Last week, one of Link Oregon’s founding members—Oregon State University—celebrated the groundbreaking of the Jen-Hsun Huang and Lori Mills Huang Collaborative Innovation Complex, which will be home to one of the world’s fastest university supercomputers capable of training very large artificial intelligence (AI) models and advancing research in other computationally-intensive domains such as materials science, robotics, and earth, ocean, and atmospheric sciences. At Link Oregon, we are excited to see this milestone development at one of our member higher education institutions and applaud the Huang family’s philanthropic support as well as public/private sector investments and emerging collaborations that are actively reshaping Oregon’s leading role in semiconductor innovation and STEM education nationally. In particular, as the research & education network (REN) for Oregon, we are interested in what AI adoption means for the future of the network.
Link Oregon is a consortium of the State of Oregon through its Enterprise Information Services (EIS) unit and the state’s four largest research universities: Oregon State University, Oregon Health & Science University, Portland State University, and the University of Oregon. AI computing research as well as adoption of AI for innovative applications has been steadily on the rise at Oregon’s largest public research universities. From the University of Oregon’s research focus on applications of AI on neural circuits and cognition to OHSU’s use of AI in early cancer detection and Type 1 diabetes management to research at Portland State University to develop reconfigurable neural networks capable of neurons on synapses on demand, each of our founding universities is actively advancing innovative use cases for AI and advancing research in the domain. Furthermore, Governor Kotek’s executive order late last year establishing a new AI advisory council to develop a plan for ethical, transparent, and inclusive use of AI signals growing intent in understanding the potential societal impacts of AI and interest in potential inclusion of this technology in government operations. Link Oregon Board vice chair Terrence Woods, who serves as the CIO for the State of Oregon, leads this Council.
With innovation in GPUs paving the way for more complex computations in high-performance computing (HPC) and AI research and blazingly fast training of large language models, both the volume and speed of data transmission exchanged over networks between on-premises and cloud data centers will see exponential growth. This trend isn’t going to be limited to HPC clusters in university research labs. Sophisticated AI capabilities are now moving to client devices/PCs, which will drive an uptick in AI-related traffic generated by all end users. As Generative AI moves out of beta and becomes more embedded into every application, every user—not just AI/HPC research scientists at Oregon’s public universities—will be impacted by AND contribute to the massive surge of data traveling over networks. This will demand “future-ready” broadband connectivity that’s designed to support extreme scalability, very low latency, security, and resilience in coming years.
Fiber optic broadband networks, such as the one we operate at Link Oregon, are built to function as data superhighways, with unconstrained bandwidth and the ability to support extremely high-volume data transmissions without signal loss over vast distances. Oregon’s higher ed member institutions engaged in computationally intensive research domains such as HPC, AI, Robotics, and Quantum Science may be operating within their “bandwidth envelope” today, but we can expect significant bandwidth usage upticks in coming years as AI gets incorporated into more research initiatives, online classroom learning, and in cloud data centers.
Link Oregon will soon be conducting trials of 400 Gbps connectivity in anticipation of this demand surge. Increasingly we’re hearing from our higher ed member universities that they need to demonstrate access to highly scalable bandwidth as part of overall infrastructure readiness to secure federal funding of large-scale data-intensive research initiatives. Our technical team at Link Oregon facilitates regular discussions among the research computing leads at our founding member universities to stay ahead of such evolving trends.
While fiber optic networks are critical to avoid bottlenecks caused by the torrent of data that will be created and used by AI systems in the near future, we can also expect AI to play a significant role in improving the operational efficiency of these networks in coming years. The Link Oregon fiber optic network already supports cybersecurity defense capabilities (namely, DDoS mitigation) that rely on machine learning to automatically detect attacks and deploy countermeasures. Autonomous, self-healing networks have long been the ‘Holy Grail’ of networking, and in particular, the ability to be predictive, not reactive, to events that require mitigation. AI can help get us there, as well as transform every facet of our business operations – from field support and network performance diagnostics and analysis to demand and capacity planning.
We are bullish about the many AI-powered innovations and research initiatives being advanced by our higher ed member institutions, the work being initiated by Governor Kotek’s AI Advisory Council to study policy implications and potential use cases for AI in government, and the broadband service innovations that we will continue to develop and offer to our members to advance Oregon’s role within the national/regional tech and AI ecosystem.
Author: Kevin Bohan is a Senior Network Architect and Network Engineer at Link Oregon.