Nvidia CEO Jensen Huang and Equinix CEO Adaire Fox-Martin, speaking at an Equinix event in San Francisco on September 2, 2026, highlighted the evolving role of data center infrastructure in the burgeoning artificial intelligence era. Long before the dominance of hyperscalers and the current AI data center boom, Equinix established itself as a critical provider of colocation services.
The company, founded in 1998 during the dot-com bubble, offers essential data center space, power, cooling, and security for over 10,500 clients housing their servers, routers, and storage systems. This extensive experience positions Equinix to capitalize on the projected over $5 trillion investment in AI data centers by 2030, as forecasted by Goldman Sachs Research.
A recent collaboration between Equinix and Nvidia, announced concurrently with the event, empowers customers to deploy and manage their AI models through the open-source cloud platform, Together AI. This strategic partnership underscores Equinix’s adaptability and its commitment to serving the dynamic needs of the AI ecosystem.
Equinix’s market performance reflects this strategic positioning. The company’s stock has surged 33% year-to-date, outperforming major technology firms and elevating its market capitalization to $100 billion. This makes Equinix the preeminent data center Real Estate Investment Trust (REIT), significantly ahead of competitors like Digital Realty, which holds a market cap of $68 billion.
**Navigating the AI Infrastructure Landscape**
Hyperscale cloud providers such as Amazon and Google, alongside specialized “neocloud” entities like CoreWeave, are currently channeling hundreds of billions of dollars annually into digital infrastructure. A substantial portion of this investment is dedicated to supporting leading AI research labs like OpenAI and Anthropic. In contrast, Equinix offers a more diversified approach, providing space within its facilities to a wide spectrum of customers. These clients can house servers across various compute platforms, leveraging hardware from industry giants like Nvidia and AMD.
During a video conference at the San Francisco event, Nvidia CEO Jensen Huang emphasized the strategic advantage of Equinix’s distributed data center locations. He noted that these facilities enable proximity to where data is generated (“where all the sensors are”) while also allowing for remote management and flexibility (“you could both simultaneously be close and be far away”).
With 281 legacy colocation facilities spanning 77 metropolitan areas across six continents, Equinix boasts a deeply entrenched global footprint. Maryam Zand, Equinix’s Vice President overseeing its AI ecosystem strategy, stated, “The companies you already use are all running on us.”
Financial specifics of the Nvidia and Together AI deal were not disclosed. However, Zand elaborated that Together AI will serve as the seller of record for its end customers utilizing the new program, named Equinix Inference Exchange. This initiative, slated for launch in the first quarter of 2027, aims to provide a streamlined platform for AI inference.
**A Strategic Shift Towards Inference**
While AI training focuses on model learning from vast datasets, AI inference – the process of making decisions based on new information – is increasingly becoming paramount. As AI applications evolve from basic conversational agents to more sophisticated autonomous systems, the demand for efficient inference processing is escalating. This necessitates a shift towards utilizing a broader range of processors, including Central Processing Units (CPUs), rather than relying solely on Graphics Processing Units (GPUs) traditionally favored for their parallel processing capabilities.
Historically, Equinix has maintained a strategy of deploying smaller data centers in urban proximity, functioning as crucial network interconnection hubs. This contrasts with competitors like Digital Realty, which have focused on developing larger-scale facilities catering specifically to hyperscale clients.
Equinix’s xScale division, dedicated to hyperscale customers, primarily operates facilities under 100 megawatts, a scale significantly smaller than the gigawatt-level power requirements of some cutting-edge AI data centers. Longtime data center analyst Vlad Galabov observed that Equinix was “too slow” to address the gigawatt-scale demand, leading to newer, more agile players entering the market to meet the rapid build-out needs of emerging AI projects. He added that established colocation providers like Equinix and Digital Realty are now re-evaluating their strategic planning to align with these evolving demands.
**Expanding Connectivity and Services**
Further underscoring its commitment to the AI landscape, Equinix also unveiled Equinix Fabric One. This new connectivity service is designed to simplify network operations across multi-cloud and AI model environments.
Zand described the Equinix Inference Exchange as an “inference platform as a service,” enabling clients to seamlessly connect to various clouds and providers, execute inference on open-source models, and optimize costs through “tokenomics.” She further indicated that Equinix’s data centers are equipped to handle Nvidia’s B300 Blackwell Ultra GPUs and are also incorporating liquid-cooled facilities for advanced chips like the Vera Rubin.
Equinix’s urban presence provides a distinct advantage, particularly for inference workloads that demand high-speed, low-latency communication between servers and end-users. McKinsey forecasts that by 2030, inference will constitute half of all AI compute and contribute 30% to 40% to overall data center demand. Zand emphasized the rapid pace of change within the AI ecosystem, stating, “Customers need to be able to move as the market moves.”
Despite these strategic advancements, not all market participants share an optimistic outlook. Noted short seller Jim Chanos expressed skepticism in May, labeling both Equinix and Digital Realty as “not great businesses” and citing their capital-intensive nature and limited profitability. He differentiated these legacy data center companies from the new wave of AI-specific data centers.
In its latest quarterly report, Equinix announced a 16% year-over-year revenue increase to $2.63 billion. While this growth is substantial, it is noteworthy that CoreWeave, a leading neocloud provider, more than doubled its revenue to $2.58 billion in the same period. However, Equinix reported a net income of $477 million, whereas CoreWeave incurred a net loss of $626 million.
Galabov characterized Equinix’s strength as its “super diversified” model, offering a broad range of general-purpose compute and services to a vast client base. This diversification, he suggests, insulates the company from specific AI bubble risks, though it might mean missing out on some of the highest-growth opportunities. He concluded that this approach represents a trade-off between high risk and high reward.
Original article, Author: Tobias. If you wish to reprint this article, please indicate the source:https://aicnbc.com/25410.html
