Nvidia, Equinix team up to offer AI supercomputers to businesses

By Stephen Nellis

SAN FRANCISCO (Reuters) - Data center operator Equinix and Nvidia said on Wednesday they had partnered up to offer the chip firm's supercomputing systems to corporate clients.

The service will make it easier for companies to own artificial intelligence computing systems and have better control over their data, instead of renting out Nvidia's market-leading chips from cloud computing providers like Amazon.com or Microsoft.

Many large businesses have shown interest in owning their AI computing system for privacy and security reasons, but Nvidia systems are different from conventional data center servers, often using different networking cabling and sometimes employing liquid cooling.

Under the deal, Nvidia has trained Equinix staff in how to build and run its systems. Corporate customers will purchase Nvidia systems but pay Equinix to build and run them efficiently while still retaining ownership.

"We want to drive privacy because what we're hearing at large customers is they are very worried about controlling their own destiny," Jon Lin, executive vice president and general manager of data center services at Equinix, told Reuters in an interview.

However, "being able to then figure out how to operate (AI data centers) efficiently and effectively - this is an entirely new set of capabilities that is not generally available."

Nvidia and Equinix said the new offering is available as of Wednesday but did not disclose any immediate customers.

(This story has been corrected to clarify that liquid cooling is not a requirement for Nvidia systems in paragraph 3)

(Reporting by Stephen Nellis in San Francisco; Editing by Edwina Gibbs)

---

Serious News for Serious Traders! Try StreetInsider.com Premium Free!

---

You May Also Be Interested In

• NVIDIA (NVDA) Says It Will Take a Cut of Some Customers’ Cloud Revenues - The Information
• Continental to sell ContiTech unit to Lone Star Funds for $4.6 billion
• Tesla Q2 deliveries crush forecasts at 480,126, erasing backlog concerns