What AI Hardware Looks Like in 2024
Explaining the CPUs, GPUs, and NPUs in Intel®’s AI PCs
Sponsored by Intel®
So there I was — an AI person without an AI laptop.
And no, not that kind of AI person; my ability to run an all-day AI workshop with barely a bio break has led a few of you to ask whether I am indeed a member of your species. (It turns out I’m an espresso-based lifeform.)
This blog post, however, is sponsored by Intel, not espresso because… there I was, an AI person without an AI laptop. My regular laptop’s battery, tuckered out by AI-enhanced video editing, had just died on my flight to Miami. In the ensuing hours of mourning (my battery and my choice to fly with an airline that doesn’t believe in electric sockets) I figured it was time for an upgrade.
The only thing you’ve ever seen me “unbox” so far is software, so I thought it might be fun to do a real and proper unboxing. My first! So I asked Intel for one ultra-powerful AI machine for home and one sleek AI laptop for travel, and they sent me these beauties (the Dell Latitude 5450 and the Dell Latitude 9450 2-in-1 respectively).
And lo! The unboxing:
What makes a PC an AI PC, exactly? How are these devices different from regular laptops? And why haven’t I owned one before?
Making AI a practical reality takes more than just math and data. It takes hardware.
Many people who talk about AI make it sound magical or mathematical or (the worst!) mathemagical. As a practical cat who spent a decade working on enterprise AI, I’m allergic to the inherent impracticality of that messaging. Making AI a practical reality takes more than just math and data. It takes hardware. It’s too easy to take AI hardware for granted, so let’s take a moment to show it some love.
Computing power has come a long way since the early days of bulky, room-sized machines. At the heart of every computer lies its processing unit — the brain responsible for executing commands and performing calculations. Over time, we’ve seen the rise of different types of processors designed for specific tasks: CPUs, GPUs, and NPUs. For those who are new to these terms, I’ve put an explainer in the footnotes for you.* Spoiler: N is for “neural”.
What’s an AI PC?
So, what’s an AI PC? It’s built with AI in mind, so it will include all three types of processors, which work together to provide the best performance across various tasks:
- CPUs handle general-purpose tasks, ensuring that the system runs smoothly.
- GPUs take over tasks that require large-scale parallel processing, such as rendering visuals.
- NPUs handle sustained AI tasks at lower power for greater efficiency
What sets Intel®’s AI PCs apart is that their dedication to AI runs both ways: their processors are built with AI and built to accelerate AI, elevating productivity, creativity, gaming, entertainment, security, and more… running AI workloads efficiently and locally.
One of the things I’m excited to do with them is take some of my time series data modeling off the cloud. But don’t get the impression that you have to be an AI person to benefit from your very own NPUs.
Don’t get the impression that you have to be an AI person to benefit from your very own NPUs.
The Dell Latitude 5450 and the Dell 9450 2-in-1 machines equipped with Intel® Core™ Ultra Processors are optimized to work beautifully with AI-laden apps like Adobe Photoshop, which means maybe I needn’t fear sudden device death next time I’m illustrating something from the back of a cab. Even simple AI tasks that you take for granted might be draining your battery to smithereens on a non-AI machine, for example blurring your background in Zoom (or setting it right to the galactic one, my favorite choice for a confusing first meeting).
A modern and mature approach to AI-powered work.
Not only are AI PCs powered by Intel® Core™ Ultra Processors and Intel® vPro built for performance on the heaviest tasks I’ve thrown at them, but they also come with beefed-up security. Enterprise AI is a nonstarter if your approach doesn’t pass muster with IT (which is why letting your people gorge themselves on a free-for-all of AI apps from every-which-where on the wild internet is a nono), but a modern and mature approach to AI-powered work is what these machines were built for.
Designed to be enterprise-ready, AI PCs powered by Intel® Core™ Ultra Processors with the Intel® vPro platform come with hardware-based features that support remote management, advanced security, and optimized performance.
Honestly, now that I have these new machines, I’m not sure why I didn’t think to upgrade sooner.
Thank you, Intel!
P.S. I still haven’t settled on what to name them. My ex-laptop is called Metropolis (do not tell me to name this one Hastings) but I think these deserve something a little more noble, especially the Dell Latitude 9450 2-in-1, which is super sleek and flips to a touchscreen. Shall I name it Chimera? Or something even better? Put your suggestions in the comments!
Footnotes
*What is a CPU?
The Central Processing Unit (CPU) is the traditional general-purpose “brain” of a computer, responsible for executing commands and performing calculations in series. Multi-core CPUs allow multiple tasks to be handled simultaneously, but what if you want to perform thousands of operations in parallel? You’re not going to do it by stringing together a few thousand CPUs…
*What is a GPU?
Unlike the CPU, the Graphics Processing Unit (GPU) is designed for parallel processing and handling heavy AI workloads, along with anything that involves crunching large datasets. Everything from neuroscience to 3D printing to cryptocurrency mining to meteorology stands to benefit from parallel operations, but where it all started was video game graphics. Yes, it’s technically true that AI owes its rise to the video gaming habits of the millennial teenager. Turns out we weren’t wasting time collecting power-ups and shooting monsters… we were building the future. Hah!
*What is an NPU?
As AI grew in popularity, it became clear that the NPU was essential, taking work away from both the CPU and GPU, which run hotter and consume more energy. Move over, video games! The Neural Processing Unit (NPU) was developed to be highly efficient at handling the unique demands of AI workloads, beyond what a GPU is capable of. This makes them much faster and more energy-efficient for AI-specific tasks, like image recognition, natural language processing, and real-time decision-making in smart devices.
