News highlights:<\/p>\n
Strategic partnership aligns Arm\u2019s leadership in power-efficient compute with Meta\u2019s innovation in infrastructure, AI products and open technologies, to enable richer, more accessible AI experiences for billions of people worldwide<\/p>\n
Meta\u2019s foundational AI software technologies \u2013 including PyTorch \u2013 now optimized for Arm, including PyTorch\u2019s Executorch runtime using Arm KleidiAI to maximize performance-per-watt for Meta and the global open source community<\/p>\n
Arm and Meta have announced a strategic partnership to scale AI efficiency across every layer of compute \u2013 spanning AI software and data center infrastructure \u2013 to enable richer user experiences to billions of people worldwide. From milliwatt-scale devices powering on-device intelligence to megawatt-scale systems training the world\u2019s most advanced AI models, the collaboration will enable AI across multiple types of compute, workload, and experiences that power Meta\u2019s global platforms.<\/p>\n
The multi-year partnership builds on the ongoing hardware and software co-design efforts between the two companies, combining Arm\u2019s leadership in power-efficient AI compute with Meta\u2019s innovation in AI-driven products, infrastructure, and open technologies to achieve significant performance and efficiency gains.<\/p>\n
\u201cFrom the experiences on our platforms to the devices we build, AI is transforming how people connect and create. Partnering with Arm enables us to efficiently scale that innovation to the more than 3 billion people who use Meta\u2019s apps and technologies.\u201d Santosh Janardhan, Head of Infrastructure, Meta<\/p>\n
\u201cAI\u2019s next era will be defined by delivering efficiency at scale. Partnering with Meta, we\u2019re uniting Arm\u2019s performance-per-watt leadership with Meta\u2019s AI innovation to bring smarter, more efficient intelligence everywhere \u2014 from milliwatts to megawatts.\u201d\u00a0Rene Haas, CEO, Arm\u00a0<\/p>\n
Scaling AI in the Cloud<\/p>\n
Meta\u2019s AI ranking and recommendation systems \u2013 which power discovery and personalization across Meta\u2019s family of apps, including Facebook and Instagram \u2013 will leverage Arm\u2019s Neoverse-based data center platforms to deliver higher performance and lower power consumption compared to x86 systems. Across its infrastructure, Arm Neoverse<\/a> will also allow Meta to achieve performance-per-watt parity \u2013 underscoring the efficiency and scalability of Arm compute at hyperscale.<\/p>\n The companies worked closely to optimize Meta\u2019s AI infrastructure software stack \u2013 from compilers and libraries to major AI frameworks \u2013 for Arm architectures. This includes the joint tuning of open source components, such as Facebook GEneral Matrix Multiplication (FBGEMM) and PyTorch, exploiting Arm\u2019s vector extensions and performance libraries, producing measurable gains in inference efficiency and throughput. These optimizations are being contributed back to the open source community, to broaden their impact across the global AI ecosystem.<\/p>\n Accelerating AI Software from Cloud to Edge<\/p>\n