and the Space Race.
An AI researcher aged 24 will earn 327x more than Oppenheimer earned while developing the atomic weapon.
Silicon Valley’s AI talent battle just reached a milestone in compensation that makes even the greatest scientific achievements of the previous look financially modest. When Meta The recent offer of $250 million to AI research Matt Deitke over four years (an annual average of $62.5) – with potentially $100 million in the very first year – shattered all historical precedents for scientific and technological compensation. This includes salaries for the development of major milestones in science during the 20th century.
The New York Times Reports stated that Deitke cofounded a startup named Vercept previously led the development Molmo, an AI multimodal system, at Allen Institute for Artificial Intelligence. His experience in building systems that combine images, sounds, text, and other media made him a top target for recruitment. But he is not alone: Meta CEO Mark Zuckerberg According to they also offered a unnamed AI-engineer $1 billion in compensation, which would be paid over several years. What’s the story?
These astronomical sums reflect what tech companies believe is at stake: a race to create artificial general intelligence (AGI) or superintelligence–machines capable of performing intellectual tasks at or beyond the human level. Meta, Google, OpenAI and others bet that whoever makes this breakthrough first will dominate markets worth trillions. This vision, whether it’s a realistic one or just Silicon Valley hype is driving compensation to unprecedented heights.
These salaries are in historical context: J. Robert Oppenheimer led the Manhattan Project, which ended World War II. Earned about $10,000 per year in 1943. Adjusted by the US Government for inflation. CPI Inflation Calculator (19459090) – that’s roughly $190,865 today, or what a senior software developer makes. Deitke, 24, who recently dropped out from a PhD program will earn 327 times more than Oppenheimer earned while developing the atomic weapon.
These numbers are beyond the reach of many top athletes. The New York Times Note: Steph Curry’s four-year contract with Golden State Warriors is $35 million less than Deitke Meta’s deal (although soccer star Cristiano Ronaldo will earn $275 million in this year’s as the The highest-paid professional athlete in the World (19459090). The comparison prompted observers call this ” Talent market, NBA-style. Except that the AI researchers make more than NBA stars.
Racing towards “superintelligence”
Mark Zuckerberg told investors recently that Meta plans on continuing to throw money at AI talent. “because we have conviction that superintelligence is going to improve every aspect of what we do.” He wrote an open letter in which he said that he was a fan of the NBA-style ” talent market. Thesuperintelligent AI was described as a technology that would “begin an exciting new era of individual empowerment,” even though it refused to define what superintelligence is. This vision is why companies treat AI scientists as irreplaceable assets, rather than professionals who are well compensated. If these companies are right, the first company to achieve artificial general or superintelligence will not only have a better product — they’ll also have technology that can invent endless products or automate millions of knowledge worker jobs and transform global economy. The company that controls this kind of technology may become the richest in history.
It’s perhaps not surprising that the salaries of early tech employees pale in comparison with today’s AI researchers. Thomas Watson Sr. was IBM’s legendary CEO. Received 517,221 dollars in 1941. This was the third highest salary in America (about $11.8million in 2025 dollars). The current AI researcher’s compensation package is more than five-times Watson’s peak salary, despite Watson having built one of the most dominant technology companies in the 20th century.
This contrast is even more striking when you consider the collaborative nature of previous scientific achievements. During Bell Labs golden age of innovation, when researchers The director of the lab developed transistor, information theory and other foundational technologies. Claude Shannon, who earned 12 times more than the lowest-paid worker, did. Claude Shannon, who In 1948, he created the information theory at Bell Labs. He worked on a standard salary as a professional while creating the mathematical basis for modern communication.
Who “Traitorous Eight” left William Shockley in order to When they first started, Fairchild Semiconductor – the company that was essentially responsible for the birth of Silicon Valley – split ownership of 800 shares of a total of 1,325 when they were founded. Their seed funding, $1.38 million ($16.11 million today), was a fraction of the amount a single AI researcher commands now.
Even Space Race salary was cheaper
Apollo program offers a striking comparison. Neil Armstrong was the first person to walk on the Moon. Earned approximately $27,000 per year, which is roughly $244.639 in current money. Buzz Aldrin and Michael Collins earned even less than him, earning the equivalent in today’s money of $168.737 and $155,373, respectively. Current NASA astronauts Earn from $104,898 to $161,141 annually. Meta’s AI research will earn more in three days than Armstrong earned in a whole year. A NASA’s technical report from 1970 gives a glimpse into these earnings through an analysis of salary data for the entire engineering field. The report, which used the Engineering Manpower Commission’s data, noted that these industry wide salary curves corresponded to the government’s General Schedule pay scale (GS), on which NASA’s employees were paid.
A chart in the 1970 Report shows that a newly-graduated engineer in 1966 began with an annual wage between $8,500 to $10,000 (about $84,622 to $95,555 today). A typical engineer with 10 years of experience earned $17,000 per year (about $169,244 in today’s dollars). Even the top-performing, elite engineers with 20 years’ experience reached a peak salary of $278,000 in today’s dollar. A top AI researcher such as Deitke could now earn this amount in a matter of days.
The AI talent market is different.
It’s not the first time that technical talent has been rewarded with a premium price. In 2012, three University of Toronto professors published AI research and were paid a premium. The auctioned themselves to Google for $ 44 million (about $62.6 in today’s money). In 2014, a Microsoft executive compared the salaries of AI researchers to NFL quarterback contracts. But today’s figures dwarf those precedents.
There are several factors. Explain the unprecedented compensation explosion. We’re in an unprecedented period of industrial wealth concentration. Gilded Age (19459090) of the late 19th Century. Today’s AI race is different from previous scientific endeavors. It features multiple companies with trillion dollar valuations competing for a very limited talent pool. Only a few researchers have the expertise to work on the most advanced AI systems, especially in areas such as multimodal AI. Deitke is an expert in this area. AI hype is currently “the next big thing” as high in technology.
The economics of the project also differ fundamentally. The Manhattan Project Cost $1.9billion total (about $34.4billion adjusted for inflation), whereas Meta alone plans to invest tens or hundreds of billions each year on AI infrastructure. The potential payoff for a company with a market cap approaching $2 trillion dwarfs Deitke’s compensation package.
An executive told The New York Times it straight: “If I’m Zuck and I’m spending $80 billion in one year on capital expenditures alone, is it worth kicking in another $5 billion or more to acquire a truly world-class team to bring the company to the next level? The answer is obviously yes.”
The New York Times reported that young researchers have private chat groups in Slack and Discord where they share details of offers and negotiation strategies. Some hire unofficial agents. The NYT reports that companies offer not only massive cash and stock packages, but also computing resources. Reports indicate that some potential employees were told that they would receive 30,000 GPUs – the specialized chips which power AI development.
Tech companies also believe they are in an arms race, where the winner will reshape civilisation. The race for artificial general Intelligence has no apparent ceiling, unlike the Manhattan Project and Apollo programs, which had specific, narrow goals. Researchers call this a “intelligence explosion” which could lead to cascading discoveries if it ever happens.
It’s unclear whether these companies are developing humanity’s ultimate labor-replacement technology or just chasing hype, but we have certainly come a long way since the The $8 per day Neil Armstrong received to go to the moon, which is about $70.51 today. Vercept cofounder Kiana Eshsani was the first to accept Meta’s offer after Deitke accepted it. Joke in social media “We look forward to joining Matt on his private island next year.”
Benj Edwards is Ars Technica’s Senior AI Reporter and founder of the site’s dedicated AI beat in 2022. He’s also a tech historian with almost two decades of experience. In his free time, he writes a nd records music, collects vintage computers, and enjoys nature. He lives in Raleigh, NC.
