The NPU in your phone keeps improving—why isn’t that making AI better?

Ars Technica — AllThursday, December 4, 2025 at 12:00:42 PM
The NPU in your phone keeps improving—why isn’t that making AI better?
  • The neural processing unit (NPU) technology in smartphones continues to advance, yet this progress has not translated into significant improvements in artificial intelligence (AI) capabilities. The challenges of miniaturizing AI for mobile devices remain complex, as the demand for processing power and efficiency grows.
  • This development is crucial for smartphone manufacturers and software developers, as enhanced NPU capabilities are expected to drive more sophisticated AI applications. However, the lack of corresponding AI advancements raises questions about the effectiveness of current approaches in leveraging this hardware.
  • The ongoing evolution of AI technology is influenced by various factors, including the reliance on optical technologies to improve performance and the exploration of decentralized solutions to overcome limitations imposed by traditional infrastructures. These themes highlight the multifaceted nature of AI development and the need for innovative strategies to unlock its full potential.
— via World Pulse Now AI Editorial System

Was this article worth reading? Share it

Recommended apps based on your readingExplore all apps
Continue Readings
Revolutionizing Supply Chain Planning with AI: Machine Learning and Agentic Frameworks
NeutralArtificial Intelligence
The integration of artificial intelligence (AI) into supply chain planning is being revolutionized through the application of machine learning and agentic frameworks. This advancement aims to enhance efficiency and adaptability in supply chain operations, addressing the complexities of modern logistics and production demands.
Can AI Look at Your Retina and Diagnose Alzheimer’s? Eric Topol Hopes So
PositiveArtificial Intelligence
Eric Topol, author of Super Agers, advocates for the potential of artificial intelligence (AI) in diagnosing Alzheimer's disease through retinal analysis, suggesting that this technology could revolutionize medical diagnostics.
Engineer proves that Kohler’s smart toilet cameras aren’t very private
NegativeArtificial Intelligence
An engineer has demonstrated that the cameras in Kohler's smart toilets are not as private as advertised, raising concerns about user privacy. The revelation highlights potential vulnerabilities in the technology designed for personal use in bathrooms.
What Happens When Your Coworkers Are AI Agents
NeutralArtificial Intelligence
In a recent episode of Uncanny Valley, writer Evan Ratliff discusses his experience creating a startup composed entirely of AI employees, revealing insights into the implications of an agentic future. This exploration raises questions about the role of AI in the workplace and its potential impact on traditional employment structures.
TrueNorth Raises $3M to Build Domain-Specific AI for Finance
PositiveArtificial Intelligence
TrueNorth has successfully raised $3 million to develop domain-specific artificial intelligence tailored for the finance sector. This funding aims to enhance the capabilities of AI applications within financial services, potentially transforming how financial data is processed and analyzed.
Anthropic CEO weighs in on AI bubble talk and risk-taking among competitors
NeutralArtificial Intelligence
Anthropic's CEO discussed the current state of the AI industry, addressing concerns about an economic bubble and the risk-taking behavior of competitors, which he described as 'YOLO-ing' in their spending strategies. This commentary reflects the heightened competition and investment in AI technologies.
An AI system for real-time fault detection in rail transport
PositiveArtificial Intelligence
A new research published in the International Journal of Information and Communication Technology reveals an AI-based automated system designed for real-time fault detection in rail transport. This system utilizes deep learning to identify issues in critical infrastructure components such as tracks, bridges, tunnels, and signaling equipment, potentially enhancing safety and reliability in railway operations.
Researchers find what makes AI chatbots politically persuasive
NeutralArtificial Intelligence
A comprehensive study on political persuasion has revealed that AI chatbots exhibit only a weak effect in influencing political opinions. Researchers conducted extensive analysis to understand the mechanisms behind AI's persuasive capabilities, ultimately concluding that these systems have limited impact in political discourse.