K-Culture GlossaryTInfrastructure and chips
Tensor Processing Unit (TPU)
Tensor Processing Unit
Google's in-house AI chip, and the leading challenger to NVIDIA's dominance.
In plain words
Tensor Processing Unit — a chip Google designed itself for its own AI. It has gone through generations since 2016, and it's what powers Gemini's training and services.
What it means is that there's a company that doesn't need NVIDIA. In an industry structure where GPU supply depends on a single company, Google controls its own costs and supply with its own chip. Other Big Tech firms are building their own AI chips for the same reason, and TPU is often cited as the success story in "de-NVIDIA" stories.
GPU vs. NPU vs. TPU: What's the Difference?
Is it a chip Google built for its own use?, Is it a general-purpose processor?, Is it a low-power dedicated chip inside a device?
AI chip articles often mention all three in the same paragraph, making it hard to tell who's actually competing with whom. They're not fighting over the same turf — one dominates the data center, one sits inside your phone or laptop, and one was built by Google to run its own services.
| Aspect | Tensor Processing Unit (TPU) | GPU | Neural Processing Unit (NPU) |
|---|---|---|---|
| Where it lives | Google data centers | Data centers. Cloud servers | Inside your phone or laptop |
| Who makes it | Google (manufactured with Broadcom's help) | NVIDIA, AMD | Device makers like Apple, Qualcomm, Intel |
| Who uses it | Google and Google Cloud customers | Nearly everyone building and selling models | Anyone using a device — usually without realizing it |
| What it's good at | Large-scale training and serving, cheaply | The heaviest training and large-scale serving | Running small models while saving power |
| In the news | "Challenging NVIDIA's dominance" | "GPU shortage race," "NVIDIA earnings" | The reality behind "AI phone" and "AI PC" |
| Can you buy one? | No — you rent it through Google Cloud | Yes, though supply falls short of demand | It's already built into any device you buy |
Rule of thumbIf a model is built in a data center, it's a GPU. If it runs inside your own device, it's an NPU. If it's behind a Google service, it's a TPU.