ProgrammingJavascriptCareerProductivityGadgetsASP.NETSecurity

The incoming storm of A.I. powered laptops is here

Written by
Published on
The incoming storm of A.I. powered laptops is here

Microsoft recently held an A.I. press event right before the annual Microsoft Build kicks off into gear and they unveiled their latest innovations in A.I. powered laptops which they are calling Copilot + PC. And for the most part, it was an entertaining 60 minutes. We even got to see the latest Surface Pro and Surface Laptop models that are soon to be released in the following month.

And I personally am a big Surface fan and have been for some years now. I started with the uber portable Surface 3 almost 10 years ago and worked my way up to the Surface Pro 6 as the most recent Surface device. I even dabbled with the Surface Laptop Go 2 more recently to test if it could handle my development workflow. And now, we have A.I. powered Surface devices. How technology advances.

The standout features of course were less to do with the latest Snapdragon X Plus/Elite chipsets, which sound impressive, but more to do with how Windows plans on using these A.I. chips in their operating system. Though the chips themselves do bring some big advantages that I'll dive into down below.

Snapdragon X Plus/Elite

Qualcomm's latest ARM based chipsets, the Snapdragon X Plus and Snapdragon X Elite are in the newest category of SOC (system on a chip) that bring a third component to the usual CPU and GPU pairing. These laptops are equipped with Neural Processing Units (NPUs), specialized hardware designed to accelerate AI tasks, enhancing machine learning capabilities, and improving overall efficiency.

The NPU will be able to handle many of the complex mathematical calculations/operations needed in order to execute various on-device A.I. features and from the publicly made benchmarks so far, it definitely sounds impressive.

A few notable features include:

- 10/12 Core CPU
- 3.8 TFLOPS GPU
- 45 TOPS NPU
- Multiday battery

And if TOPS is new in your vocabulary, that is 45 trillion operations per second. Definitely an impressive feat, numbers wise, regardless of the use case.

Outside of just A.I. related keywords and stats though, one feature that has people exited is the promise of multiday (at least all-day) battery life. And while still relatively new and hard to actually determine right now, estimates are that these new line of SOC's can deliver up to 22 hours of computing.

This is not the first time that I've seen 20+ hours of battery life promised by a manufacturer however, and from past experience those 20 hours, were more like 8-9 hours during a real-world work day. But I'm hoping that we can at least start to push to the 14-15 hour mark.

Copilot + PC

The chips themselves are great, but they can't really do much without an environment to play around in. And this is where Microsoft steps in. Copilot + PC is a new kind of laptop in which Microsoft Copilot sits at the forefront powering your day to day work.

And if that "salesly" description didn't make too much sense to you, it's because I'm personally still trying to wrap my head around it. But for the most part, laptops that are a part of this new line, feature the following:

- Powerful NPU based CPU's (Snapdragon X Elite)
- All day battery life (or multiday)
- A minimum of 16GB of RAM
- A minimum of 256GB of storage
- On device A.I. functionality and features
- Dedicated Copilot button

So it's still a Windows laptop for the most part, but it's a better one. Which makes us wonder why didn't we just make everything better in the past.

But let's dive in to some of the A.I. features, because this is more or less what you're going to be using those 45 trillion operations for.

Recall

Recall essentially allows you to search (using natural human language) for anything on your PC. And by anything, I mean anything. If you saw an article on 'butterflies' 2 weeks ago, you can search for 'butterflies' and have the operating system bring up the browser page of what you were looking at.

If you wrote some notes down a month ago and have lost track of the file, then you can bring it up in Recall if you remember just a few of the words. Overall it sounds very interesting.

Initially, it sounded like a feature that I'd never personally use. Particularly after hearing just how this is handled, because essentially, once you turn this feature on, Windows will begin to take screenshots periodically throughout your work day in order to build an index that it can later search through as needed.

As expected, comment sections were flooded with privacy concerns. But when you hear just how this is achieved, maybe there is slightly less to worry about. And more than likely you're probably giving away more information to 3rd parties anytime that you buy a pen online.

Recall works entirely on the device, meaning that no images are ever pushed to the cloud in any way. And all searches are done through the NPU directly, so that you're not bogging down your machine with such an expensive query. You can also control just how long you want the history to date back to, and you can delete your entire history whenever you choose. And Microsoft has stated that none of this data is used to train any model at any point in time.

And that's really the most interesting aspect of this whole A.I. laptop thing for me, because if my device is fully in charge of compute using the power that I'm supplying to it (and paying for), then all of a sudden there's a little more freedom to build in isolation without having to rely on 3rd party security or added fees.

Cocreator

Cocreator is your digital drawing assistant, in a sense. It isn't quite like Dall-E or any of the other image generation models where you give it a prompt and then get a random result. With cocreator, you give the model a prompt, but then you have to initiate the drawing yourself. The diffusion model will then get to work on rendering what it thinks you are trying to draw.

And once again, most interestingly, the entire thing happens on device. And from the demonstrations showed at the press event, the rendering happens pretty fast.

Again, the use case seems kind of limited. From a web development standpoint, I don't typically rely on drawings for web assets. I need cold hard SVG's and smooth icons for much of my work.

And as a hobby artist at times, I don't usually incorporate any elements outside of my paint brush in order to create, as that would defeat the purpose.

But maybe there's a use case somewhere in the future that will make more sense.

Live captions

This is not the first time that we've seen on-device live captions or translations as some of the latest smartphones have baked that feature into their devices.

But it is the first time we've seen it running locally on a PC. And that is worthy of applause, particularly as more and more people from around the world begin to comingle in this new remote work economy that we find ourselves in.

I personally take on many meetings during the course of a month, and I can definitely see myself using live captions as often times people are on spotty connections making communications an all too difficult task.

The one question left to be answered in all this though, is just how much compute would you need to live-transcribe an hour long meeting.

Voice assistant

There wasn't too much detail given on this specific feature, but the demo that was showed definitely left me asking more questions.

In the demo, we see a Minecraft player fire up Copilot and seamlessly ask it questions about how to play the game. Copilot was able to scan the page and, in detail, tell the player how to continue with crafting items.

Just recently, we saw something similar during the OpenAI press event in which people were able to launch a native ChatGPT  application on macOS and then continued to have it answer questions about it was seeing.

And seeing how Microsoft is leveraging the GPT4o model for various things, it makes sense that we'll see something similar. Now whether this is running on-device or not, is a different story and we'll have to wait and see for future use-cases. But it definitely starts to steer more in the direction of A.I. assistants hovering around us 24/7 answering our questions.

New laptops

Because of the new requirements for Copilot + PC based laptops (more RAM and storage), and the power efficiency of the new Snapdragon chips, we got a slew of pretty powerful and lightweight laptops that are soon to be released.

We saw new devices from Microsoft, with their latest gen Surface Pro and Surface Laptop models. And we saw similar devices from Acer, ASUS, Dell, HP, Lenovo and Samsung.  Lenovo unveiled 2 models, including  a very slim and portable Yoga Slim 7X model, which I personally gravitated towards for various reasons, price being at the top of the list.

Seemingly every laptop came equipped with OLED displays and 3K resolution out of the box and came in weighing under 3lbs. Pricing varied across manufacturer, though surprisingly Microsoft came in with possibly the most affordable configuration at just under $1000.

It being the first gen of these Copilot + PC devices though, we can expect that the hype will settle down just a bit in the near future when tech reviewers start to take them through their paces.

Will the battery lives not be up to par, or will the A.I. translations and image generations leave us disappointed. Hopefully we'll get more answers soon, because if the hype is even remotely real, then I might have just found my next programming laptop.

Walter Guevara is a software engineer, startup founder and currently teaches programming for a coding bootcamp. He is currently building things that don't yet exist.

Comments

No messages posted yet

Developer Poll

Q:

Stay up to date

Sign up for my FREE newsletter. Get informed of the latest happenings in the programming world.

Add a comment

Keep me up to date on the latest programming news
Add Comment

Stay up to date

Get informed of the latest happenings in the programming world.

No thanks