In our October 2022 Semiconductor Digest webinar, Yuji Minegishi, General Manager at Gigaphoton, and Jacob Strock, a Gigaphoton data scientist, explained how the Fabscape platform works to solve the semiconductor industry’s greatest puzzle.
At first glance, it looks like today’s semiconductor manufacturers have no shortage of data whatsoever. But while data in its simplest forms is easy to come by, data that is unified, insightful, and actionable is not. The culprit? A siloed and complex software environment that causes even the biggest OEM software budgets to fall short.
Below are a few highlights from the presentation. You can watch the full recording here.
The Challenge of Silos in the Semiconductor Industry
Today’s dilemma is that the deeper the insight you seek from your equipment, the more and more you will need to rely on equipment vendor-specific solutions. The key is to find a way to gain deeper multi-vendor equipment insight. While there are SEMI Standards that exist, sometimes standards on paper are not enough.
There is a whole other layer of advanced actionable analytics—many of them AI-powered—that can be unlocked if equipment and software system vendors and customers have a better environment to work together to harness deeper actionable insights from equipment data.
How the Fabscape Open Platform Works
In the simplest terms, Fabscape is a collection of microservice containers each providing a specific service within their overall system.
- A container is kind of like a package that contains an executable program and all the dependency it needs to run on its own to provide a specific function or service.
- A microservice is just a term that refers to a broader system that is made up of a set of interlinked containers.
The Fabscape plugin and driver system allows users to create their own feature extensions as containers that can easily be hot installed or uninstalled from the system dynamically without bringing it down or restarting.
Why Successful AI Models Need to go Beyond Algorithms
According to data scientist Jacob Strock, the semiconductor industry needs to begin investing in predictive and prescriptive technologies that allow semiconductor manufacturers to predict into the future, make recommendations, and achieve desired outcomes.
It’s not just about what’s happening today. How will your equipment work a month from now if you change XYZ configuration? While it’s attractive to talk about the latest algorithms and domain knowledge, it takes a lot more to bring a model to production and make it usable.
The Path to Open Platform Success
At the end of the day, a successful open platform needs to be owned and driven by the industry, both chip manufacturers and vendors alike. A truly open platform can’t really depend on one vendor alone, but will depend on the industry banding together to evolve a new software ecosystem.
The above content represents just a small portion of the original webinar. Watch the full Semiconductor Digest webinar recording now!