Google Shows Off Its Tensor Chip, the Powerhouse Behind Pixel 6

Google Tensor SoC
Google

Do you ever feel like something’s holding you back? Phone manufacturers have been stuck using Qualcomm Snapdragon chips for years, and while these processors may be powerful, they aren’t one-size-fits-all. That’s why the new Pixel 6 and Pixel 6 Pro run on Google’s first in-house processor, the Tensor chip. But why is Tensor so much better for Google?

According to Google, it all comes down to AI and machine learning. No, Google isn’t trying to turn your phone into the Terminator—it wants to use AI to boost camera quality, intelligently reduce battery usage, and enable lightning-fast Live Translate features for real-world conversations or livestreams.

How Does the Tensor Chip Work?

The Tensor chip labeled to show its architecture layout.
Google

As explained by Google, modern smartphone apps and software are “heterogeneous,” meaning that they need to pull resources from several parts of a processor to run at an acceptable speed. Advanced machine learning algorithms are especially demanding, and as a result, Snapdragon chips can’t unlock the full potential of Google’s advanced AI systems.

The 5nm, eight-core Tensor chip is a bit different. It can accommodate more heterogeneous processes than competing mobile processors thanks to a unique design—other mobile chips have just one ARM Cortex-X1 core, but the Tensor packs two ARM Cortex-X1 cores running at 2.8GHz. These powerful cores can work together to juggle demanding tasks, delivering AI performance that just isn’t possible on Snapdragon.

Tensor also features two medium-sized 2.25GHz Cortex A76 cores, plus four small 1.8GHz A55 cores. These extra cores should help Pixel phones juggle multiple AI processes without slowing down app or system performance. And like many Android SoCs, the Tensor packs a Mali-G78 graphics core, likely to improve game performance and reduce computational photography rendering times.

But that’s not all. Google also slipped a Titan M2 security core in the Tensor chip to help protect customers from attacks. The Titan M2 core helps Pixel phones withstand voltage glitching, laser fault injection, and other advanced attacks that very dorky criminals may use to try and steal your private data.

One last thing that’s worth noting—Google can optimize Tensor in small ways that we may not know about. The company quietly mentioned that parts of its HDRNet feature are embedded in Tensor, and that this inclusion allows for 4K 60FPS recording on the Pixel 6.

How “Fast” Is Tensor?

Pixel 6 Pro in Cloudy White
Google

It should come as no surprise that Tensor is the fastest chip ever used in a Pixel phone. Previous Google handsets used very modest processors, such as the Qualcomm Snapdragon 765G that we got with the Pixel 5.

But Google’s claim that Tensor is “80% faster” than the Snapdragon 765G is a little wonky. Early benchmarks show that, in a traditional sense, Tensor should offer performance comparable to the Snapdragon 888 or Exynos 2100. (Especially when gaming, given that the Tensor has a pretty standard GPU setup.)

It seems like Google’s “80% faster” metric has more to do with AI processes than everyday tasks or gaming. So is the Tensor fast or not?

The answer is simple—Tensor offers flagship performance for everyday tasks and speeds up processes that you didn’t realize were slow. That means Google Assistant, photography, voice to text, and other Google-centric features should feel a lot snappier on a Pixel 6 than on previous Google phones.

What Does Tensor Make Possible?

The Pixel 6 running Live Translate on a road sign.
Google

I’ll be the first to say that AI is confusing and unattractive. Most people just want their phone to work, and big-brain software is rarely user-friendly. But the AI processes enabled by Tensor benefit Pixel users in several simple, straightforward ways. Google isn’t over-complicating its phones; it’s making phones more useful and usable to regular people.

Computational photography is an obvious example here. Google relies on AI to process photos on the Pixel, and more powerful AI will result in better photos. It also opens the door to simple photo editing tools, such as the new Magic Erase and Face Unblur features on Pixel 6.

But Tensor also brings major improvements to speech recognition, leading to more accurate speech-to-text features and a faster Live Translate tool. For those who hate typing on a phone, the Pixel 6 sets the gold standard for talk-to-text. And if you need to speak with someone in another language, Live Translate lets you do it with typing or waiting for crap to load.

Tensor brings several other improvements to Pixel users, and it gives Google the ability to introduce new AI-powered features in the future.


Clearly, Google is setting itself apart from all other phone brands by focusing on these practical AI features. The new Pixel 6 costs just $600, but considering its speed, camera quality, and unique AI features, it can compete with $1,000 flagship phones.

We’re not sure how Google will use Tensor in the future. Perhaps it will try to put the chips in its Chromebooks, or even sell custom Tensor chips to other phone manufacturers. Either way, Qualcomm should be worried. Snapdragon may be on the way out.

Andrew Heinzman Andrew Heinzman
Andrew is a writer for Review Geek and its sister site, How-To Geek. Like a jack-of-all-trades, he handles the writing and image editing for a mess of tech news articles, daily deals, product reviews, and complicated explainers. Read Full Bio »

The above article may contain affiliate links, which help support Review Geek.