AI-Driven Analog Verification from Mach42 Targets IP Reuse
Cette publication existe aussi en Français
Startup Mach42 is taking aim at one of analog design’s persistent pain points: slow, manual verification processes that don’t scale well when it comes to IP reuse. The company’s newly launched Discovery Platform combines machine learning with high-performance simulation and cloud infrastructure to help engineers quickly validate analog circuit performance across a wide range of conditions.
For eeNews Europe readers working on analog, mixed-signal, or custom IC design, the news is notable. By automating design space exploration and streamlining spec compliance checks, Mach42 could deliver faster time to market, fewer re-spins, and better support for reusing IP blocks across different PVT corners.
Tackling IP reuse and spec violations with ML and SPICE
The platform takes in a design’s netlist and user-defined specs, then uses machine learning to pinpoint out-of-spec behavior before validating any issues using Cadence’s Spectre Simulation Platform, a trusted reference in SPICE accuracy. According to Mach42, this gives designers confidence that violations aren’t just flagged quickly—but verified accurately.
“Mach42’s AI-powered solution is accelerating the path to reliable, scalable, and efficient IC design verification,” said Bijan Kiani, CEO of Mach42. “Recent enhancements to the Discovery Platform specifically target IP reuse challenges by enabling robust performance validation across a wide range of operating conditions, including voltage, temperature, and process corners. These recent innovations are redefining the potential of IP reuse by enabling designers to confidently validate circuit correctness across diverse operating conditions.”
Cadence partnership
The platform is tightly integrated with Cadence tools, and Mach42 has joined the Cadence Connections Program to further align its offering with existing design workflows.
“We’re pleased to have Mach42 as a new member of our Connections Program,” said Aparna Dey, Sr. Product Marketing Group Director, Ecosystem and Standards, at Cadence. “Through the Connections Program, Cadence is committed to providing complementary solutions to our customers to help tackle critical challenges in performance, accuracy, and interoperability.”
Behind the scenes, Mach42’s team has built proprietary neural network models that deliver significant speedups over traditional simulation flows—with minimal data and high fidelity. The end goal is a verification process that’s not just faster, but more scalable and repeatable, especially in analog-heavy applications where every spec matters.
The platform is already in use with early customers, and Mach42 sees it as a fit for companies looking to unlock IP reuse without sacrificing confidence in performance validation.
If you enjoyed this article, you will like the following ones: don't miss them by subscribing to :
