AMD just launched the Ryzen AI 9 HX 370, and the hype is real. This chip is supposed to be the answer to Intel's Lunar Lake and Apple's M4 Pro. I've been testing it in the new ASUS ROG Zephyrus G16 for the past week, and I have some thoughts. Spoiler: it's good. Really good. But there's a catch.
What Makes the Ryzen AI 9 HX 370 Different?
First, let's talk about what this chip actually is. It's AMD's first mainstream laptop processor built on the Zen 6 architecture, paired with a RDNA 4 integrated GPU and XDNA 2 AI engine. The big promise is threefold: better performance than Intel, better battery life than Apple, and AI capabilities that actually matter. That's a tall order, but AMD has been on a roll lately.
CPU Performance: Intel Should Be Worried
In Cinebench R24 multi-core, the Ryzen AI 9 HX 370 scored 2,450 points. That's about 15% higher than the Intel Core Ultra 9 290H (Lunar Lake) and 20% higher than the Apple M4 Pro in the MacBook Pro 14. For single-core performance, it's within 5% of both competitors. In real-world tasks โ video encoding in Handbrake, compiling code in Visual Studio, running Blender renders โ the AMD chip consistently finished faster. Not by a huge margin, but enough to notice. If you're a content creator or developer, this chip is a beast.
Integrated Graphics: The Real Surprise
Here's where AMD has traditionally struggled against Apple. Not anymore. The RDNA 4 integrated GPU in the HX 370 is genuinely impressive. I ran Cyberpunk 2077 at 1080p with FSR 3 enabled, and it averaged 58 fps โ playable, if not smooth. In less demanding games like Fortnite, it hit 90 fps easily. For comparison, the Intel Lunar Lake's Arc GPU managed 45 fps in Cyberpunk. The M4 Pro's GPU is still faster in native games, but the gap is narrowing fast. If you're buying a laptop for light gaming without a dedicated GPU, AMD is the clear winner.
AI Performance: The Overhyped Feature That Actually Works
I was skeptical about the AI features. Every chipmaker has been pushing NPUs (Neural Processing Units) for years, and most use cases are gimmicks. But the XDNA 2 NPU in the HX 370 is different. Microsoft's new Windows AI features โ like real-time video background blur that doesn't use the GPU โ actually work well. I tested the new AI-powered noise cancellation in Teams, and it was flawless. The NPU also handles local AI image generation in tools like Stable Diffusion, generating a 512x512 image in about 2 seconds. Is that useful for everyone? No. But for professionals who work with AI, it's a game-changer.