During the Cloud Next, Google unveiled its new TPU AI accelerator chip, Ironwood.

Ironwood, Google's seventh-generation TPU, is scheduled to be available to Google Cloud customers sometime in 2025. According to the company, it will come in two configurations: a 256-chip cluster and a 9,216-chip cluster.
“Ironwood is our most powerful, capable, and energy-efficient TPU yet,” Google Cloud VP Amin Vahdatin a blog post. It’s reportedly designed to power AI models that rely on reasoning and inference.
See also: Ocelot: Amazon Web Services' (AWS) quantum computing chip
It's worth noting that the Ironwood chip is being revealed at a time when competition in the AI accelerator space is fierce. Nvidia may have the lead, but other companies, such as Amazon and Microsoft, are also pushing their own solutions. Amazon has its Trainium, Inferentia, and Graviton, available through AWS, and Microsoft has its Cobalt 100 AI chip.
According to Google, Ironwood can deliver 4,614 TFLOPs of computing power at its peak. Each chip has 192 GB of dedicated RAM with a bandwidth approaching 7.4 Tbps.
See also: Gemini gets smarter with Google Photos integration
The company also said that the Ironwood TPU AI accelerator chip features an enhanced specialized core, the SparseCore. This is designed to process the types of data typically found in “advanced ranking” and “recommendation” workloads (e.g., an algorithm that suggests clothes you might like). The TPU architecture is designed to minimize data traffic and on-chip latency, resulting in power savings.

Google plans to integrate Ironwood into AI Hypercomputer , a modular computing cluster on Google Cloud.
See also: Google Docs updates Help me create with more languages
Google's announcement of its new TPU AI accelerator chip Ironwood is a significant development in the field of artificial intelligence and cloud computing. Ironwood is Google's seventh generation of TPUs and is designed specifically to accelerate AI applications, particularly in inference tasks.
This development underscores Google’s commitment to developing hardware solutions in-house to support its AI applications, reducing its reliance on third-party vendors. The launch of Ironwood is expected to strengthen Google’s capabilities in the field of artificial intelligence, offering Google Cloud customers improved performance and efficiency for their AI applications .
source: techcrunch.com
