The partnership announced Thursday by the Defense Advanced Research Projects Agency addresses concerns about the long-term consequences of the country’s lack of microchip production. The U.S. relies largely on Taiwan for microelectronics manufacturing, prompting fears that China may tamper with chips for weapons and other critical defense platforms.
The announcement dovetails with a global semiconductor chip shortage that led the Biden administration to commitment of $37 billion to speed production in the U.S.
DARPA will pay an undisclosed sum to Intel to help boost production, and the total cost will be “tens of millions” for the program titled Structured Array Hardware for Automatically Realized Applications, or SAHARA, an agency spokesperson said.
DARPA and Intel will work with researchers at the universities of Florida, Maryland and Texas A&M to automate processes to boost production of a type of chip — structured application-specific integrated circuits — that allows unique security features, performs better and consumers less powers. Those features make them “an efficient and effective alternative for defense electronic systems,” the DARPA announcement said.
With the researchers, Intel will develop new ways to protect data against reverse engineering and counterfeiting attacks. The researcher “will use rigorous verification, validation and new attack strategies to test the security of these chips,” an Intel statement said.
Today, the military widely uses a different kind of integrated circuits — field-programmable gate array chips — that can be programmed for different applications. The project seeks to speed up conversion of those chips to the application-specific circuits that the Pentagon would prefer to use.
Automating the conversion process would “shorten the design process, reduce associated engineering costs, and enhance chip security,” the DARPA news release said. The project aims to reduce design time by 60 percent, decease engineering costs ten-fold, and halve power consumption, Serge Leef, program manager in DARPA’s Microsystems Technology Office, said in the statement.
Manually converting chips to the better-performing version “is a complex, lengthy, and costly process, making it difficult to justify the economic burden at the volume of custom chips required by DoD applications,” the release said.