A Hard Look at Strategic Chip Investment

Estimated reading time: 3 minutes

Jan Vardaman is a key contributing author to an IPC report detailing the capabilities gap in advanced packaging capabilities in the U.S. manufacturing ecosystem. We talked with Jan about the report and the current dynamics in the U.S. market. Jan’s comments provide additional detail and insight to the findings published in the report. 

Nolan Johnson: Jan, how do we take that huge report on gap analysis with IPC and put it into context?

Jan Vardaman: The bottom line is you can put all the silicon fab capability you want in the U.S., but if your goal is to not be dependent on going overseas for your semiconductors, and you don’t have the packaging and assembly done here, you’ve defeated that purpose. It all depends on your goal. We have found there are several smaller outfits here that can do assembly quite capably on a small scale. When there’s large volume, however, it typically goes to Asia where there’s large volume capability and where the investment has been made.

The report indicates there are not any substrate manufacturers that use the ABF material—a buildup film to create substrates here in the U.S. that would serve the high-performance computing market. If you want the buildup substrate to do the assembly, you must go overseas.

There are some people who have assembly capability here for some of their planned activities. For example, Intel says that it is putting its Foveros—basically a 3D chiplet stack with microbumps that goes into a package—assembly in the U.S. The substrates would still be made overseas. Even if they put that assembly facility in New Mexico, your substrate would still not be made here.

Now, it’s a global supply chain and you can make use of the global supply chain if you want to. But if your goal is to have everything done in the U.S., then the plan does not currently achieve the goal.

Johnson: That really does set up the issue, doesn’t it? You can make all the chips you want, but if you can’t get them packaged, then what?

Vardaman: Then what’s the point? What’s the point of spending $52 billion on silicon foundry capability if you’re just going to send it back over to Asia for packaging and assembly?

You can capture this in a quick couple of sentences: First, “We’re going to use our global supply chain.” That’s fine. But you need to recognize that you are not bringing everything back to the U.S. It’s much more expensive to do things here. For example, manufacturing is done here for medical and defense purposes, but the costs are much higher. Assembly and other manufacturing in the U.S. are also more expensive, which is why companies moved overseas in the first place.

In the beginning, we manufactured in the United States. Apple had a huge board assembly plant in Fremont, California, that received visitors from all over the world. I’m here in Austin, Texas, and we have board level assembly for a lot of the Cisco routers. When people talk about electronics manufacturing you need to understand every aspect of the supply chain.

Many people read the report and concluded, “I guess we’re going to put in substrate capabilities.” Not quite, that’s very expensive. If you’re targeting this high-performance market, the feature sizes will require you to produce substrate in a cleanroom. Intel says that putting in a state-of-the-art substrate facility using the build-up film would require about a billion dollars in investment. The equipment suppliers are not in the U.S., typically. The equipment lead-times are two years. Even if you start today, you wouldn’t have that high performance buildup film capability until 2024.

It’s fine to say, “We’re just going to use the global supply chain.” But everybody needs to be on the same page and understand what we’re doing here.

To read this entire conversation, which appeared in the March 2022 issue of SMT007 Magazine, click here.