Recommended Free Tools
iTechGuides is reader-supported. When you buy through links on our site, we may earn an affiliate commission. As an Amazon Associate I earn from qualifying purchases. Learn more
Google’s custom AI server chip is Ironwood, its seventh-generation Tensor Processing Unit (TPU). The first Ironwood product, TPU7x, is designed for data-center-scale AI training and inference and is available to customers through Google Cloud—not as a chip or server card for consumers to buy and install.
What is Google’s AI chip called?
Ironwood is the name of Google’s seventh-generation TPU family; TPU7x is its first release. Google introduced Ironwood in 2025 as its first TPU designed specifically for inference—the phase when a trained model generates answers or other outputs. Google Cloud’s current product documentation says TPU7x supports both large-scale training and inference, so it is not limited to serving models.
A TPU is an application-specific integrated circuit (ASIC): a processor designed for particular computing workloads. Ironwood is best understood as a complete AI-computing system, not just a chip. Google describes it as combining accelerators, high-bandwidth memory, inter-chip networking, liquid cooling, and software within its AI Hypercomputer architecture.
How can customers use Ironwood?
Customers access TPU7x capacity through Google Cloud, using Google Kubernetes Engine (GKE) or Compute Engine. Google marked TPU7x generally available on March 31, 2026. Compute Engine support for creating and managing TPU VMs and slices was marked generally available on June 1, 2026, according to Google Cloud release notes.
#1 Best Overall
- NVIDIA Volta GV100 Architecture — 4,608 CUDA Cores, 640 1st-Gen Tensor Cores delivering 14 TFLOPS FP32 and 112 TFLOPS deep learning performance for AI training, inference, HPC, and scientific computing workloads
- 32GB HBM2 ECC Memory — 900 GB/s Bandwidth — High-bandwidth memory on a 4096-bit bus with ECC error correction provides the memory capacity and throughput required for the largest AI models, simulations, and datasets
- PCIe 3.0 x16 Interface — 250W TDP — Standard PCIe Gen3 connectivity with passive cooling designed for enterprise rack server deployment in HPE ProLiant, Dell PowerEdge, and Supermicro platforms with adequate chassis airflow
- NVLink — Scale to 96GB Unified Memory — Connect two V100 GPUs via NVLink at 300 GB/s bi-directional bandwidth to scale GPU memory from 32GB to 96GB for larger AI training and HPC workloads
- Multi-Precision Computing — Supports FP64 (7 TFLOPS), FP32 (14 TFLOPS), FP16 (112 TFLOPS) and INT8 precision modes for flexible deployment across training, inference, and scientific simulation workloads
Framework support is a key migration consideration: Google documents JAX and PyTorch for TPU7x, but not TensorFlow. A team considering a move should check whether its existing models, operations, and tooling work with the supported frameworks before estimating the engineering effort.
Because Google deploys Ironwood as cloud infrastructure at data-center scale, it is not a retail accelerator card or a consumer CPU. A full Ironwood pod can scale to 9,216 chips, according to Google’s TPU7x documentation.
Rank #2
- High-Performance AI Processing: The MX3 is designed to handle the most demanding AI computer vision workloads, delivering exceptional performance and efficiency.
- Flexible Integration: The MX3 can be easily integrated into your existing systems via its M.2 M-key form factor and support for Linux operating systems.
- Energy Efficient: The MX3 is designed to provide high performance while minimizing power consumption.
- Comprehensive Software Development Kit (SDK): The MX3 is supported by a comprehensive SDK that simplifies development and deployment.
- Hardware compatability: The MX3 is compatible with the PCI-SIG M.2 M-key 2280 Specification. It can be used with the Raspberry Pi 5 with a M-key 2280 HAT.
What are TPU7x’s specifications?
The figures below are Google Cloud’s documented hardware specifications for TPU7x, not independent workload benchmarks.
| TPU7x specification | Google-documented value |
|---|---|
| Chips per pod | 9,216 |
| Peak compute per chip, BF16 | 2,307 TFLOPs |
| Peak compute per chip, FP8 | 4,614 TFLOPs |
| High-bandwidth memory (HBM) per chip | 192 GiB |
| HBM bandwidth per chip | 7,380 GB/s |
| Bidirectional inter-chip interconnect (ICI) bandwidth per chip | 1,200 GB/s |
| Documented four-chip VM configuration | 224 vCPUs and 960 GB RAM |
Ironwood’s interconnect is designed to let chips exchange data directly. Google says its custom networking supports remote direct memory access (RDMA), allowing chips to communicate without routing that data through the host CPU. Google’s engineering description specifies eight HBM3E stacks per chip and rounds peak HBM bandwidth to 7.4 TB/s; that is consistent with the TPU7x documentation’s 7,380 GB/s figure.
Rank #3
- ✅Powered by 26 Tera-Operations Per Second (TOPS) Hailo-8 AI Processor. 2.5W typical power consumption
- ✅Scalable, enabling simultaneous processing of multi-streams & multi-models
- ✅Enabling real-time, low latency and high-efficiency AI inferencing on the edge devices
- ✅Supports TensorFlow, TensorFlow Lite, ONNX, Keras, Pytorch frameworks
- ✅Supports Linux and Windows. Supports the temperature range of -40°C to 85°C
Peak compute and bandwidth figures describe hardware capability; they do not by themselves predict throughput, latency, cost, or energy use for a particular model. Those outcomes depend on the workload, configuration, and software.
What does Google claim about performance and efficiency?
In a November 2025 product update, Google said Ironwood delivers ten times the peak performance of TPU v5p and more than four times the per-chip performance of TPU v6e (Trillium) for training and inference. These are Google’s comparisons, not a promise that every application will run that much faster. A useful comparison requires measuring the intended model and configuration against a defined target, such as throughput or response latency.
Rank #4
- 48GB AI graphics accelerator
Google also reported a 3.7× improvement in carbon compute intensity (CCI) for Ironwood compared with TPU v5p, based on fleet measurements from January 2026. Google’s calculation combines life-cycle emissions with utilized BF16 FLOPs. Its stated scope includes cooling electricity, but excludes peripheral rack, shelf, and network equipment and auxiliary compute and storage. Operational calculations use one month of observed TPU fleet machine-power data and Google’s 2024 average fleetwide carbon intensity. Google says results vary by workload location and that the analysis is not a full quantification of Google AI emissions. Treat the figure as a Google-reported fleet comparison, not an independent life-cycle assessment or a guarantee for a particular customer workload.
How does Ironwood compare with Google’s newer TPU 8 chips?
In April 2026, Google announced two eighth-generation systems with different intended roles. The announcement says both would be available to Cloud customers “soon”; it does not establish that either is generally available. By contrast, Google’s release notes document TPU7x general availability in March 2026.
Best Value
- Professional AI & Creator Workstation: AMD Radeon AI PRO R9700 GPU with 32GB GDDR6 is engineered for AI development, professional content creation, and compute-intensive workloads.
- Massive 32GB Memory Capacity: 32GB of GDDR6 memory on a 256-bit bus provides ample bandwidth for large AI models, 8K video editing, and complex 3D rendering.
- Advanced RDNA 4 with AI Accelerators: 64 Compute Units with 3rd Gen Ray Tracing and dedicated 2nd Gen AI Accelerators for groundbreaking AI performance and visual computing.
- Professional Blower Cooling: Efficient single blower design exhausts heat directly out of the chassis, ideal for multi-GPU workstation and server configurations.
- Enterprise-Grade Thermal Solution: Vapor chamber heatsink with industrial Honeywell PTM7950 thermal interface material ensures reliable cooling under sustained professional loads.
| System | Google’s announced focus and figures | Availability established by the cited announcement |
|---|---|---|
| TPU 8t | Training-focused; Google says a superpod has 9,600 chips and provides 121 exaflops | Google said it would be available to Cloud customers “soon”; general availability is not established |
| TPU 8i | Inference and reinforcement learning; Google highlights 384 MB on-chip SRAM, 288 GB HBM, and 19.2 Tb/s interconnect bandwidth | Google said it would be available to Cloud customers “soon”; general availability is not established |
These are announcement specifications, not independent performance measurements. The two systems are not simply one all-purpose replacement: Google positions 8t for training and 8i for inference and reinforcement learning.
Independent reader supportYour contribution helps us test, update, and keep practical guides available for everyone.How should teams assess Ironwood for their workloads?
There is no universal winner among AI accelerators. Google Cloud also offers NVIDIA-based systems, and the evidence does not support a blanket claim that TPU is faster, cheaper, or better for every workload. Compare options using the actual model, software stack, service capacity, and deployment requirements.
Quick Recap
- Workload: Decide whether the priority is training, inference, or both, and identify the relevant batch sizes and response-time targets.
- Software fit: Verify framework and model support. For TPU7x, Google documents JAX and PyTorch support and says TensorFlow is not supported.
- Measured performance: Test the intended model and configuration rather than relying on peak FLOPs or vendor comparisons alone.
- Memory and scale: Check memory capacity and bandwidth, interconnect needs, and whether the workload benefits from scaling across a pod.
- Practical constraints: Compare cloud cost, capacity availability, software porting effort, and any energy or emissions figures using clearly stated methods.
Product prices and availability are accurate as of the date/time indicated and are subject to change. Any price and availability information displayed on Amazon at the time of purchase will apply.
