The Golden Age of AI: Tech Giants Delegate Power, Top Talents Earn a Seat at the Table

The Golden Age of AI: Giants Devolve Power, Prodigies Take Their Seats In the tech industry, two photos have sparked widespread discussions recently.

The first photo was taken in Silicon Valley, the United States. Mark Zuckerberg, CEO of Meta, wears his usual gray T-shirt with a bright smile; standing next to him is Alexandr Wang, in a black hoodie with slightly messy hair. This founder, who dropped out of MIT at 19 to found Scale AI, became a core figure of Meta's Superintelligence Lab after Meta acquired a 49% stake in Scale AI for approximately 14.3 billion US dollars.

The second photo was taken in Shenzhen, China. At a high-profile official symposium, Ma Huateng, CEO of Tencent, sat upright with a red nameplate in front of him. Next to him was a young man wearing black-rimmed glasses and a dark shirt, with the name "YAO Shunyu" on the nameplate in front of him. This post-95s (born in 1998) PhD from Princeton University, former OpenAI researcher, and technical researcher who proposed the *Tree of Thoughts* framework, officially joined Tencent at the end of 2025 as the Chief AI Scientist of the "CEO/President's Office". At this symposium gathering figures from the political and business circles, Ma Huateng sat next to this young technical leader.

These two photos spanning the Pacific, though taken in different scenarios, jointly point to a repeatedly discussed business proposition in the AI era: in this new world dominated by computing power and models, traditional business giants and top technical talents are forming a rarely seen deep cooperation and symbiosis relationship: giants begin to devolve power, and prodigies take their seats at the table. This is not a simple move to poach talents with high salaries, but a redistribution of resources, computing power and decision-making space.

The Failure of the Old Order: When Capital Is No Longer Omnipotent

In the mobile Internet era over the past two decades, the bosses of large tech companies were the rule makers. Capital was the leading force, and huge R&D teams and engineers formed the cornerstone of large enterprises. As long as the business model was proven feasible, large companies could recruit tens of thousands of engineers with high salaries and conquer any track with their scale advantage.

However, the advent of the large model era has fundamentally impacted this set of logic.

On the journey to AGI, what determines the ceiling of the model is no longer the number of people, but the cognition of a very small number of top researchers. There are only a handful of scientists in the world who are truly capable of leading the training of cutting-edge models at the trillion-parameter level and have in-depth understanding of underlying architectures such as Transformer variants, MoE sparse architectures, and RLHF alignment algorithms.

For business magnates like Zuckerberg and Ma Huateng, they can call the shots in capital operation, computing power procurement and application distribution. But when facing problems that touch the boundary of human intelligence such as "how to break through the bottleneck of Scaling Law" and "how to design a more efficient inference engine", huge checkbooks cannot directly get the answers.

Capital is still powerful, but pure capital itself can no longer directly buy the most cutting-edge judgment. Big bosses begin to use a more equal attitude, top-tier resources and even core powers and responsibilities in exchange for opportunities for in-depth cooperation with these young researchers. The so-called "giants devolve power" refers to the core decision-making space that used to belong only to founders and senior veterans.

Zuckerberg and Alexandr Wang: Deep Alliance Under Huge Chips

Zuckerberg's anxiety about AI has become particularly obvious in the past two years. Against the background that the market response after the release of Llama 4 was less than expected, he needed a disruptive move to stabilize Meta's position in the AI competition.

Acquiring a 49% stake in Scale AI for approximately 14.3 billion US dollars and bringing Alexandr Wang and part of his team into Meta is a key card Zuckerberg played. Scale AI is one of the world's leading data labeling and data service companies, occupying an infrastructure-like position in the data link of large model training. Alexandr Wang not only has operational experience in the data service field, but also has quite keen business and technical intuition.

In the photo, Zuckerberg wears a casual T-shirt, and Alexandr Wang wears a hoodie. This casual dressing style is interpreted by the outside world as that the relationship between the two has gone beyond the traditional "superior-subordinate" relationship, and is closer to a Silicon Valley-style partnership alliance.

Zuckerberg provides Alexandr Wang with huge funds, large-scale computing power clusters and Meta's user data; Alexandr Wang, with his team, builds the next-generation model for Meta. Zuckerberg also specially set up the Superintelligence Lab, granting it a considerable degree of autonomy. This can be regarded as a major concession of capital to top technical judgment, and also a typical sample of "giants devolving power": what Zuckerberg gave up is not only the budget, but also the organizational structure and R&D autonomy.

Ma Huateng and Yao Shunyu: The Chinese Giant's Eagerness for Talents

Looking back at China, the photo of Ma Huateng and Yao Shunyu appearing in the same frame reflects another kind of eagerness of Chinese tech giants in the deep water zone of AI.

In the last round of Internet wave, Tencent accumulated a huge user base and cash flow relying on WeChat and its gaming business. But in the large model era, facing the expansion of ByteDance's Doubao and DeepSeek's competition in underlying technologies, although Tencent has launched the Hunyuan large model, the outside world has been discussing its progress in the reserve of top AI talents and cutting-edge research.

Yao Shunyu's joining is an important move of Tencent in its AI talent strategy. As a former OpenAI researcher, he has profound accumulation in AI Agent and complex reasoning. Tencent introduced him as the Chief AI Scientist of the "CEO/President's Office", and he initially served as the head of the AI Infra Department and the Large Language Model Department at the same time. Later, his powers and responsibilities gradually expanded, and by September 2026, he had taken full charge of the three lines of model, computing power and data. It should be noted that the previously circulated statement of "salary of over 100 million yuan" on the Internet has been refuted by Tencent's official, so it is not appropriate to specify the salary, but the importance Tencent attaches to this talent introduction can be seen from the position arrangement.

In that photo, Yao Shunyu sits next to Ma Huateng. This is not a simple "audience seat", but a public endorsement: Yao Shunyu is not the person arranged to observe at the table, but the person placed at the core table. Ma Huateng is using this arrangement to send signals to the outside world and the inside of the company: Tencent is willing to hand over core resources to young technical leaders like Yao Shunyu.

Big bosses understand business, scenarios and product experience; young researchers understand algorithms, architectures and underlying logic. Tencent has rich landing scenarios (social networking, games, office, video), and people like Yao Shunyu are expected to inject stronger AI capabilities into these scenarios. This is a typical complementary advantage.

Why the "Young People"?

Many people will wonder why big bosses no longer give priority to relying on those senior veterans with deep qualifications and rich experience, but bet on these researchers in their twenties and thirties?

An important reason is that in the large model era, the iteration speed of underlying technologies has, to a large extent, exceeded the rhythm of traditional experience accumulation.

In traditional software engineering, the amount of code and technical years are important moats. But in the AI field, the cycle of paradigm shift has been greatly shortened. From the early GPT-3 to reasoning models, AI is reconstructing its own logic at an unprecedented speed.

These young researchers are not bound by the thinking of the old era. At their most perceptive age, they stand directly in the world's most cutting-edge laboratories. They dare to break the rules, dare to subvert themselves, and can absorb the latest knowledge at a relatively fast speed. Big bosses may just value this point: in this AI competition, the pure curiosity and willingness to break through of young researchers may be more capable of coping with the unknown technical fog than step-by-step experience.

Conclusion: In the Golden Age of AI, Who Sits at the Table

These two photos are a microcosm of the intersection of Silicon Valley and Shenzhen in the AI era.

In the AI era, the way capital and technology combine is being redefined. The traditional employment relationship is evolving towards a deeper partnership. Big bosses provide ammunition and stages, and young researchers output hardcore judgment and technology. The two are not simply one leading the other, but need each other and advance and retreat together in a tech competition that determines the future direction.

Whether at Meta's headquarters in Mountain View or in Tencent Building in Shenzhen, this kind of cooperation points to a reality: in this golden age of AI belonging to superintelligence, the scarcest resource may not be money, nor computing power, but the young brain that can understand and create the future. Giants devolve power, prodigies take their seats at the table, which is not only the survival strategy of tech giants, but also shapes the direction of technological evolution to some extent.

This article is from the WeChat Official Account "Tech Business", author: Tech Business, published with authorization from 36Kr.