Google vs Nvidia AI showdown: ‘We’re a generation ahead,’ says Nvidia after possible Meta-Google chip deal

Nvidia’s stock dropped on Tuesday after a report by The Information said that Meta is thinking about switching to chips designed by Google. Nvidia’s shares slid by up to 7% before improving slightly and ending the day about 4.3% lower. Alphabet, Google’s parent company, rose 4.2% after jumping more than 6% on Monday.

According to the report, Meta is looking at using Google’s tensor processing units (TPUs) in its data centers starting in 2027. The report also said that Meta might rent TPUs from Google’s cloud service as early as next year.

A Google spokesperson said the company is seeing strong demand for its own TPUs as well as Nvidia’s GPUs, and that it plans to continue supporting both.

Meta looking to step away from its reliance on Nvidia?

Adam Sullivan, the chief executive of data-centre operator Core Scientific, told the Wall Street Journal that the biggest story in the AI world right now is the intense competition between Google and Nvidia. According to him, both companies are trying to lock in as much data-centre capacity as possible.

The report, citing people familiar with the matter, said that Meta is in discussions to use Google-made chips for its AI work which would help Meta reduce its heavy dependence on Nvidia.

It also said that the potential deal could be worth billions, but the talks are still ongoing and may not lead to an agreement. One of the people added that it is still unclear whether Meta would rely on Google’s TPUs to train its AI models or to handle inference. It’s the stage where a trained model produces an answer, which needs less computing power than training.

‘NVIDIA is a generation ahead’

Nvidia responded on Tuesday to the Wall Street Journal report which suggested that Google’s technology could challenge its lead in AI infrastructure. Nvidia said that its technology is still a full generation ahead of the rest of the industry.

In a post on X, Nvidia said it is happy to see Google’s progress in AI and noted that it continues to supply chips to Google. The company also said its platform can run every major AI model and works across all types of computing environments.

Nvidia’s comments came after its shares dropped 3% on Tuesday following reports that Meta, one of its major customers, may partner with Google to use Google’s tensor processing units (TPUs) in its data centers.

In the same post, Nvidia argued that its chips are more powerful and flexible than ASIC chips like Google’s TPUs, which are built for a specific company or task. It highlighted that its newest generation, called Blackwell, offers stronger performance and can be used in more ways.

“NVIDIA offers greater performance, versatility, and fungibility than ASICs,” Nvidia posted.

Nvidia vs Google

According to CNBC, analysts have said that Nvidia controls over 90% of the AI chip market with its graphics processors. But in recent weeks, Google’s in-house chips have drawn more attention as a possible alternative to Nvidia’s powerful but costly Blackwell chips.

Google doesn’t sell its TPUs to other companies. Instead, it uses them for its own projects and lets outside businesses rent them through Google Cloud.

Earlier this month, Google launched Gemini 3, which is a highly praised new AI model trained on TPUs rather than Nvidia GPUs.

A Google spokesperson said the company is seeing fast-growing demand for both its own TPUs and Nvidia’s GPUs, and that it plans to support both, as it has done for years, the report said.

Nvidia CEO Jensen Huang also addressed the growing interest in TPUs during an earnings call earlier this month where he pointed out that Google still buys Nvidia chips and that Gemini can also run on Nvidia hardware.

Huang had added that he has been in touch with Demis Hassabis, the head of Google DeepMind. According to Huang, Hassabis texted him saying the belief that bigger AI models can be built by adding more chips and more data, often called “scaling laws”, still holds true. Nvidia believes this trend will continue to drive demand for its chips and systems.

Google emerging to be too dog in AI race – Here’s why

Google has suddenly become one of the strongest players in the AI race. Nothing is guaranteed in this fast-changing field, but the company’s position looks better than it has in years. Here are the key reasons:

1. A strong launch for Gemini 3

Google released Gemini 3 last week, and the model received very positive reviews. It performs better than the earlier version in coding, design, and analysis, and it beats many rival models in tests. Business Insider found it can even build simple websites and video games, which shows it can be used for more than just coding. This successful launch eased concerns that Google had fallen too far behind or that scaling laws were slowing the company down. Since Gemini 3 came out on November 18, Alphabet’s stock has risen more than 12%.

2. Google’s TPUs are gaining attention

Google has been building its own chips – TPUs, for more than ten years. These chips were used to train the Gemini models, which serves as a strong showcase for their power. Companies can rent these TPUs through Google Cloud, and Google has stepped up efforts to bring in more customers. This could become a serious challenge to Nvidia over time.

The reported multibillion-dollar deal that could place Google’s chips inside Meta’s data centres is also gaining attention.

3. A favourable outcome in its antitrust case

In September, Google received penalties in a long-running antitrust case targeting its search business, but the impact was far smaller than expected. The court said Google could continue paying partners like Apple to stay the default search engine, though not on an exclusive basis. It also required Google to share some search data with competitors. For a moment, the case risked breaking up parts of Google’s business, including the Chrome browser, a key part of its search advertising system. Even though the ruling called Google a monopoly, the company avoided major damage.