Celera samples analog IC designed by AI

New Products | March 18, 2025

By Peter Clarke

Cette publication existe aussi en Français

Celera Inc. (San Jose, Calif.) has started sampling an analog IC designed by an autonomous AI-enhanced software platform.

Celera, a startup founded in 2018, has been developing its AI enhanced “fully-automated” analog and mixed-signal IC development platform for several years.

Celera has developed a software platform called ChipHub that it claims improves engineering productivity by a factor of ten. This has allowed the design of 60V down to 3.3V DC-to-DC buck converter from specification to manufacturing release in a matter of days, the company said.

“This is a major milestone for Celera and an important breakthrough for our customers,” said Pat Brockett, Celera’s CEO. “Celera has demonstrated that end-to-end automated design of high-performance analog ICs can be done.”

Brockett has served as an advisor to the company since 2019 but has now joined as CEO. He emphasized the Celera’s business model is to develop custom ASSPs for customers as a chip product company under the Celera brand. Brockett previously led Zarlink and Summit Microelectronics as CEO after spending many years at National Semiconductor.

The buck converter targets a 130nm BCD manufacturing process at Vanguard Semiconductor but Brockett declined to name the customer. Celera is foundry agnostic and can develop AI expertise across the breadth of analog circuits including op-amps, comparators, filters, converters and power management functions, he said.

“This first customer product is a state-of-the-art high-voltage step-down converter for industrial and automotive applications,” said Calum MacRae, CTO and founder at Celera. “Our Nesto technology simplifies analog IC design, enabling even non-IC designers to generate custom silicon. A huge benefit of our Nesto technology is that the same algorithm can be used to quickly produce whole families of buck converters in hours.”

MacRae added: “Our IP is all in digital form. This allows us to generate data to train machine learning (ML) models, producing AI agents for analog design, layout, and modeling. The ability to generate large amounts of synthetic data positions Celera as the only company able to apply ML to analog design.”

The company claims that product designs can be superior in die size, cost and performance as well as being designed and verified more quickly.

Since its formation in 2018 Celera has raised US$14 million and is now looking to close a Series A round of funding for about US$20 million by the end of 1H25, Brockett told eeNews Europe. “We are already engaged with major customers designing products for consumer, datacenter, wireless , industrial, solar and automotive applications,” Brockett said.