Google unveils Ironwood, a new AI chip to challenge Nvidia

Google (GOOG) has launched its new Ironwood chip — the seventh generation of its Tensor Processing Unit (TPU). This move aims to challenge Nvidia’s strong position in AI hardware. The chip is designed for both training and inference, meaning it can help build and run AI models. After months of testing in Google Cloud, Ironwood will roll out widely in the coming weeks.

Google becomes a serious contender in AI silicon

Ironwood is built to handle all kinds of AI work — from training large models to running live chatbots. Each Ironwood pod can connect over 9,000 TPUs. This setup cuts data delays and boosts performance for huge models. Google says Ironwood is more than four times faster than its previous TPU version, marking a big jump in speed and efficiency.

AI startup Anthropic plans to use up to a million Ironwood chips to run its Claude models. This shows strong trust in Google’s custom hardware. The chip enters a crowded AI hardware race among major cloud players — Google, Microsoft (MSFT), Amazon (AMZN), and Meta (META) — all building the systems that power generative AI.

AI hardware becomes the new battleground

Ironwood’s launch shows a shift in the cloud market. Nvidia (NVDA) still leads in GPUs, but Google’s TPUs aim to offer faster and cheaper options for specific AI tasks. Google Cloud earned $15.15 billion in Q3 revenue, up 34% from last year. The company also says it has signed more billion-dollar cloud deals this year than in the past two years combined.

Google raised its capital spending forecast to $93 billion to meet rising demand for AI infrastructure, including TPUs and GPUs. CEO Sundar Pichai called AI hardware “a key driver of our growth.” With this, Google is positioning itself as a full-stack AI company — building both the chips and the cloud systems that use them.

Investors eye the new AI race

For investors, Ironwood highlights how the AI hardware market is splitting into multiple players. Nvidia remains the leader, but Google’s in-house chips could slowly chip away at its lead, especially in large cloud workloads. This trend could be slightly negative for Nvidia in the long run and positive for Google, which can now control more of its hardware and earn higher margins.

Amazon and Microsoft may also expand their own chip programs, which could pressure GPU prices and Nvidia’s market share. If Google succeeds in scaling Ironwood, it could gain more control over pricing and stand out in AI cloud services. The key question is how fast developers will adopt it — and whether they’ll move away from Nvidia’s popular CUDA software platform.

In short, Ironwood marks a new stage in the AI compute race. Dependence on one chipmaker may be ending. Google’s push into AI chips could reshape how cloud companies compete for leadership in model training. For long-term AI investors, it’s a development worth close attention.

Google (GOOG) and Nvidia (NVDA) have a Disruption Score of 5 and are part of the Disruption Aristocrats.