Google Cloud TPU 8t (Trillium 2 / Training-Dedicated)
Overview
Google TPU 8t (codenamed Trillium 2) is the latest training-dedicated TPU, announced on 2026-04-22 (forming a split architecture with the simultaneously announced TPU 8i inference-dedicated TPU). It features 216GB HBM (12% more than TPU v7 Ironwood), 6,528 GB/s bandwidth, and an integrated Arm Axion CPU (Google's custom 64-core Arm processor).
TPU 8t is the core training chip for Google Gemini 3 / Gemini 4 frontier models, with key improvements over TPU v7 Ironwood in training paradigm optimization (MoE training, long-context training, RLHF post-training).
Core Specifications
| Item | Specification |
|---|
| Architecture | TPU 8t (Trillium 2) |
| Type | Training-dedicated (distinct from 8i inference-dedicated) |
| BF16 Compute (dense) | ~3,500 TFLOPS (estimated, ~50% higher than Ironwood's 2,307 TFLOPS) |
| FP8 Compute (dense) | ~7,000 TFLOPS |
| HBM Capacity | 216 GB |
| HBM Bandwidth | 6,528 GB/s |
| ICI Interconnect | 1,400 GB/s (bidirectional) |
| DCN Bandwidth | 200 Gbps (estimated) |
| Integrated CPU | Arm Axion (Google custom, 64-core) |
| Pod Size | 9,216 chips (estimated) |
| Topology | 3D Torus |
| Announcement | 2026-04-22 |
📌 8t naming: TPU 8th-gen + t = training. 8t and 8i are same-generation; 8t is for training only.
TPU 8t vs TPU v7 Ironwood (Training Comparison)
| Metric | TPU v7 Ironwood | TPU 8t | Improvement |
|---|
| Type | Training + Inference | Training-dedicated | Type split |
| BF16 Compute | 2,307 TFLOPS | ~3,500 TFLOPS (estimated) | 1.5× |
| FP8 Compute | 4,614 TFLOPS | ~7,000 TFLOPS | 1.5× |
| HBM Capacity | 192 GB | 216 GB | 1.13× |
| HBM Bandwidth | 7,380 GB/s | 6,528 GB/s | Slight decrease |
| ICI Interconnect | 1,200 GB/s | 1,400 GB/s | 1.17× |
| Integrated CPU | None | Arm Axion 64-core | New |
| Announcement | 2025-11 | 2026-04-22 | — |
💡 TPU 8t bandwidth slightly decreased (7,380 → 6,528 GB/s) but compute increased 50%, indicating Google traded some bandwidth for higher compute on 8t (better suited for compute-intensive training phases: dense FFN, attention computation).
TPU 8t Training Paradigm Optimization
| Optimization | Details |
|---|
| MoE Training | Native Expert Parallel support (DeepSeek / Mixtral style) |
| Long-context Training | Optimized for 1M+ token context training |
| RLHF / Post-training | Native optimization for Online RL (DPO / PPO / GRPO) |
| Multimodal Training | Vision-language joint training (ViT + LLM synchronized) |
| AXIOM | Arm Axion CPU co-processing (data preprocessing / weight initialization) |
Arm Axion CPU Integration
| Item | Specification |
|---|
| Architecture | Arm Neoverse V2 (64-core) |
| TDP | ~100 W |
| Role | Host CPU + Data loading + Preprocessing + Inference scheduling |
| Significance | Google's custom Arm CPU enters TPU nodes for the first time |
Axion integration = TPU nodes evolving towards "super nodes": TPU 8t is no longer a pure accelerator, but a TPU + Axion CPU co-processing system, competing with NVIDIA Vera CPU.
Recommended Deployment Configurations
| Scenario | Recommended Configuration |
|---|
| Gemini 3 Training | TPU 8t pod 9,216 chips (single pod can train frontier models) |
| Llama 4 Training | TPU 8t pod (hundred-billion-scale models) |
| Multimodal Training | TPU 8t + Vision Transformer |
| Scientific Computing | TPU 8t + JAX 0.5+ |
| RLHF Post-training | TPU 8t (natively optimized) |
Software Ecosystem
- JAX 0.5+: Google's primary training framework
- PyTorch/XLA 2.5+: PyTorch compatibility
- TensorFlow 2.17+: Legacy framework
- Paxml / Orbax: Google internal LLM training stack
- MaxText: Google reference implementation
- vLLM 0.8+ (experimental): Inference support
Use Cases
- ✅ Frontier model training (Gemini 3/4, Anthropic, external customers)
- ✅ MoE large-model training (native support)
- ✅ Long-context training (1M+ token)
- ✅ Multimodal training (ViT + LLM)
- ❌ Inference scenarios (use TPU 8i instead of 8t)
- ❌ Non-Google Cloud deployments
| Item | Details |
|---|
| Vendor | Google Cloud |
| First Announced | 2026-04-22 (Google Cloud Next 2026) |
| Product Page | https://cloud.google.com/tpu |
| Cloud Deployment | Google Cloud only |
| Codename | Trillium 2 |