Wednesday, November 18, 2026
AI 인프라 · 뉴스 & 분석
홈 › 데이터센터 › 리포트
데이터센터 · 리포트

Cal Poly has launched an AI data center on its San Luis Obispo campus, and the San Luis Obispo Tribune has taken a look inside the facility.

Campus-based AI facilities show universities building their own compute capacity, a small but growing source of on-premises demand for power and cooling.
업계 전문지Slicast · 2026년 11월 17일 18:00 UTC · 미국 · 출처: San Luis Obispo Tribune
중요도 45

Cal Poly has launched a new, one-room AI data center designed to serve current and future generations of students and faculty who want to develop and train artificial intelligence models.

Housed in the Frank E. Pilling Computer Science Building, Cal Poly's $3 million "AI Factory" contains four large-scale artificial intelligence systems powered by chipmaker Nvidia. The factory was officially unveiled in October during the university's AI Day event.

The servers are powerful enough to store and process the large volumes of data needed for Cal Poly students and faculty to experiment with AI in an academic setting and to train in-house AI models.

Some of those projects are already underway. According to Cal Poly director of tech and training Ryan Matteson, who works in the Noyce School of Applied Computing, the servers currently store data used to power Digital Democracy, a Cal Poly creation operated in partnership with the nonprofit news outlet CalMatters. Digital Democracy tracks the legislative activity and financial backings of officials elected to statewide office in California. Other projects powered by the system include analysis of scientific drone imagery and the search for patterns in medical data, according to a Cal Poly news release.

Matteson told The Tribune he was excited to see how students will use the systems in the future.

"I'm excited for students to get access to build their own models, and I think the creativity to focus on one problem and solve it really well," he said. "I'm very excited about that."

**How does Cal Poly's AI Factory work?**

The AI Factory is located inside the computer science building, in the heart of campus.

Visitors are admitted through a lobby, where the university supplies earplugs to deal with the loud whirring noise constantly coming from the processors.

Inside, the Nvidia servers are housed in tall, black compartments amid a sea of colorful blinking lights, wires and cords. Walking around the room, visitors may feel a slight breeze or notice a difference in temperature as they move closer to the front or back of the servers, due to the heat they emit.

The Nvidia servers have space for about 1,300 terabytes of information, roughly 6,000 times a typical cellphone's storage, Matteson said. The internet connection used by the servers is about 500 to 1,000 times faster than a typical home's connection, he said.

Each of the four servers includes eight integrated graphics processing units, which Matteson described as "the things that do the heavy lifting for AI and machine learning models."

"All to say, any model that we want to run, we have plenty of capacity, and then also the capacity to train our own models and to fine-tune models," he said. "And what one of the big benefits of that is, we can be more accurate than a general-purpose model, and we can also be a lot more power efficient."

AI data centers have faced widespread pushback in recent years, largely due to energy and natural resource consumption. While Cal Poly's AI Factory is a small-scale data center, Matteson said the university has taken steps to run it in an environmentally efficient way. It uses power from the university's solar farm and recycles water through the system to cool the servers without evaporating any of it. Matteson said all of that energy capacity already existed on campus before the AI Factory debuted.

Data privacy is another major AI concern, and Matteson said Cal Poly has addressed it because the university has full ownership of its new systems and can control who can access the system and what they can access. That ownership also benefits students and staff. For example, users can analyze public health issues using medical information that will not be released publicly until it has received proper authorization.

Cal Poly's system can also be used to create models that address specific topics that widely available AI models are programmed not to discuss.

"There's topics that public models just won't talk about," Matteson said. "They may be things that we need to talk about with the model to do research, especially around public health, drug addiction, mental health issues, and so we have the ability to keep all of our data and all of our programming private."

Matteson said he is also excited to see how the factory will be used by campus community members beyond computer science students. They will be able to use a more user-friendly version of the technology by logging into a website, selecting the software they want and connecting.

"It's something for every level of technical expertise," he said. "Everybody will have a place in this."

원문 보기
Cal Poly has launched an AI data center on its… · Slicast