Latest generation chips from AMD, Intel, and Qualcomm push NPU capabilities beyond 40 TOPS, with real-world performance improvements reaching 27% over previous architectures
The artificial intelligence computing race has reached a new milestone in 2024, with the latest processor generations from major chipmakers delivering unprecedented neural processing unit (NPU) performance that’s reshaping the landscape of AI-enabled personal computers.
Leading the charge is AMD’s Ryzen AI 300 series, which delivers an impressive 50 TOPS (trillion operations per second) of AI processing power.
This represents a significant leap forward in on-device AI capabilities, enabling everything from real-time language model inference to advanced video processing without relying on cloud services.
Intel’s response comes in the form of its Lunar Lake architecture, featuring the new NPU 4 that achieves up to 48 TOPS of peak AI performance.
Intel claims up to 48 TOPS of peak AI performance from its fourth-generation neural processing unit, marking what the company describes as “a giant leap in enhancing power and efficiency” compared to the previous NPU 3 generation.
Not to be outdone, Qualcomm’s Snapdragon X Elite processors pack a powerful 45 TOPS Hexagon NPU. Featuring the integrated Qualcomm Hexagon NPU, which delivers an unprecedented 45 trillion operations per second, the Snapdragon X Elite is the fastest NPU for laptops globally, according to Qualcomm executives.
Real-World Performance Gains Tell the Story
While TOPS specifications provide a useful benchmark, real-world testing reveals the practical impact of these NPU improvements.
When testing the Ryzen AI 9 HX 375 model in the LM Studio development environment, the Strix Point APU achieved up to 27 percent higher performance compared to the Intel Core Ultra 7 258V, according to recent independent testing.
This 27% performance advantage demonstrates that the NPU improvements aren’t just theoretical—they translate to measurable benefits in applications like large language model (LLM) inference, image generation, and other AI workloads that consumers and professionals rely on daily.
The competitive landscape has intensified significantly compared to previous processor generations. Intel’s existing Core Ultra “Meteor Lake” processors have NPUs that support 10 TOPS, while AMD’s 8040-series “Hawk Point” goes up to 16 TOPS. The jump to 45-50 TOPS represents a 3x to 5x improvement in just one generation cycle.
Beyond Just NPUs: Total AI Performance Matters
Intel is pushing a broader narrative about AI performance that extends beyond just NPU capabilities. Intel also revealed that several architectural improvements made to the CPU and GPU will push the chip’s total AI performance to 120 TOPS, at least on the highest-specced model.
The implications extend far beyond raw performance numbers. These NPU improvements are enabling new categories of applications, from real-time video enhancement and background blur to on-device language translation and advanced voice recognition—all without the latency, privacy concerns, or connectivity requirements of cloud-based AI services.
Market Impact and Industry Response
The NPU performance race reflects broader industry trends toward edge AI computing, driven by privacy concerns, latency requirements, and the desire to reduce dependence on cloud services.
Microsoft’s Copilot+ PC initiative, which requires a minimum of 40 TOPS of NPU performance, has helped establish clear performance targets that chipmakers are now exceeding.
Early benchmark results suggest that both Lunar Lake and the Snapdragon X Elite are pretty evenly matched; Intel’s chip does better with AI on wall power, but Qualcomm is the superior AI chip on the go, highlighting how different architectures excel in different usage scenarios.
The competition is spurring rapid innovation cycles, with each manufacturer pushing to deliver not just higher TOPS numbers but also improved power efficiency, better software optimization, and enhanced compatibility with popular AI frameworks and applications.
As these processors roll out in laptops and desktop systems throughout 2024, consumers can expect a new generation of AI-powered features that were previously impossible or impractical on personal computers.
The NPU performance improvements of 2024 mark a clear inflection point in making sophisticated AI capabilities accessible to everyday users, setting the stage for even more dramatic advances in future processor generations.
The NPU market, estimated at $15 billion in 2025, is projected to witness a Compound Annual Growth Rate (CAGR) of 25% from 2025 to 2033, reaching approximately $75 billion by 2033.
Performance figures and specifications based on manufacturer data and independent testing by Tom’s Hardware, AnandTech, Igor’s Lab, and other industry sources as of late 2024.
