What Are TPU Chips – and Why They’re Putting Pressure on Nvidia?

The escalating rivalry in the Artificial Intelligence (AI) chip market has intensified following reports that 元 is in advanced talks to spend billions on Google’s (NSDQ: GOOG) AI chips, a move that challenges the long-held market dominance of Nvidia (NSDQ:NVDA).

A potential deal with Meta, one of the world’s largest spenders on AI infrastructure, would signal major momentum for Google’s Tensor Processing Units (TPUs) and validate the search giant’s case as a serious rival to Nvidia in AI technology.

The Emergence of Google’s TPU as a Contender

For as long as Nvidia has dominated the AI chip space, customers have sought competition. The formidable alternative has emerged from Google: the TPU.

• TPU vs. GPU: Unlike Nvidia’s general-purpose Graphics Processing Units (GPUs) – originally designed to render video game images and prized for their adaptability across various workloads – TPUs are highly specialized. TPUs are a type of specialized product known as Application-Specific Integrated Circuits (ASICs), built for a discrete purpose.
  
  
• Specialized Efficiency: TPUs were built specifically for matrix multiplication, the main operation involved in training neural networks for large language models (LLMs). This specialization allows them to be less power-hungry and more efficient when running these specific operations compared to the more flexible but costlier-to-operate Nvidia GPUs.
  
  
• The Ironwood Chip: Google’s latest version, the Ironwood Tensor Processing Unit, was introduced in April 2025 and is designed for running AI inference workloads (using the models, rather than training them).

Google’s strategy has benefited from its in-house AI development; as the company and its DeepMind unit developed cutting-edge models like Gemini, they could customize the chips to benefit their AI teams.

Market and Corporate Reactions

The news of a potential Meta-Google partnership has shifted investor sentiment, putting pressure on Nvidia’s stock while boosting Alphabet’s valuation.

• The Meta Deal: Meta is in discussions to use the TPUs in its data centers starting in 2027 and may rent chips from Google’s cloud division as early as next year.
  
  
• Stock Impact: Nvidia’s shares were down about 4% in premarket trading on the news. The stock has faced a “tricky November,” dropping 10% over the past month (as of 17 November) amid worries about high AI valuations and the mounting threat from Google. In contrast, Alphabet Inc. is benefiting from the market validation, with its shares soaring and the company on track to hit a $4 trillion market valuation for the first time.
  
  
• Anthropic Validation: The momentum for TPUs was already growing after Google agreed in October to supply up to 1 million chips to Anthropic PBC, a deal an analyst called a “really powerful validation” for TPUs.

Analyst Consensus and Nvidia’s Outlook

Despite the competitive headwinds, Wall Street remains largely confident in Nvidia’s future.

• Strong Analyst Support: The majority of analysts covering the stock remain bullish. According to data on Bloomberg, 74 of the 80 analysts covering Nvidia rate it a Buy, with a consensus price target of $253.
  
  
• Continued Demand: Mizuho analyst Vijay Rakesh noted that demand for Nvidia’s GPUs still looks “rock solid”. The next-generation Blackwell chip is reported to be sold out, and CEO Jensen Huang is targeting half-a-trillion dollars in revenue in 2026.
  
  
• The Long View: Analysts believe that no company, including Google, is currently looking to replace Nvidia GPUs entirely. Nvidia’s market position is cemented by the versatility of its GPU platform, which is “better suited to handle a wider range of workloads” if a customer’s AI algorithm or model changes. Even Anthropic, a major TPU customer, announced a “big deal” with Nvidia just weeks after its Google TPU tie-up.

Ultimately, the best hope for Google’s TPUs may be to establish themselves as a crucial part of the product range required to power the exponential growth of AI, rather than a full replacement for Nvidia’s dominance.

Trade AI related and other US stocks on Phillip Nova 2.0 now! 立即点击开户