Subtitles section Play video
Hello, and welcome to another episode of CNBC's Beyond the Valley, I'm Arjun Kharpal in Guangzhou,
China now, Beyond The Valley listeners, you or someone you know, may have struggled to
get their hands on, say, a consumer electronics product recently, maybe a games console.
Or maybe you've struggled to get your hands on a new car.
Well, where has this all come from?
Well, this is a result of currently, what is a global semiconductor or chip shortage
around the world.
And that's the topic of Beyond the Valley today.
Now, these chips are or micro processors are very tiny components that go into many of
the items we use on a daily basis, such as your smartphones, but even things like fridges,
or washing machines as well.
So they're incredibly critical.
And that's really why this global semiconductor shortage is such an issue and a crisis right
now that has impacted industries across the board.
So far, the automakers have really borne the brunt of it at this point, but you are seeing
expand to so many different industries as well.
So how has this happened?
And what's next?
Well, I'm glad to be joined by Sam Shead, who is our technology correspondent, out in
London, Sam, good to have you on beyond the valley for the first time.
Thanks for having me.
I'm pleased to be here.
Well, look, this is a topic you've been looking at for several months now digging deep into it.
And what's happening behind the scenes here, so just give us an overview of you know, how
we've got here?
Well, the simple answer is demand has been outstripping supply.
There's basically not enough chips to go around.
It's a little bit more complicated than that, though.
These chips are made at huge factories or foundries as they're otherwise known, sometimes
fabs too.
And these, these big factories can only make so many chips basically, they have to decide
which chips they want to make months in advance.
And this is where the problems come in.
So last spring, car companies started to reduce their production targets and chip purchases
as the virus spread around the world.
At the same time, chipmakers saw a pickup in demand for semiconductors used to support
things like remote working, and gaming, then the demand for automotive chips rallied back
a lot quicker than most people anticipated.
But the foundries were already busy making other more sophisticated chips for companies
like Apple.
And to make matters worse, Chip factories themselves have also been hit by Coronavirus lockdowns.
That's a great overview and some of the points we're going to dig deeper into with our guest
who we will introduce momentarily.
But Sam, I just want to ask you as well, a lot of major executives of companies around
the world have been commenting on this issue recently.
What have they had to say about what they're facing right now?
And sort of how they see this playing out even when it might end?
So opinions vary, but most people are kind of saying it's going to go on to next year
and possibly 2023.
Elon Musk, CEO of Tesla, he said last week that he sees things getting better next year
as more chip plants come online, but then yeah, as I said, there are some analysts who
think 2023 is more realistic.
That's a, that's a great setup Sam and you know, there's a lot of uncertainty as well.
Of course those are those are what these executives are saying there's so many variable factors
in all of this around the world, including you know how the Coronavirus situation continues
to play out.
And that's certainly going to feed into this.
Well, Sam, that was a great setup and I want to dig deeper into some of those topics with
our guests for this episode Peter Hanbury, a partner at Bain and company at the company's
technology, media and telecommunications practice focusing very heavily on semiconductors.
Peter, thanks so much for joining us on this episode.
Thanks for having me.
So Peter, let's just kick off the conversation with the question, you know, why are we facing
this chip shortage right now?
Yeah, that's a great question.
And a lot of industries are asking this exact question.
We like to think about it really in two phases with slightly different positives.
At first phase was really highlighted by the automotive chip shortage.
And what happened there is when COVID hit, they pulled back of their orders and the capacity
that they had reserved, went to other industries.
Combine that with a lack of traceability and visibility into their supply chains, lack
of inventory and, and they were hit pretty quickly and hard by the semiconductor shortage.
The second phase is a bit of a different story.
And this is where you're seeing books like PC makers, apple, Samsung Qualcomm being impacted,
and the story there is a bit more challenging.
What's driving that is the demand for electronics as folks have shifted to work from home have
increased significantly.
And these industries are now being impacted by just structurally there's more demand than
there is supply available.
And so instead of the auto industry losing a small portion of the semiconductor market
that they used to have access to, this is really the overall semiconductor market has
demand significantly above supply.
And that's a much more challenging structural problem, just given how long it takes to add
capacity in the industry.
So when it comes to the length of the chip shortage, Peter, you've got people like Elon
Musk saying this is a short term phenomenon.
And that things will be back to normal next year, because more and more plants are going
to come online.
But the factories that have been announced by the big chip makers like tsmc, and Intel,
they're going to take years to build.
So those two things seem to be at odds with one another.
What's the true picture there?
That's it's a good point.
And I think there are two ways in which this crisis is all the way number one is, you know,
we build more factories.
Typically to build a new factory, it takes about three to four years.
And so there is some capacity that's coming online, for example, Bosch and Dresden and
Infineon and Villach announced fabs in the 2018 timeframe, and those are just starting
to come online, and they're gonna help this situation a bit.
But like you noted, a number of the announcements made recently, you know, are things that are
not going to show up online until 2024.
So in the short term, you know, the destiny of this crisis from a supply perspective is
really driven by decisions that were made back in 2018.
So there will be some relief from a supply perspective, probably not enough to solve
the crisis before q2 of 2022.
The other way the crisis gets solved a bit faster is from a demand perspective.
You know, if we start seeing demand for PCs, smartphones, servers start declining, for
example, as folks go back to work, that's the other way that you could solve this crisis
maybe a bit faster.
There aren't a lot of indications that that's happening right now.
But that's really the best lever for the crisis to be solved faster than what we currently think.
And as you touched upon there, there are many different types of chips in the world.
There's the high end expensive ones that go into your phone, and then there's the much
cheaper, slightly less sophisticated ones that end up in cars.
Where is the shortage?
And how much does it come from chip makers prioritizing these high end chips?
That are higher margin products?
Yeah, that's a great point.
And you know, if you can think about the semiconductor, people think about the semiconductor market
is one market and it's really not it's 20 to 30 separate markets that are really oriented
by technology.
So a lot of folks think of the bleeding edge chips that go into your smartphone or your
PC.
Those are really the very most advanced chips that are available.
And those chips are designed every two years in magic with a Moore's Law events, and then
capacity is built for them.
So those bleeding edge chips have some shortage, but the factories are essentially built as
the process goes on and demand materializes.
So that very, very bleeding edge is not where you're seeing as much of the challenge.
There are also little way lagging edge shift, you know, things that sit on old six inch
wafers, those are in shortage.
But you know, the capacity is a little bit faster there, there's more players who can
do this, you know, a lot of China's push for self sufficiency means that they're building
a lot of capacity there.
So again, shortage there, but you know, there's some potential relief insight, it's really
those chips that sit in between that is the challenge.
And these would be, for example, from 28 nanometers to 130 meter, there's not many players who
can build these type of chips, because it's a very sophisticated technology, there's a
lot of overlap, and the demand for those chips from PC makers need those type of chips for
display drivers.
But automakers need those type of chips to control their brakes and ECU that go into,
an automobile.
And so there is where you're really seeing a lot of the crunches, you've got kind of
duplicative demand across lots of different industries, and very limited ability to add
supply on that space.
So that middle range of eight nanometers at 130 nanometers, where a lot of the shortage
is most pronounced.
And one of the big issues I guess, with with the the semiconductor industry is the concentrated
supply chain, you know, you may have sort of one company or two companies that are able
to create equipment or make equipment that goes in to eventually manufacturing those
chips, you've got the likes of Samsung tsmc, and Intel that can only make these sort of
leading or bleeding edge semiconductors.
You know, a good example is is asml.
Over in the Netherlands making that extreme ultraviolet lithography machine, you know,
they're one of the only companies in the world that is able to do that as well.
And, you know, we've seen instances reports of the US trying to block shipment of, you
know, asml equipment out to China, and that could hold back China's development in the
semiconductor industry as well.
So as you look at the concentration of the semiconductor industry and the supply chain,
even if some of these short term bottlenecks get resolved, are these structural issues
a concern over the long term?
And you know, could that eventually sort of spout another crisis where we have further
shortages?
Yes.
So the semiconductor industry in general has a number of structural challenges that lead
to these types of issues.
You know, one, the lead times to add capacity or produce an individual chip are extremely
long, you know, to once you design a component in it's very hard to change.
And then three, as you noted, the value chain is very specialized, and concentrated.
And so the most pronounced example that many people focused on is tsmc, at the bleeding
edge of the logic industry, you know, they have about 80% share for the most bleeding
edge technologies right now.
But you're totally right, that if you look at other places in the value chain, you'll
see similar dynamics.
So for example, asml and lithography at the ??? layer deposition.
The resist industry is a material that's used in the manufacturing and highly concentrated
in Japan.
And so you know, you see a lot of this and really the driver is these are very specialized
process.
You know, some of the materials used are multiple nines of guarantee significantly more pure
and complicated than any other industry in the world uses.
And the r&d required to create those as well as the scale required to amortize those upfront
costs are huge.
And so you see a lot of what I would consider natural monopolies in different parts of the
semiconductor industry, driven by the economics associated with producing these complicated
materials, equipment, manufacturing processes, and the specialization of skill, there's just
not that many people who can do it.
And part of the the conversation around supply chain has moved on to countries like the US,
for example, talking about reshoring and in particular, reshoring manufacturing as well
trying to bring back some of that manufacturing onto us soil.
And at the same time, you've got China talking about trying to become more self reliant in
semiconductor industry as well, you know, how much of a challenge those two things you
know, for these countries, and how practical is it?
Yeah, no, it's it's a it's a great question.
And a lot of the emphasis in news recently has been about you know, different countries
and make different types of investments that industry at a high level You know, there's
really two different goals that you could use, or could be pursuing from a government
intervention perspective.
You know, one is what the auto industry wants, you know, they want to add capacity and a
lot of existing technologies, they don't need the bleeding edge.
So they want to go invest, to basically build new capacity, or the industry on all the existing
technologies.
That's an expensive proposition, it's probably $40 billion to add five or 10% of capacity
to every existing node and technology that's available.
But it's possible.
And for a lot of these technologies, you know, you might have 2030 different companies that
could actually do it.
And so that goal is actually you know, quite feasible.
The second goal is more oriented around national security.
And that one's a lot trickier, because what you wouldn't be hoping to achieve is access
to the most advanced technology for artificial intelligence, robotics, to code, crafting,
things like that.
To do that, you need to constantly be at the leading edge, you need to be able to invest
three or $4 billion a year into r&d, every two years, you've got a new technology, so
you got to go build a $15 billion facility, it's very expensive, it's, you know, probably
in the range of $150 billion over just 10 years, to stay at the bleeding edge of Moore's law.
And then added to that you've only got three players who can do that today.
tsmc, Intel and Samsung.
And so if you think about those different goals, it becomes very challenging to think
about, well, how would a European player or Japan, do this?
Well, they don't have one of those three kind of champions or remaining players left.
And so it's very challenging for them, they have other levers they can pull, like asml
sits in Europe, that's a very important pinch point in the supply chain, a lot of the resist
coming from Japan, so another big point of supply chain, so they have other levers that
they can pull.
But being at the bleeding edge and becoming the fourth player with tsmc, Intel and Samsung,
that's going to be very challenging.
And if you've had a crystal ball Peter, and you were looking into the future, how do you
think this chip shortage resolves itself?
You know, there's, there's all of these bottlenecks that still exist, and there's more and more
demand for chips to be put into every single thing we have.
So where do we go now?
Yeah, it's, it's probably what the, I would say, million dollar question, but that I was
thinking billion, though, it's probably more like $100 million question.
Yeah, it's, uh, you know, if you think about kind of where the industry is going, there's
a lot of intervention going on, there's a lot of big claims around investment, you know,
at least, our point of view is that in the short term, it will continue to be very challenged.
So over the next 12 months, we're still going to see a imbalance of supply and demand, a
lot of shortages across industries, you may see some industries like auto do slightly
better, just given by the level and push that they have with the government intervention.
And being a small part of the industry, you can actually increase their production and
only, you know, go from 4% of the industry to eight, and it doesn't really drop the entire
industry where you can't really do that with a bigger part of the industry.
So in the short term, we continue to see it being quite challenged, you'll see different
industries, recover at different paces, different types of chips will recover a different type paces.
So those more basic chips will recover a bit faster than those middle edge chips that we
talked about.
And then long term, you know, we think, given a lot of the subsidies that are being pursued
the focus on adding more supply, and the increased recognition by customers that, you know, these
challenges exist, and you know, they may need to prepare reserve capacity, we think the
industry will eventually return to a more normal balance, and may even go into a bit
of oversupply if companies end up building more capacity in response to these substitutes.
But that's not until probably 2024 2025.
And just on the automaker front, the electric vehicles that are in such high demand today
require significantly more chips.
Do you think that will extend the shortage that we're currently seeing?
Yeah, so the chip auto industry need for semiconductors is a super interesting part of the market.
In general, the need within auto for semiconductors and related electronic components is increasing,
driven by battery electric vehicles hybrid autonomous driving, and so the type complexity
and volume of chips required is increasing significantly.
As you think about you know, the historic model within the auto industry.
A lot of these chips that they use, were, very lagging edge custom designed, you know,
about 100 to 150, what are called ECUs or engineering control units in the vehicle,
everyone uses different chips.
And that's a big part of why they're struggling, they don't know what chips are on those board
boards, there are lots of individual and customized chips there.
So their supply is very fragmented.
And so we think as you look forward in the automotive industry, this is a really important
time for them.
And many of them are starting to do this.
But to basically change the architecture of the vehicle and move towards a model where
we have a more modern design philosophy and design that analogy, where they use more leading
edge chips, more standardized chips, enabled greater software portability, move towards
a model where instead of having 150, ECUs, you know, throughout the vehicle, you might
have more like 10, or 20, and a smarter kind of brain of the vehicle up front.
And so this is a really important time for the auto industry as they make this transition
to move towards more modern design, which will enable their a much more resilient semiconductor supply chain.
And Peter, we've spoken about, you know, the auto industry in relation to this semiconductor shortage.
We've spoken a bit about consumer electronics, are there other industries where you see this
chip shortage potentially impacting at all?
Yeah, I mean, the impact to date has been felt pretty much everywhere that consumes
electronics from, you know, small startups to washing machines.
And so that impact has already been felt.
throughout the industry, there's been a lot of focus on automotive, just given how visible
it is the level of employment at those facilities, the tech industry is being impacted.
And you're starting to see, see that and, you know, lack visibility, excuse me up.
And so the impact is being felt throughout lots of different industries.
Right now, it's just most visible, the automotive and tech space.
And how much of semiconductor prices risen during the chip shortage?
Where do you see them going from here, and who wins and loses out as prices rise?
Now, there's a bunch of different ways in which prices have risen.
The most obvious one, you know, folks really focus on is tsmc.
They have over the last year cut, or increased prices and a couple of subtle ways like taking
away discounts or removing the typical price down curve that they they offer.
And then recently, they've gone and actually, you know, raise the starting price of their,
their wafers.
And so net net, we think those overall prices are increasing about 20 to 30%.
For the fabulous customers who are buying their wafers from tsmc.
In the back end, you're seeing some similar price increases.
That's the assembly and test portion of manufacturing, the packages that go with the chips are also
rising in price.
And so as you look at the impact on, you know, the end products, you're looking at the semiconductor
cost of these components going up about 20 to 30%.
And the impact on the cost of a product will vary depending on how many semiconductors
are in there.
And so a phone where 80 to 90% of the value is semiconductors, you'll see a pretty big
impact for automobiles where you know, the semiconductor content might be 10% of the
vehicle cost, you'll see a more muted impact.
And so the cost of devices will depend on how many semiconductors are in there.
Now the price that consumers pay may behave totally differently.
So for example, automobiles, you know, the cost of an automobile may have gone up 510
percent, the price that a consumer sees may go up much more than that, because the vehicle
is in shortage and it's an opportunity to fully realize the price of the vehicle.
And so there can be a disconnect between price and cost due to the shortage of the components.
Peter, that was a fantastic insight into what is an incredibly complicated topic.
Thanks so much for joining us today on beyond the valley.
Thank you
It's going to be be very interesting to see where this goes how long it takes to resolve
itself.
It should hopefully be in a much better place by this time next year.
But the fact of the matter is, most of today's chips are still made in Asia and TSMC has
a lot of power.
There are a couple of other players like Samsung and Intel as well, but if there's another
pandemic or if there's another supply chain breakdown somewhere, this whole thing...there's
a lot of power concentrated in one particular corner of the world...and things may not go
as smoothly as they have on this occasion.
Yeah, I think that's a great point, you know, structurally in this market.
You know, there are still some issues, though we were talking about it in that conversation
with Peter there.
And that is the concentration of power in certain companies hands and the concentration
of players in certain parts of the supply chain, as well.
And I think that's going to be a key issue going forward.
I guess that feeds into the geopolitics of it always why the US is talking about wanting
to reassure manufacturing, it's why you know, you're in Europe, the EU are talking about
this issue as well, right Sam?
Yeah, that's right.
But these things will take time, and Europe can't just all of a sudden start competing
with Asia, you know, they need talent, they need a lot of money.
And this is going to take a long time, whether governments, and politicians will kind of
make the right investments, remains to be seen.
absolutely, yeah.
Sam, It was excellent to have you on beyond the valley for this episode.
I think, you know, we'll definitely get you back soon.
Fantastic, yeah, look forward to joining you again Arjun.
Yeah, and thank you for all our viewers and all our listeners for watching this episode
and listening to this episode.
Of course, you can subscribe to our YouTube channel, and also download the podcast episode
on Spotify, Apple and Google's platforms.
That's it for another episode of CNBC's beyond the valley.
Thanks for listening and watching and we'll catch you next time.