In 2025, Silicon Valley witnessed a talent battle of unprecedented proportions, one capable of rewriting industry records. When Mark Zuckerberg extended a six-year, $1.5 billion salary package to Andrew Tulloch, the entire tech world was shaken. The scale of this figure is staggering: it’s three times the total career earnings of an NBA top-salary player, exceeds the valuation of some mid-sized tech companies, and is nearly 200 times Tulloch’s original salary at Meta. On the AI track, where capital and technology intersect, such an offer would be hard for most to refuse.
Yet, Tulloch’s response—an outright “no”—flipped the script on the industry’s understanding of talent value. This wasn’t a simple salary decision, but a direct confrontation between technological ideals and commercial interests. In the accelerating pace of AI technology and the fierce competition between industry giants, Tulloch’s decision ripped open the underlying layers of the industry’s power struggle, offering a critical perspective on the future of AI, talent dynamics, and ecological transformation, with impacts far beyond an individual career choice.
The Value Battle Behind the Sky-High Salary
Zuckerberg’s poaching attempt was no spur-of-the-moment decision. By 2025, the AI race had entered a white-hot phase, and Meta was willing to go to any lengths to secure key technical talent in order to strengthen its ecosystem. Meta’s baseline for top AI researchers had already climbed to $2 million annually. But the $1.5 billion offer for Tulloch was an “epic” one—$1 billion guaranteed over four years, plus stock and bonuses—a signal of the company’s anxiety over Tulloch’s open-source tech approach at TML. As a former Meta employee of 11 years who led neural network optimization research (with papers cited over 7,000 times) and later contributed to the development of GPT-4.5, Tulloch’s expertise was essential to Meta’s AI ambitions. His specialties in distributed training systems, model compression, and multimodal interaction were exactly the gaps Meta needed to fill to build its AI ecosystem.
However, the reason behind his rejection ran far deeper than just the salary. In Tulloch’s view, the offer was essentially a “technological sellout” deal. Meta was asking him and his team to refrain from certain areas of research for six years, trying to bring his distributed training framework into a closed ecosystem—this clashed directly with TML’s open-source vision of making AI as freely available as water or electricity. More critically, he was frustrated by the bureaucratic processes at Meta. The efficient “write code in the morning, run it in the afternoon” research pace that TML thrived on had become a luxury in Meta’s corporate structure. From his earlier rejection of OpenAI’s first offer in 2016 due to “insufficient technical autonomy,” to leaving Meta in 2023 for OpenAI, Tulloch consistently placed control over his technical direction above material rewards. His “no-compromise on ideals” stance made his refusal of $1.5 billion seem entirely logical.
The Entrepreneurial Blueprint of a Technological Idealist
Tulloch’s decision wasn’t a gamble but a calculated move based on a firm belief in the future of technology. In November 2024, he co-founded Thinking Machines Lab (TML) with former OpenAI CTO Mira Murati, securing over $2 billion in investments. TML’s core mission is to build the next-generation AI operating system, targeting key use cases for AGI such as enterprise AI automation and complex decision systems. As the co-founder and chief researcher, Tulloch’s vision for the company is disruptive: through open-source architecture, TML aims to break the AI infrastructure monopoly of the tech giants, enabling developers to build intelligent agent applications just like using Windows or Android. If successful, this approach would directly challenge Meta’s “walled garden” AI ecosystem—an outcome Zuckerberg was willing to pay handsomely to prevent.
The 37-year-old Australian scientist, whose academic journey took him from a bachelor’s degree at the University of Sydney to a master’s at the University of Cambridge, and who spent 11 years at Meta, has always centered his work around the freedom of AI infrastructure. Known as the “invisible architect” of AI infrastructure, his research in neural network optimization and model efficiency has already had a profound impact. His role in GPT-4.5’s development gave him deep insight into the technical bottlenecks and breakthrough paths for AGI. This strategic foresight has made TML’s entrepreneurial vision far from a pipe dream. The team plans to launch the first version of its intelligent operating system in 2026, with industry analysts speculating that it could become the “water and electricity infrastructure” for AGI, potentially reshaping the competitive landscape of the AI industry. It’s this faith in technological change that allowed Tulloch to remain clear-headed in the face of $1.5 billion: he values building the technical foundation that will change the industry far more than short-term wealth.
Reconstructing the Value Framework of the AI Industry
Tulloch’s rejection sent ripples throughout the tech world, significantly reshaping the value logic of the AI industry. The most immediate impact was the collapse of the talent pricing system. According to Menlo Ventures, the salaries for top AI researchers in 2025 had surged from the $1 million range to $200-500 million. However, Tulloch’s case demonstrated that salary figures were no longer the sole measure of talent value. Alignment with technical vision, organizational freedom, and strategic influence—these “non-material assets” are becoming the core demands of top-tier talent. The fact that no one from TML chose to leave for Meta further underscores this point. A flat organizational structure (where senior researchers are still considered “technical staff”) and a culture driven by a shared mission are proving to be more valuable than a high salary.
The deeper impact, however, lies in the intensifying battle between open-source and closed ecosystems. TML’s open-source ideology is in direct contrast to Meta’s closed ecosystem, and Tulloch’s rejection is like a shot in the arm for the open-source movement. As he wrote in his email: “It’s truly absurd to offer $1.5 billion to buy a chance to change the rules of the industry.” This sentiment is resonating widely—by 2025, the financing scale for open-source AI projects had grown 300% year-over-year, with more and more top scientists choosing to join startups rather than tech giants. The industry is beginning to recognize that the ultimate competition in AI is not just a matter of technical power but also an ideological battle: while closed ecosystems might monopolize technology in the short term, open-source collaboration is more likely to accelerate the breakthrough of AGI.
In the end, Tulloch’s “billion-dollar rejection” leaves behind a profound question about the tech industry’s values. In an age of runaway capital and technology, Tulloch’s decision proves that top talent isn’t driven by astronomical salaries, but by the opportunity to guide technology along an idealistic path that can change the world. When $1.5 billion becomes Silicon Valley’s most expensive “rejection memento,” a new industry consensus is forming: those who refuse to compromise on short-term profits in favor of their technological ideals will ultimately shape the future of AI.
