Google’s Capitalization Approached $4 Trillion Amid Talks With Meta to Sell AI Chips

• Meta is negotiating a multi-billion dollar purchase of Google’s TPUs, increasing pressure on Nvidia.
• The potential agreement could make TPUs an alternative to Nvidia’s dominant GPUs.
• Alphabet has moved closer to a market capitalization of $4 trillion amid expectations of a deal.

Technology giant Meta is discussing a multi-billion dollar deal with Google to acquire TPU AI chips, The Information writes. The company plans to use them in its data centers and is considering leasing the processors from Google Cloud as early as 2026.

The deal will be an important step in Meta’s strategy to reduce dependence on Nvidia, the authors of the material write. GPUs of the mentioned company remain the industry standard for training and outputting large language models, but other industry players are trying to find alternatives.

According to analysts, the interest in TPUs is logical, as Google previously entered into an agreement to supply up to 1 million chips to Anthropic. This confirmed the company’s ability to compete in the AI infrastructure market and attracted additional institutional inquiries, the media outlet noted.

The background of the negotiations also played on the growth of Alphabet shares (Google’s parent company), which for the first time approached the capitalization of $4 trillion, according to TradingView. Over the past 24 hours, the corporation’s securities rose by 1.6%.

In contrast, Nvidia shares slipped about 4% at the premarket as investors assess the risks of slowing demand for GPUs and increasing diversification of vendors. However, the securities partially regained some of their lost value later in the day.

Nvidia also did not ignore reports of a possible sale of Meta chips. The company welcomed Google’s “success” in TPU development, but noted that their AI capabilities are “a generation ahead”.

Analysts note that the potential deal would reinforce Google’s status as a potential secondary gas pedal vendor for AI model output.

Bloomberg Intelligence estimates that Meta’s capital expenditures could exceed $100 billion in 2026. Of that, $40-50 billion will come from infrastructure for AI.

The decision to choose TPU will depend on whether Google’s chips can prove their efficiency, experts believe. In addition, the factor of performance on the scale of Meta data centers may play an important role in the choice.

TPUs have been developed by the company for more than a decade and are specialized integrated circuits designed specifically for AI tasks. Their demand is growing amid concerns in the industry about excessive dependence on Nvidia and limited availability of GPUs.

Today, TPUs are not only used for Google’s internal models, including Gemini, but are also being made available to developers via Google Cloud. The company emphasizes that it will continue to support its Nvidia chips and GPUs.

Recall, we wrote that Google stated the need for a thousand-fold growth of AI power by 2030.