‘Intel shelves plan for $700 million R&D facility’

‘Intel shelves plan for $700 million R&D facility’

The decision to kill the ‘mega lab’ in Oregon is the latest of the company’s cost-cutting measures.

Intel won’t build a $700 million ‘mega lab’ it had planned in Hillsboro, Oregon, according to a report in The Register. The tech giant plans to pursue cheaper options instead as it implements billions of dollars in cost cuts amid declining revenue.

Intel announced the plans for the site, located 20 miles west of Portland, last May. The company planned to use the facility as a research and development center to prototype, qualify, test and demo its datacenter portfolio using a variety of cooling tech.

Part of a cost-cutting campaign

“We are looking to reduce costs and increase efficiencies through multiple initiatives. This includes exploring more cost-effective real estate options to continue our data center R&D work in Oregon that is already in progress”, Intel spokesperson Penelope Bruce said in a written statement.

The chipmaker is seeking $3 billion in spending cuts for 2023 and billions more in future years, responding to sales that were at least 16 percent below targets last year. The company saw its revenue plunge 20 percent year over year in the last quarter.

Observers note, however, that while Intel is making cuts in some areas, the tech titan is still moving ahead in others. Its planned production facilities in Europe, for example, are still going forward.

Scalable cooling technologies

The report explains that among the headline developments planned for the US West Coast facility was an open reference design for immersion cooling systems. These processes involve submerging whole systems in a tub of dielectric fluids.

Heat from components such as CPUs and GPUs are captured by the fluid and dissipated using either a liquid-to-liquid heat exchanger or evaporation of specialized two-phase fluids developed by the likes of 3M.

Immersion cooling is nothing new, the report says, as the technology dates back decades. Implementing it at scale, however, has proven difficult due in part to a lack of agreed-upon standards.