A New Era of AI Chips: Google Challenges Nvidia

Alphabet recently announced that it is developing its own TPU processors to train AI models, and now reports indicate that Meta is seriously considering using these chips in its data centers. This development could have significant implications for the AI hardware market and directly impact Nvidia’s position, which has so far practically monopolized the GPU segment in data center infrastructure.

Data centers are a key pillar of Nvidia’s business, accounting for approximately 90 percent of the company’s revenue. Nvidia’s dominant GPUs have been used to train and operate AI models in the largest cloud environments worldwide, providing a stable and high-margin source of revenue. Meta’s decision to shift part of its infrastructure to Google’s TPUs means that Nvidia could start losing key clients in this strategic segment. Even a small reduction in computing power in data centers could have a measurable impact on the company’s financial results and growth rate in its most profitable business area.

At the same time, Alphabet is expanding its TPU infrastructure and, in collaboration with Broadcom, is developing chips that offer higher performance with lower energy consumption. The flagship Gemini 3 model, trained primarily on TPUs, demonstrates that Google is actively testing and monetizing its processors in real-world AI scenarios, giving the company a technological edge and potentially new revenue streams. For Nvidia, this translates into growing competition not only in raw computing power but also in energy efficiency and cost effectiveness, which are critical factors for scaling data centers.

The market has already reacted to these developments. Alphabet’s stock rose in after-hours trading, while Nvidia’s shares declined, reflecting investor concerns about potential threats to GPU dominance in data centers. Broadcom, as Google’s partner in TPU production, received an additional boost, with its stock climbing more than ten percent, highlighting the increasing importance of the ecosystem surrounding the new AI chips.

For investors in the markets, Nvidia’s situation is becoming particularly significant. As Meta and other technology leaders potentially adopt TPUs on a larger scale, Nvidia may need to reconsider its pricing strategy and accelerate investment in new GPU generations to maintain its edge in the most profitable segment of data centers. Such actions could directly affect both margins and revenue growth in this critical area.

The potential collaboration between Meta and Google on TPUs signals a major reshuffling of power in the AI industry. Alphabet and Broadcom are strengthening their positions, while Nvidia faces a real risk of losing market share in a strategic data center segment that accounts for nearly ninety percent of its revenue. In the coming months, the pace of TPU adoption by large companies and further development of Gemini 3 could determine how the balance of power in the AI hardware market evolves.