Qualcomm and Hugging Face Expand Strategic Alliance to Deploy Open Source AI from Energy Efficient Hardware to Global Cloud Networks
Qualcomm introduced to expand on its strategic alliance with Hugging Face to enable deployment of open source artificial intelligence from hardware device up to cloud networks. As per official corporate release published under NASDAQ QCOM ticker a partnership sought to blend Qualcomm high performance computing silicon with the Hugging Face global developer platform. The joint initiative would focus to create all in one system where software applications could arbitrage computing expenses, latency in processing, and security parameters dynamically.
The first core component of this business collaboration entails moving on premises and developer 0 workloads of Hugging Face to Qualcomm Dragonfly data center platforms. The present storage / cloud inference infrastructure of Hugging Face will be aligned to the energy efficient data center platforms. This move would facilitate 16,000,000 developers of Hugging Face network to move from model experimentation to production with agility. The integration of the products family of Qualcomm Dragonfly forms an enterprise ready pipeline for application scaling on energy efficient hardware.
The 2 nd initiative is to quicken the open models adaption for edge devices. Over the services provided by Hugging Face, there are over 3,000,000 open models. With this partnership, an autonomous software agent will oversee the set up, the configuration and optimization, and reduction of these models unto Qualcomm chips without manual integration.
This autonomous set up will enhance on the easiness of designers creating advanced AI processing on phones, Computer systems, automobiles, wearables, and industrial equipment.
To compliment the work, Open Hugging face will enjoy Premium PRO service to developers utilizing Qualcomm compatible hardware to provide better collaboration tools.
The 3 rd pillar of the partnership builds a distributed computing infrastructure for agentic AI. This platform allows intelligent agents to perform models locally on device or offload computation to remote cloud servers taking into account the moment by moment cost of processing, local power limit, and the need to keep the data local. Developers will be able to directly access Modular's building blocks and utilities created for the MoE infrastructure within the Hugging Face hub.
The leadership on both sides of the deal suggest the joint venture is just a natural evolution of the industry moving toward local processing.
Cristiano Amon, President and CEO of Qualcomm Incorporated told The Register that, local high performance, low power computing combined with a huge software community will accelerate the arrival of a new wave of applications that work seamlessly across device and in the cloud.
Clment Delangue, Co Founder and CEO of Hugging Face agreed by saying, local models are becoming even more critical as they bring a private and cost effective alternative for companies to proprietary centralized APIs.
More and more the world is utilizing open and local models because they cost less than the big API’s and are private from the get go. We are working with Qualcomm Technologies, to enable our 16,000,000 developers and counting use Modular software and tools to deploy open models everywhere – from that device in your hand to that warehouse full of rackmount servers in the data center…with agents that operate across the compute continuum.
This technical collaboration is meant to facilitate the cooperation between hardware manufacturers and software developers by allowing open source models that run properly on Qualcomm Snapdragon, Dragonwing and Dragonfly processing units.
