The AI-Native Era: Building the Future on Arm - Six Five Summit

Arm's Ubiquitous Role in Computing

• 💡 Arm's compute platform is used by 70% of the world's population, spanning from large data centers to small devices like earbuds.
• 🚀 The platform supports all legacy software, including operating systems, applications, and hypervisors, which have run on Arm for decades.

Powering the AI-Native Era

• 🧠 AI workloads are being added on top of existing compute, covering both training in the cloud and inference at the edge.
• 🎯 This shift presents a gigantic opportunity for Arm, as AI will run on its platform across virtually all devices.
• ✨ Arm is focused on optimizing the hardware experience and creating a seamless software environment for developers to run AI workloads.

Innovations with Compute Subsystems (CSS)

• 🛠️ Arm introduced Compute Subsystems (CSS) to deliver pre-integrated IP, accelerating chip development and guaranteeing highest performance levels.
• 📈 CSS helps customers shave significant time off market for chip development, especially with advanced geometries like 7nm, 5nm, and 3nm.
• ✅ This platform approach allows for earlier developer engagement and faster utilization of new features, like those needed for AI workloads such as Gemini Nano.

Evolving Platform Strategy

• 🔑 Arm is transitioning from a product-focused approach to a platform-first company strategy, with distinct platforms for data centers, mobile, and automotive.
• 🌐 This evolution clarifies how Arm's ecosystem elements fit together, making it easier for partners to understand the comprehensive solutions offered.
• ⚡ The goal is to establish Arm as the compute platform for AI, capable of providing solutions from milliwatts to megawatts.

Addressing AI Adoption Challenges

• 🧩 Arm helps overcome AI adoption challenges by providing hardware hooks for acceleration and software libraries like Clyde AI.
• 💡 These libraries abstract hardware complexity, allowing developers to easily run AI without needing to know the specifics of the underlying NPU.
• 🚀 The focus is on inference, which will run ubiquitously on devices (the "student"), rather than training, which is primarily cloud-based (the "teacher").

Vision for 2030

• 🔮 By 2030, Arm envisions remaining the dominant platform for an increasing AI workload, potentially becoming the primary workload.
• 🌍 Success means running the vast majority of software (AI or non-AI) on Arm, providing the best solutions for customers across all device types.
• 💡 The company is focused on the developer community and enabling a wide breadth of devices, adapting to future forms of ambient computing.