Gemini enters wartime management! Google (GOOGL.US) concentrates the AI command chain to Silicon Valley Mountain View's $205 billion AI gamble enters the cashout period

The Zhitong Finance App learned that US tech giant Google (GOOGL.US) is concentrating the AI business leadership at its headquarters in Mountain View, Silicon Valley, California to gain stronger momentum in competition with Anthropic and OpenAI's accelerating AI models and agents, and compete for the world's most dominant cutting-edge artificial intelligence technology. Google's move comes at a time when the market is worried that “AI veteran” Jeff Dean, as well as senior employees such as Sanjay Ghemawat, Oriol Vinyals, and Quoc Le, will weaken the continuity of updates and iterations of Google's flagship AI technology Gemini, which also caused Google's stock price to drop more than 4% on Wednesday.

The tech giant appointed Koray Kavukcuoglu (Koray Kavukcuoglu), who moved to Mountain View in the past year, to take charge of its massive artificial intelligence research and operation system. This also highlights the extreme urgency of Google compressing the decision-making chain and concentrating AI's day-to-day operations on Mountain View in Silicon Valley under model delays, loss of core talent, and competitive pressure from OpenAI and Anthropic.

Meanwhile, Demis Hassabis (Demis Hassabis), the co-founder and long-term CEO of DeepMind, has withdrawn from daily operations to become the chairman of Google DeepMind and chief scientist of Alphabet. Sebastian Borgeaud (Sebastian Borgeaud), who is in charge of a key artificial intelligence programming project, also moved from the UK to California. Others have left the company's London headquarters altogether.

Google's AI Power Center Returns to Silicon Valley: Gemini Rallies in California to Face OpenAI and AnthroPic

Google is concentrating the AI leadership's efforts to reverse trends that have plagued the company since at least 2023. Back then, Google merged two highly regarded science labs that originally operated independently: Google Brain, which is based at the company's Mountain View headquarters, and DeepMind, which is rooted in London. Although the two laboratories merged under the name of Google DeepMind, the researchers are still working on different continents. According to people familiar with the matter, this arrangement made the decision-making process more complicated, and also made talents from both places feel dissatisfied.

As Kavukciuolu officially took over Google's artificial intelligence ambitions, the company was lagging significantly behind its rivals in the AI application competition. The release of the strongest version of its flagship artificial intelligence model is several months behind schedule; in the AI programming automation market, one of the earliest applications of artificial intelligence to achieve high commercial value, many people think of Google's products as an afterthought.

Researchers' dissatisfaction with Google's competitive position continues to grow, triggering a wave of departure and entrepreneurship — including Google's legendary engineer Jeff Dean (Jeff Dean, Google's veteran AI leader) — and recent continuous talent switching to Anthropic and OpenAI. These competitors can not only keep talent at the cutting edge of the artificial intelligence technology market, but also provide huge wealth opportunities brought about by upcoming IPOs. Meanwhile, Google has promised to invest up to $205 billion in capital expenditure this year alone to support its artificial intelligence ambitions, putting pressure on Kavukkuolu to deliver results quickly.

Some of the company's leavers include senior employees from DeepMind's London office. David Silver (David Silver), an outstanding researcher at the office, left his job earlier this year to start a new AI startup called Ineffable Intelligence, and received $1.1 billion in seed funding. According to a source familiar with the matter, DeepMind's Chief Experience Officer Simon Bouton (Simon Bouton) is also planning to leave his job and join the London-based startup CuSPai. Bouton has been with the company for 13 years and is responsible for overseeing operations personnel. Bouton did not immediately respond to media comments.

According to Google, London remains an important hub for the company's AI talent and strategic ambitions. The company recently opened a new office building in London, and Hassabis will stay there to focus on artificial intelligence vision and strategy.

These adjustments over the past year mean that many of Google's key AI leaders driving the Gemini model are now concentrated in the same place. Gemini is the core pillar of Google's artificial intelligence strategy.

Since the start of the artificial intelligence boom, Google CEO Sundar Pichai and co-founder Sergei Brin have often appeared on a floor dedicated to Gemini employees, but according to a Google internal employee, Hassabis has rarely appeared there. Over time, Gemini's Silicon Valley oversight was gradually transferred to Kavukkuolu.

The new arrangement is also reflected in Brin's description of his role in Google's artificial intelligence work. In June of this year, at an event for artificial intelligence researchers and industry leaders, Brin was asked how he divided his work with Kavukkuolu and Hassabis. Brin said he is deeply involved in the Gemini project and works closely with Kavukkuolu.

Brin said, “I'll say, I'll keep poking and tweeting him and the team, like, 'Hey, are you really doing that? ' Sometimes it does get a little intrusive, and I don't deny it. But Corey was really responsible for organizing the teams and letting them deliver perfect results.”

As a longtime leader of DeepMind, Kavukciuolu created a deep learning team and pioneered several major research breakthroughs within the company. Compared to Khasavis, he has always kept a relatively low profile, but has quietly moved up the DeepMind leadership in recent years. Last summer, Kavukkuoglu was promoted to Google's chief artificial intelligence architect.

Deep learning pioneer Yann LeCun (Yann LeCun) served as his mentor during his PhD studies at New York University in Kabukkuolu. Lecoen called him “an excellent engineer” and said he “developed excellent research management skills” during his promotion at DeepMind. Lequin also added that Kavukkuolu's early research helped people believe that artificial intelligence can fundamentally change the way machines process language.

Now, as Google deals with the subsequent impact of the departure of many of its best-known AI leaders, the task ahead will be extremely difficult. Kavukchuolu is currently the only senior Gemini co-leader remaining at Google within the company.

Google quickly centralizes the AI command chain to compete with AI programming agents and enterprise AI

The first focus of Google's future business strategy is to transform the “research federation” that used to be scattered in the traditional systems of DeepMind in London and Google Brain in California into a Gemini product delivery machine with Mountain View as the core and directly responsible to Pichay. Kavukkuolu also oversees Google DeepMind's day-to-day operations and chief AI architect responsibilities, which means model pre-training, post-training, evaluation, computational power scheduling, and collaboration with Cloud, Search, and developer products will be incorporated into a shorter decision chain; Hassabis was transferred to the position of Chairman of Google DeepMind and Chief Scientist of Alphabet, focusing on AGI strategy, social impact, and Isomorphic Labs.

These latest personnel adjustments and transfers do not weaken basic research, but rather split management of “long-term scientific exploration” and “quarterly product delivery” in an attempt to solve the problems of slow decision-making and slow commercialization of research results by cross-continental teams after the 2023 merger.

The second strategic focus is to concentrate resources to make up for the shortcomings of cutting-edge models, AI programming, and intelligent execution capabilities, and upgrade Gemini from a single chat model to a basic operating system throughout Google's entire stack. Gemini 3.5 Pro was delayed by several months from the original June release due to programming capabilities falling short of internal targets, indicating that Google's most urgent task is not to continue to stack generic benchmark scores, but to improve code generation, complex tool calls, long-range task planning, and enterprise-level reliability; at the same time, the company has initiated large-scale pre-training for Gemini 4 and covered developers and enterprise scenarios through the Antigravity intelligent development platform, CodeMender security agents, and low-cost Flash models.

Google's current real AI business goal is to simultaneously embed the same set of model capabilities into Google's search engine, advertising, Workspace, Android, Google Cloud, and Vertex AI, so that a single investment in model development can be repeatedly monetized across billions of consumers and enterprise customers, rather than continuing to let OpenAI and Anthropic take the lead in defining high-value AI applications.

The third focus is to transform unprecedented AI capital expenditure into enterprise AI-side cloud computing revenue, search increments, and verifiable return on capital. Google's parent company Alphabet has raised its 2026 capital expenditure guidelines to US$195 billion to US$205 billion, with capital expenditure of US$44.9 billion in the second quarter and a free cash outflow of US$5.9 billion; however, Google Cloud's cloud computing business revenue also increased sharply by 82% year over year to US$24.8 billion, with backlog orders reaching US$514 billion. Nearly 90% of the Fortune 100 companies have already used Gemini Enterprise, and the model API processing capacity has reached about 22 billion tokens per minute. This also highlights that Google's future growth path is not simply “focusing on the big model,” but rather reconstructing the company's growth path around vertical integration paths such as “self-developed TPU and high-performance network infrastructure - Gemini big model - cloud computing cloud enterprise platform - Google search engine and advertising commercialization - Waymo and AI Pharmaceuticals.”

For investors, organizational centralization has raised the execution limit, but the continued loss of core talent has also amplified the risk of single-point decision-making and innovation gaps. Therefore, the investment-level verification indicators surrounding Google will not only be a model ranking, but whether Gemini's release schedule can accelerate, whether programming and smart devices can increase market share, whether cloud backlog orders can be converted into cloud computing revenue, whether AI+ search engines can maintain advertising monetization, and whether high capital expenditure can be re-converted into positive free cash flow.