The chip shortage is a manufacturing problem that won’t be easy to solve

So, here’s what’s going on with the chip shortage thing. First, among U.S. chipmakers, only Intel fabricates its own chips in the U.S. The rest contract with big companies, mostly in Taiwan or South Korea, known as fabs, which is short for semiconductor fabrication plants. The biggest are TSMC and Samsung. These facilities are incredibly expensive and take years to build and even upgrade.

Add in the pandemic, lots of people at home buying computers and slowdowns in the actual manufacturing, and there aren’t enough chips for cars, medical equipment or all those other devices.

Last week, President Joe Biden requested $37 billion from Congress to kick-start the domestic supply chain for chips and issued an executive order calling for a larger review of U.S. supply chains.

Anshel Sag covers the semiconductor industry at Moor Insights & Strategy. I asked him if $37 billion will be enough. The following is an edited transcript of our conversation.

Anshel Sag: I think $37 billion is very likely a down payment. I’m not really sure it’s enough. Because ultimately, if we want to move enough chip-producing capacity to the U.S., there needs to be a considerable amount of investment to incentivize the TSMCs and the Samsungs of the world to do that.

Molly Wood: So the companies that develop chips in the United States, many of them all contract with the same foundry, right? TSMC. So it’s not like they’re completely in charge of their own supply. They are constrained by how many chips TSMC can produce and what other arrangements it has to sell chips to other companies?

Sag: Exactly. So there’s TSMC, there’s Samsung and there’s GlobalFoundries. Those are the bigger ones. TSMC is a provider of everyone’s silicon. And based on the customer relationship that a company has with TSMC ultimately determines what priority they get in that pecking order of what’s available.

Wood: And then tell me about Intel, which does have its own foundry, but it doesn’t seem like that’s going that well.

Sag: Yeah. They had some of their own challenges with these newer process nodes. And they’ve been struggling to overcome them, and they just recently got a new CEO. And they’re really aggressively pushing for resolving these issues. But they’ve shown that they’ve been able to overcome some of their challenges, and increase their supply, because they had shortages of their own prior to the pandemic. And now they seem to be kind of getting back on track to a degree. But ultimately, Intel was only producing Intel products.

Wood: This is a complicated problem to solve. Do you think there is a best solution? Is it just that we have to wait out COVID and this shortage and beef up companies like TSMC? Or is the Intel model, of companies essentially producing their own chips, going to be the future?

Sag: I think there is a valid concern in wanting to have a certain percentage of chip capacity in the U.S. We have to just figure out what is the right number for how much chip capacity we need in the U.S., and how much we as U.S. taxpayers are willing to front that bill to ensure that national-security level of chip supply, right? Because it will most likely come down to a, you know, minimum necessary investment. Because I just don’t think that it makes sense for the U.S. to try to replicate what TSMC and Samsung have done in Korea and Taiwan because it’s double-digit billions. And that’s just a short-term cost.

Wood: What role does Apple play? I mean, wasn’t Apple pretty early to go fabless, and maybe kick-start some of this trend? And now Apple is like Intel, creating its own chips for many of its products and building this kind of big Apple-only business. I’m just kind of curious about where they sit in this whole landscape.

Sag: Well, it’s interesting because like you said, Apple is fabless, and they do design their own chips, but they ultimately still do depend on TSMC to fab those chips, and I believe they are TSMC’s No. 1 customer. But they do have a role in the industry, where they do influence what kind of tools are necessary for the fabs to purchase, what kinds of materials and new technologies are being adopted. Because ultimately, you know, what Apple wants does drive how the market moves in terms of sensors and chips. So I think Apple’s influence is one, its size. They do influence how much the industry will grow. And they also influence what kinds of technologies get adopted because their scale makes certain types of technologies profitable. And I think with this chip shortage, you’re seeing Apple not really talk about it very much at all because they’re kind of the first ones in the pecking order. And I think that they’ve anticipated chip demand just enough that they probably don’t really have much to be worried about.

Wood: So on the pecking order question, are automakers just like low on the priority list?

Sag: I don’t think that they’re necessarily at the bottom, but I think by canceling their orders, I think they became the bottom, and they’re not necessarily anywhere near high a volume as smartphones are. So I think they kind of hurt themselves pretty badly when it came to that. And it’s been very difficult for everybody to find chips, be it for a PC or for other things. If you ask anyone who’s tried to build a computer lately, it’s not very easy or fun. And I think that if you’re trying to find chips for a car, it’s probably even worse just because a lot of these auto manufacturers canceled their chip orders, which means their chip suppliers had to also cancel orders and it takes two to three to four months to get those things spun back up again.

Related Links: More insight from Molly Wood

There’s a story in the Financial Times (behind a paywall) on this superinteresting thing that Anshel Sag noted, about how device makers actually increased their orders for chips at the beginning of the pandemic, while carmakers furloughed workers and canceled a lot of orders. But now that the carmakers can’t get back into the queue, so to speak, more of those furloughs are becoming permanent, and that’s becoming a bit of a political problem for President Biden, who’d like to be reopening American manufacturing as soon as possible.

It’s also an interesting tech story because the chips in tablets and computers and phones are smaller and more sophisticated than the chips in cars, so these fabs are increasingly gearing their facilities toward the device chips and away from the car chips. But the automakers, according to this story about it in the Washington Post, are trying to save money by sticking to the older chips because there are a lot of upfront durability and safety costs when you switch out a car component.

And of course there’s a 5G element here too. You need even more chips of various sorts in 5G phones for managing power so your battery doesn’t die, and the better phone cameras get, the more display chips you need, and all of those are also needed in cars, and eventually you just come back to the one inescapable problem, for which $37 billion is a somewhat anemic down payment, and that problem is: Man, people need to make more chips.