Equinix Turns To Big Tech As Corporate AI Demand Is Slow To Materialize

Under new leadership, data center REIT Equinix is accelerating its short-term shift toward large-scale artificial intelligence data centers for the world’s largest tech companies — even as it insists its long-term opportunity lies mainly in its traditional colocation business. 

Equinix CEO Adaire Fox-Martin took the helm of the world’s largest data center REIT in June after more than two years in top positions at Google Cloud. Given her Big Tech background, it is perhaps appropriate that the early months of her Equinix tenure will see the retail colocation giant pivot more aggressively toward developing large-scale campuses for cloud providers in the U.S. — a market the firm had no foothold in just one year ago. 

Just months after opening its first single-tenant hyperscale project in the U.S., on Wednesday, Equinix announced plans to build a hyperscale campus totaling hundreds of megawatts of capacity in the Atlanta market, the largest such development in the company’s history. 

With Equinix executives hinting that there will be more to come, the projects represent a departure from the firm’s traditional business and development strategy in the U.S.

But Fox-Martin and other leaders at the REIT say this push into the U.S. hyperscale market isn't a pivot away from the firm’s core colocation business lines. Rather, it is a reactive shift to capitalize on unprecedented demand for AI-enabled hyperscale facilities, even as the firm sees its long-term AI growth coming from its more traditional multitenant offerings. 

“Our approach to the market opportunity created by AI is multifaceted, and we believe it will deliver value in the short term and sustainable growth over the medium to long term,” Fox-Martin said on its second-quarter earnings call Wednesday. “I think retail will have a much broader demand and a more meaningful long-term upside from AI as use cases move from proof of concepts into production.”

Equinix, which operates 264 data centers in 72 markets globally, is coming off a second quarter in which the REIT posted record bookings and generated $2.2B in revenue, up 7% from a year prior.

The company anticipates the same pace of growth will continue through the end of 2024. It has continued to expand its global footprint, with $648M in quarterly capital expenditures helping fund 54 projects underway. 

While Equinix’s business model has traditionally focused on retail colocation and other multitenant business lines, 15 of the developments the firm has underway are part of its xScale program, in which the REIT forms development joint ventures to build single-tenant facilities for cloud and social media giants, companies like Amazon, Microsoft and Meta. To date, Equinix has spent $4.7B on the initiative and has 20 xScale data centers in operation globally. 

Equinix leadership said on the earnings call that the firm will be accelerating this program and “meaningfully augmenting and extending our xScale portfolio” in the coming months, particularly in the U.S.

This push into U.S. hyperscale marks a strategic shift. As recently as the start of this year, Equinix’s xScale initiative exclusively targeted overseas markets. 

Equinix opened its first xScale campus in the U.S. this spring, fully leasing a 14-megawatt hyperscale facility in San Jose developed through a $600M joint venture with PGIM. The project may eventually expand to as much as 28 MW. 

Equinix says it is embarking on a larger xScale project in the U.S., this time in Atlanta. Fox-Martin told analysts Wednesday that the firm has already closed on land and power for what she described as Equinix’s “first multihundred-megawatt xScale campus.”

While she didn't provide further detail on the location of the project, the scope of the campus or the identity of Equinix’s partner in the joint venture, the project marks an escalation in scale from anything the firm has built previously. The firm’s entire xScale portfolio today totals 405 MW of capacity.

With the planned Atlanta project, Equinix appears to be ramping up its push into the market for the increasingly massive campuses that are spearheading the unprecedented wave of data center development in the U.S. Demand for ever-larger campuses with access to hundreds of megawatts of power has exploded over the past 18 months, as cloud providers and social media giants scramble to secure the infrastructure to support the development and training of energy-intensive generative AI models. 

It is the REIT’s first foray into a space that has been almost exclusively the realm of self-development by the world’s largest tech companies or private third-party developers backed by institutional capital, companies like Stack Infrastructure, PowerHouse Data Centers and Compass Datacenters. 

“We plan to lean into this program, as we've seen strong demand for this offering evidenced by both our cloud and AI bookings momentum,” Equinix Chief Financial Officer Keith Taylor said. “We continue to believe this off-balance-sheet JV structure with our equity partners is the right model to pursue this significant opportunity.”

But Equinix’s leadership said this isn't a pivot away from the multitenant facilities in high-demand markets and enterprise-focused business lines that have been the firm’s bread and butter. Ultimately, they say, their enterprise-focused colocation businesses will be the biggest beneficiaries of the AI revolution. 

For now, the major cloud providers like Microsoft and Google and social media giant Meta are driving the vast majority of AI adoption, and their cloud-based AI services will likely continue to account for a significant percentage of corporate spending on AI. But many companies will want to control some or all of their AI computing themselves, with around a third of corporate AI workloads expected to go toward colocation rather than cloud.

A July report from Flexential suggests the vast majority of large companies have AI road maps in place, most involving AI infrastructure separate from the public cloud services like Amazon Web Services or Microsoft’s Azure. 

Equinix leaders said the firm’s data center portfolio leaves it uniquely positioned to capture a key chunk of the emerging AI marketplace.

They pointed to Equinix’s global portfolio of AI-capable multitenant data centers in major markets and robust network connection to public cloud providers, which is critical for firms balancing their infrastructure between computing they operate themselves and the cloud. Equinix also offers private cloud services for companies concerned about the security or compliance risks that come with giving their data to major cloud service providers. 

“In the context of AI and AI demand, that initial short-term demand is coming, indeed, from the service providers,” Fox-Martin said Wednesday. “But many CIOs … are looking to ensure that they have an AI strategy, and we are beginning to see the beginnings of enterprise training and enterprise funnel as customers look to evolve their strategy into proof of concept and beyond that into working production systems.”

Equinix may also benefit from a gradual transition from demand driven by training AI models to demand driven by AI inference, or the computing to apply those models for end users. Unlike AI training, which can be located almost anywhere, inference computing will often need to be hosted in facilities in major markets near large concentrations of end users. 

According to Fox-Martin, Equinix’s portfolio is ideally positioned to absorb the coming wave of inference demand, along with enterprise AI training. 

Still, she acknowledged that this demand has yet to materialize or translate into leasing activity, a slow pace of corporate adoption that has been corroborated by numerous recent studies. A push into hyperscale campuses serves as a bridge until this demand materializes, even as Fox-Martin insisted the firm is seeing the early signs of that demand wave taking off.

“We're seeing early traction in the enterprise-level inference and those types of workloads beginning to occur in our data center environment,” she said. “But that is something that we'll see extrapolate and grow over the medium to longer term.”

These shifts in development strategy and leadership come as Equinix faces scrutiny from the Department of Justice and the Securities and Exchange Commission over allegations of improper accounting practices. It received federal subpoenas this spring in the wake of a report published by a prominent short seller accusing the firm of, among other things, manipulating key accounting metrics to inflate the company’s share price and executive compensation. 

An internal audit conducted by Equinix’s audit committee concluded that the company’s financial reporting had been accurate, but the federal inquiry reportedly remains open. 

On its Wednesday earnings call, Equinix leadership brushed off the lingering allegations. They said they are cooperating with the probe and expressed confidence that the firm’s financial disclosures will be found to be accurate. 

“Overall, we feel very confident about it,” Taylor said. “I feel very comfortable with our financial disclosures.”