Equinix will work with South Korean AI chip company FuriosaAI to expand the AI ecosystem. It will provide infrastructure in Europe to support FuriosaAI's entry into overseas markets. It will also respond to the agentic AI era with a "distributed AI hub" and "Fabric Intelligence" strategy.

Equinix held a media briefing titled "Network infrastructure trends and future strategy in the AI era" at its Korea office in Yeongdeungpo-gu, Seoul, on Sept. 15, and disclosed the status of its cooperation with FuriosaAI.

FuriosaAI has built servers based on its second-generation AI accelerator, Renegade, at an Equinix data centre in Lisbon, Portugal. Ji-hoon Yoon (윤지훈), a product manager at FuriosaAI, said it has been about a month since the servers were deployed and it is onboarding customers and expanding the target base. He said FuriosaAI chose Equinix because it judged it important to work with a specialised partner so end users can use the services comfortably and for infrastructure operations.

FuriosaAI support for Europe expansion... connectivity automation with "distributed AI hub"

FuriosaAI will use local Equinix data centres in Europe to connect customer service environments and widen touchpoints with potential customers within the Equinix ecosystem. After conducting proof-of-concept work for European customers, it will either move to customers' own infrastructure or link with cloud partners to expand into commercial environments. Yoon said it is securing an additional foundation by working with Equinix in markets where local infrastructure is important.

The two companies are also reviewing expanded cooperation in South Korea. FuriosaAI plans to mass-produce 100,000 AI accelerator cards from this year through next year and step up efforts to target overseas markets including Europe, the United States and Singapore. Yoon said it is also planning various forms of cooperation with Equinix in South Korea beyond colocation.

Equinix is also expanding its "distributed AI hub" strategy that connects distributed infrastructure such as AI models, data and cloud. Anthony Ho (앤서니 호), product management director for Asia Pacific at Equinix, said how to connect AI infrastructure is becoming more important as the industry evolves into the agentic AI era.

The distributed AI hub addresses security and cost control, management complexity and other issues that arise when multiple departments within a company use different AI models, services and data. Equinix has already built an "AI Discovery Hub" in Hong Kong with HPE, Nvidia, Dell and Schneider Electric. It is a solution that supports companies in testing and verifying the performance of actual applications and AI models before adopting large-scale AI infrastructure.

Equinix has also recently unveiled "Fabric Intelligence" for automating network operations. It is a platform that uses AI to automate the connection and operation of distributed AI infrastructure, including cloud, neocloud, on-premises and data platforms.

Fabric Intelligence uses Fabric Super Agent, a Model Context Protocol (MCP) server, a Fabric Application Connector and Fabric Insights. Ho said the goal is to provide AI networks at the speed of AI.

"Connecting data, cloud and networks is important"

Equinix is accelerating the expansion of its domestic colocation centres in addition to its cooperation with FuriosaAI. It currently operates 3 data centres in South Korea and will open an additional data centre in the Seoul metropolitan area in the first quarter of 2028. It is also reviewing building an AI Discovery Hub in South Korea similar to the Hong Kong case.

It also shared its view on the government's large-scale AI data centre project. Hye-deok Jang (장혜덕), country manager of Equinix Korea, drew a line by saying it is a market different from Equinix's business model. He said large-scale AI data centres differ from Equinix's retail approach because they target a small number of hyperscalers and large AI companies.

He said AI competitiveness will now be determined not only by semiconductor performance but also by the ability to efficiently connect data, cloud and networks.