# Google TPU 73 Hacker News commenters recommended Google TPU and 15 criticized it, November 2025 to November 2026. It was recommended in 11 of 13 months. It meets the Hacker Approved badge rule. **Verdict:** Google's custom AI chips, rented through Google Cloud, are recommended for cheap, energy-efficient inference and large-scale clusters. Common caveat: hard to program and behind Nvidia at top performance. ## Why people recommend it - Cheaper, efficient inference at scale (10 comments): "even if google TPUs aren't competitive in a pure performance sense they are surely competitive from a pricing perspective" ([source](https://news.ycombinator.com/item?id=48841724)) "TPUs are the economically-conscious inference choice if you've already separated your training/inference workflows." ([source](https://news.ycombinator.com/item?id=47878686)) "Google is crushing them on inference. By TPUv9, they could be 4x more energy efficient and cheaper overall" ([source](https://news.ycombinator.com/item?id=46994308)) - Owns hardware and full stack (4 comments): "Google’s TPU is going to be a massive advantage that’s almost impossible for Amazon or Microsoft to replace" ([source](https://news.ycombinator.com/item?id=47918247)) "Owning your hardware and your entire stack is huge, especially these days with so much demand." ([source](https://news.ycombinator.com/item?id=47864273)) "They get to make pickaxes that are made exactly for the mine they are working." ([source](https://news.ycombinator.com/item?id=46070422)) - Strong interconnects for large clusters (4 comments): "until they can match that interconnect scale, they simply don't compete in this weight class" ([source](https://news.ycombinator.com/item?id=46070648)) "GOOGLE has more experience with datacenter clustered TPUs and tensor acceleration" ([source](https://news.ycombinator.com/item?id=46062534)) "TPUs don't have enough memory either, but they have really great interconnects" ([source](https://news.ycombinator.com/item?id=46994147)) - Proven in Google's own workloads (4 comments): "Was a solid bet on energy efficient dense matrix multiplication and has stood the test of time." ([source](https://news.ycombinator.com/item?id=48674946)) "they've had some initial success with TPUs which could be a major differentiator in the future" ([source](https://news.ycombinator.com/item?id=48590479)) "Google is already using their own TPUs at scale for both training and inference." ([source](https://news.ycombinator.com/item?id=45935915)) ## What people warn about - Hard to program; immature compiler tooling (5 comments): "Nobody wanted to mess with an immature crashing compiler and very steep performance cliffs" ([source](https://news.ycombinator.com/item?id=47933598)) "last we tried, so much of our kernels needs rework that it’s not worth the effort" ([source](https://news.ycombinator.com/item?id=47899514)) "TPUs can be more efficient, but are quite difficult to program for efficiently" ([source](https://news.ycombinator.com/item?id=46630314)) - Trails Nvidia at top performance (5 comments): "If 'this is faster than a CPU for massively parallel AI training' is the best you can say about it, that's not very impressive." ([source](https://news.ycombinator.com/item?id=45966445)) "Google had nigh on a decade to boost TPU production, and they're still somehow behind the curve." ([source](https://news.ycombinator.com/item?id=47538244)) "It can't compete with NVidia's best offerings for sure but it gives them a path to independence." ([source](https://news.ycombinator.com/item?id=50312032)) ## Alternatives hackers recommend - [Taalas HC1](https://hackerapproved.com/p/taalas-chip.md): 59 recommend - [Taalas hardwired LLM chip](https://hackerapproved.com/p/taalas.md): 45 recommend - [Cerebras wafer-scale chip](https://hackerapproved.com/p/cerebras-wafer-scale-chip.md): 29 recommend - [Cerebras CS-4](https://hackerapproved.com/p/cerebras-cs4.md): 12 recommend - [Groq LPU](https://hackerapproved.com/p/groq-lpu.md): 9 recommend ## Where to get it - [Visit the official site](https://cloud.google.com/tpu) ## Cite as Hacker Approved, "Google TPU", https://hackerapproved.com/p/google-tpu/ (counts from 1,795,640 Hacker News comments, November 2025 to November 2026).