🧔♂️ A friendly human may check it before it goes live. More news here
Arm CEO: Shifting AI tasks from cloud to devices can cut energy use
Arm Holdings CEO Rene Haas said shifting some AI tasks from the cloud to local devices could help reduce energy consumption.
Speaking on CNBC, Haas noted that the growth of multi-gigawatt data centers is unsustainable, and suggested that running AI inference on devices such as phones, computers, or smart glasses could ease power demands.
He added that, while AI training is likely to remain in the cloud, hybrid models—where inference happens locally—could become standard.
Arm supplies chip technology used by companies including Microsoft, Amazon, and Meta, with Nvidia holding a significant stake in the firm.
Arm and Meta announced an expanded partnership to improve AI efficiency across software and data center infrastructure.
Haas said this partnership focuses mainly on data centers and software, but also highlighted that Meta’s Ray-Ban Wayfarer smart glasses run some AI features locally on Arm-powered chips.
🔗 Source: CNBC
🧠 Food for thought
Implications, context, and why it matters.
Arm’s AI revenue model unclear despite on-device momentum
Rene Haas, Arm’s CEO, pushes AI work onto devices. The on-device revenue path is murky, and Arm has not broken out income from NPUs versus partner-built accelerators in sources reviewed 1.
– Cortex-A320 and Ethos-U85 run models with over one billion parameters 2, though only over 20 partners had licensed Ethos NPUs by April 2024 1. That lean uptake hints revenue favors Arm’s CPU platform, which it calls the most ubiquitous compute base 3.
– Work with Meta and others centers on software tuning and data center gear, which signals bets across cloud plus edge instead of reliance on device inference 3.
Recent Arm developments
Stay updated on the go with our mobile app.
Get latest insights with smoother, more personalized experience through TIA mobile app.