Google’s in-house AI chip TPU has welcomed a couple of new additions to the 8th generation portfolio, which are the TPU 8tx and TPU 8i Max, that join the existing 8t and 8i variants to form a 4-option lineup for customers to access.
The new selections tackle the hottest problems in enterprise AI, as businesses are running into issues as they put autonomous agents into production: keeping sensitive data secure and handling million-token context windows.
So to start, the TPU 8tx is built for regulated enterprises and governments that run frontier or proprietary models and agents alongside their most sensitive data. As data is being encrypted in storage and transit but has to be unlocked in memory during processing, 8tx can close that gap by building security into the hardware layer, using the same silicon as TPU 8t paired with custom AMD CPU hosts, so data and model weights stay protected and isolated even while in use.
It also uses the low-latency Boardfly network created for real-time reasoning, delivering 2x lower latency than TPU 8t with a near-zero performance penalty. Because it pairs training-grade compute with a fast inference network, teams can fine-tune models on private data, distill large models into smaller ones, and train agents with reinforcement learning on a single system, without moving sensitive data between clusters.
Meanwhile, the 8i Max scales compute and memory to new highs by doubling both of them in addition to HBM bandwidth, compared to the base 8i. It also fits twice the silicon in the existing rack footprint, so data centers get significantly more computing power per square foot and per watt.
Demand for Google TPUs has been rising over the years, as the firm is using them for its own Gemini models and agentic workloads while being offered to customers working in frontier labs, trading firms, retailers, and universities. Google also gave simple recommendations on which TPU to choose based on workload type:
- TPU 8t – For scale-up training
- TPU 8i – For low-latency reasoning
- TPU 8tx – For confidential computing
- TPU 8i Max – For massive context windows and highest compute density
Like our Facebook Page here at NasiLemakTech.com for more news and in-depth reviews! Also, join our Facebook Group for insightful information and memes!
Subscribe to our YouTube channel too!
