For decades, the division of labor was clear: the CPU was the generalist for everything, the GPU (graphics card) the specialist for images and games. But modern software demands a third discipline: artificial intelligence.

Classic CPUs are extremely inefficient when it comes to the massive matrix calculations of modern AI. Graphics cards can handle it, but they consume so much power that the battery is drained in no time.

This is where the NPU (Neural Processing Unit) comes in. It is physically integrated on the chip and specialized in processing AI workloads extremely quickly and, above all, energy-efficiently. A laptop without an NPU in 2025 is like a car without air conditioning – it will still run, but comfort and resale value will suffer massively.

Change is not optional; it is being dictated by the software giants.

• Microsoft has introduced a two-tier system with its “Copilot+ PC” label. Many of the new Windows features (local AI search, real-time translation, Cocreator) are not even enabled on devices without a powerful NPU (at least 40 TOPS).
• Apple is fully committed to the Neural Engine of its M-series chips with “Apple Intelligence.” Anyone still using an old Intel Mac is effectively excluded from the entire next generation of the operating system.

The message is clear: Without an NPU, you are a second-class guest in the operating system of the future.

Why would you buy this if you don’t even want to generate AI images? The answer is battery life.

In a traditional PC, the main processor has to power up for background tasks like noise reduction in video calls or facial recognition. In a modern AI PC, the NPU handles these “always-on” tasks while the system is running. The result is a massive leap in efficiency. Current chips from Qualcomm (Snapdragon X), Intel (Core Ultra Series 2), and AMD (Ryzen AI) enable battery life that was previously unthinkable because the power-hungry CPU is relieved of this burden.

The market has already decided.

A look at the current offerings shows: The “classic PC” is disappearing from store shelves.

• Premium segment: There are practically no high-end devices without an NPU anymore.
• Mid-range: The technology is rapidly trickling down to lower segments.

With “Lunar Lake,” Intel completely rebuilt its architecture to compete in the NPU race against Apple and Qualcomm. There’s no going back to the old x86 dominance without AI acceleration.

These devices use the “Snapdragon X Elite/Plus”. They were the first true “Copilot+ PCs” and often offer the longest battery life.

Microsoft Surface Laptop 7:

The reference device. Microsoft has gone all in on the “ARM architecture” here. It’s extremely fast, wakes up instantly (like a smartphone), and easily lasts a full workday.

Dell XPS 13 (9345):

For design lovers. It’s extremely compact and futuristic, but uses the Snapdragon chip for maximum efficiency. (Note: The XPS 13 is also available with an Intel chip – a good example of the confusion in the market!).

Lenovo Yoga Slim 7x:

The price-performance winner. Offers a superb OLED display and full NPU power, often cheaper than the competition.

These devices use the “Core Ultra Series 2” (e.g., Core Ultra 7 258V). Intel has made significant strides here: The RAM is integrated directly onto the chip (like Apple’s), making them almost as efficient as the Snapdragons, but without compatibility issues.

ASUS Zenbook S 14 (UX5406):

One of the flagships for Intel’s new era. Ultra-thin, with a chassis made from a new ceramic-aluminum alloy (“Ceraluminum”), and proof that Intel laptops can also be quiet and cool.

HP OmniBook Ultra Flip 14:

A 2-in-1 convertible (laptop & tablet). HP utilizes Intel AI power for robust security features and a 9-megapixel AI camera.

Samsung Galaxy Book5 Pro 360:

Shows the strength of the Intel ecosystem in conjunction with Android smartphones.

Devices with “Ryzen AI 300” (Strix Point). These chips have extremely powerful NPUs (50 TOPS), but also very powerful integrated graphics cards. Ideal for people who want AI and a bit of gaming.

ASUS Zenbook S 16:

An exciting device – a very large 16-inch screen, yet as thin as a tablet. It demonstrates how efficient the new AMD chips are.

ASUS ROG Zephyrus G16 (2024/25 Refresh):

Here, the NPU is often combined with a powerful NVIDIA graphics card. This is an example for “power users” who want to train AI locally or edit videos.

Apple uses the M4 chip family. They don’t talk about “Copilot+”, but about “Apple Intelligence”.

MacBook Pro (M4 / M4 Pro):

The ultimate choice for creatives. The M4 chip has an incredibly powerful Neural Engine that is deeply integrated into the system.

MacBook Air (M4):

The entry into the world of AI for the masses. Fanless and silent.

Note: We would like to point out that when buying a laptop in a store, you shouldn’t just look at the brand name (e.g., “Dell XPS”), but also at the processor name. A “Dell XPS 13” could have an older, hot-running Intel chip or a newer, cooler Snapdragon chip. The casing looks almost identical, but the internal components are radically different.

Search terms for buyers: “Snapdragon X”, “Intel Core Ultra Series 2” (or “200V”), “Ryzen AI 300”, “Apple M4”.

The title might sound drastic, but it reflects the technological reality. Buying a laptop without an NPU today is comparable to buying a 3G smartphone just before the 4G breakthrough. It still works, but it’s obsolete from day one.

If you’re investing today, there’s no way around an NPU. Not necessarily because you want to chat with AI every day tomorrow—but because without one, your computer won’t be able to optimally handle updates and features in two years.