Efficiently Enabling AI Workloads on Arm
On-Device Generative AI
The arrival of sophisticated Generative AI models has sparked a new wave of innovation, and offer an opportunity for technology step change.
Arm’s compute platform offers the efficient and highest-performing capabilities, which enable GenAI to run on phones, PCs and datacenters.
AI Inference
With the rise in number of AI applications, there will be an exponential increase in the need for AI inference capabilities. Arm CPUs provide the technology foundations for inference to run on all compute touchpoints, bringing AI into the hands of billions of people around the world.
Arm’s Heterogeneous Solutions Offer a Path to Customization
We know that developers target 70% of AI workloads in third-party applications in the smartphone space on the Arm CPUs, as it’s flexible, ubiquitous, and portable. But there are some workloads that benefit from an heterogenous AI approach, which implements accelerators such as GPUs and NPUs.
Arm’s heterogenous AI encourages customization, choice and flexibility whether you’re integrating with Arm’s GPUs (such as Arm Mali and Arm Immortalis), Arm Ethos NPUs, or partner solutions (such as NVIDIA’s Grace Hopper and Grace Blackwell), so you can seamlessly complement our CPUs to achieve accelerated ML.
Arm technology is crucial to enabling AI to happen everywhere.
“We’ve invested in technologies that enable flexible compute solutions that balance efficiency and performance to match your workloads wherever they are running. Whether it’s features that equip our CPUs and GPUs to run AI workloads, or the creation of the world’s first line of microNPUs, we continue to make strategic investments that enable AI in future devices.”
- Richard Grisenthwaite, EVP Chief Architect, Arm
Enabling the Developer Ecosystem to Deploy at Pace
Software developers are at the heart of AI innovation. Whether they are experimenting to gain insights via audio, visual or text-based data or leveraging generative AI, they need to trust they can deploy their AI-based workloads in the most efficient way, on widely available platforms, with the confidence that code and data will run securely.
Arm is accelerating this transformation, enabling rapid and trusted AI innovation by integrating with major operating systems and AI frameworks, to ensure seamless delivery of performance and efficiency in every use case.
As the foundational architecture for over 280 billion digital devices globally, with the widest choice of hardware platforms at many price points, Arm delivers efficiency without compromising performance and provides a consistent, supported and security-backed foundation for running current and future AI workloads everywhere.
“Software developers are the driving force in delivering the digital solutions of the future, and a large proportion of those solutions will leverage AI. We work across our broad ecosystem of partners, and in thousands of collaborative software projects to provide an easy-to-use software toolbox for AI workloads. We contribute to all the key enabling projects to provide an essential framework for accelerating AI innovation and performance on Arm, enabling AI from endpoint to cloud.”
- Mark Hambleton, SVP Software, Arm