensical.
    • So you hire 10 engineers for 1 billion in expenses to the company.

      So, I'm having a hard time seeing how this works in reality.

      I bet you are. Math works very differently where you're from than where the rest of us are. That's going to make reconciliation with our reality very difficult for you.

  • If I were starting out I sure would be hitting the books learning everything I could about the various aspects of AI technology. I would be looking at all of the open source implementations and getting a good feel for how they work. I would be attempting to craft my own LLM's, and try to build 'small language models' that are specific to limited domains where a company might have loads of training data but no expertise. I would attend all of the AI conferences and get to know as many people in the business

  • The end of this hype myst be near. And the collapse at its end will probably a lot worse than expected.

  • Certainly those payments look hyped up a lot, but when I compare them to the salaries of CEOs who are just spouting generic sentences like a very poorly made LLM all day, those numbers are not high at all.

The next person to mention spaghetti stacks to me is going to have his head knocked off. -- Bill Conrad

Working...