Katherine Blunt

Inference, it's the computing process necessary to answer a query on behalf of a user.

For example, you ask ChatGPT a question, the process that's going on behind the scenes there before a response is generated, that's inference.

And inference is becoming increasingly important as we move into an era of agents.

which are capable of doing things on behalf of a user without explicit instruction from one.

So you could instruct the agent to clean up your inbox, then it will go decide how to do that by itself.

This particular chip, it's very efficient, and it's supposed to be tailored for this process that's becoming...

increasingly important.

Catherine says the development sharpens Google's rivalry with AI chip giant NVIDIA.

These are custom-made in-house hardware that's been developing for many years, largely for its own use.

Google needs a lot of these for the AI models and other things that it's developing.

There's questions about the degree to which Google will eventually externalize access to its chips.

But we've seen the company strike deals to allow anthropic access to its chips, meta access to its chips.

And we'll see whether the company takes any sort of other strategic directions in making its chips available.

One of the biggest announcements today out of the conference is that Google is rolling out this AI mode capability, and it's meant to create an experience for users akin to what they might find if they use ChatGPT or one of ChatGPT's competitors or Google competitors.

One of the biggest announcements today out of the conference is that Google is rolling out this AI mode capability, and it's meant to create an experience for users akin to what they might find if they use ChatGPT or one of ChatGPT's competitors or Google competitors.

And then other announcements that were made today were previews of what the company hopes to achieve in AI, for example, with a universal AI assistant who could be very helpful in managing different areas of your life and the administrative tasks we all have to do. There was a lot as it related to images and videos and the creative capabilities that can come with more advancements in AI.

And then other announcements that were made today were previews of what the company hopes to achieve in AI, for example, with a universal AI assistant who could be very helpful in managing different areas of your life and the administrative tasks we all have to do. There was a lot as it related to images and videos and the creative capabilities that can come with more advancements in AI.

And this is a really interesting moment for Alphabet and for Google specifically. Google has struggled to figure out how exactly to embrace AI, given that it does pose at least some sort of threat to the traditional search model that's really been its largest driver of revenue.

And this is a really interesting moment for Alphabet and for Google specifically. Google has struggled to figure out how exactly to embrace AI, given that it does pose at least some sort of threat to the traditional search model that's really been its largest driver of revenue.

And so this is a big announcement today because it's Google finally conceding that the future of search is really going to be very AI driven and it's going to have to adapt.