US Puts in $50bn (Rs.
3.50 lakh crs) to Push Chip Manufacture
·
Chip Industry is Cyclical
And yet – the CHIPS Act, as drafted, gets both
of these points wrong.
The bill proposes a generic subsidy program, to
channel money to build up onshore (domestic) production capacity. It targets
what is essentially a nonexistent “problem” – the private sector is already
addressing this vulnerability with resources far in excess of what CHIPS would
provide. (See below.)
First, what
is “strategic” in this context?
Vulnerabilities in the semiconductor industry are
either short-term or strategic. Supply chain constraints – of the sort
bedeviling the automotive industry right now – are generally short-term. This
problem is actually endemic in the chip business. Given the economics and
logistics of adding capacity to address sudden shifts in demand, the
semiconductor industry becomes highly cyclical. Boom-and-bust patterns prevail,
with shortages and gluts alternating as the headline crisis du jour.
The pandemic whiplash has produced an unusually severe cycle of this sort, but
the natural corrections are already underway.
·
“Historians will probably spend decades
picking apart the consequences of the COVID-19 epidemic. But the shortage of
chips that it’s caused will be long over by then. A variety of analysts agree
that the most problematic shortages will begin to ease in the third or fourth quarter of
2021, though it could take much of 2022 for the resulting chips
to work their way through the supply chain to products. The supply
relief will not be coming from the big, national investments in the works right
now by South Korea, the United States, and Europe but from
older chip fabs and foundries running processes far
from the cutting edge and on comparatively small silicon wafers.”
This is good old supply-and-demand stuff. Market
mechanisms do resolve these problems, as painful as they may be in the near
term, and the case for government intervention is not strong. The timing alone
works against it; government action under our democratic/technocratic system is
normally much slower than the natural market adjustment process. Passing
legislation, setting up programs, qualifying the recipients of funding,
disbursing the money… by the time all that is accomplished, the new programs
will be solving last year’s problems.
The true strategic advantage of the U.S. is the
deep expertise in the private sector, embodied in decades and millions of
man-years of experience with an incredibly complex technology. Which means that
the soil from which sound strategy will spring is not in the government’s
domain. The leadership of the effort should emerge from this foundation of
private sector experience, not from government mandates.
Software is
more important than hardware
So, for example, the CHIPS bill should not
call out specific technology targets — like “3 nanometer processes.” 3 nm is
today’s headline must-have, but it is of limited relevance to the goal of
maintaining technological leadership. The Big Fact about semiconductors today
is that Moore’s Law is, finally, yes, coming to its end, its physical limit.
New paradigms like neuromorphic computing (a fascinating prospect, by the way,
described here in Science magazine,),
and perhaps quantum computing, are among other approaches that side-step
(rather than assaulting frontally) the physical limits of traditional processor
designs. These new approaches will soon displace today’s “process node”
fixation. “Process node” competition (which is what the CHIPS 3 nm mandate
plays to) is hardware-intensive, physics-intensive, materials-intensive
technology. It is not the only way to drive progress. Design-driven solutions
(software) can sometimes overleap hardware-intensive technologies, and with far
superior economics. ARM (a core IP supplier) has achieved near-monopolistic
dominance of some applications (95%+ of the smartphone market – the largest
single processor market today), not by brute-force hardware innovation, but by
the opposite – subtle-force software innovation. CHIPS’ technology targets are
based on outdated sloganeering — and by writing them into the law, we risk
baking this obsolescence into the industry and handicapping it for the next
generation.
Nor should CHIPS focus so exclusively on the
fabrication segment. Yes, we need to construct more domestically focused,
“on-shored” supply chains – that is one of the big lessons of the pandemic, and
it applies to all industries, not just semiconductors. But we don’t need
government to lead that charge. The chip industry is already addressing it,
with much bigger investments than CHIPS can project.
Non-U.S. competitors often benefit from foreign
government subsidies. This supposedly creates a serious structural cost
disadvantage for American producers. For example, the Semiconductor Industry Association states
·
“the total ten-year cost of ownership of a new fab located in
the US is approximately 25- 50% higher than in Asia, and 40-70% of that
difference is attributable directly to government incentives.”
This is lamentable, but does it matter? These subsidies are the modern
version of mercantilism – a grievance as old as global capitalism, usually
resorted to by the “losers” in the game, and long since understood to be
ineffective for the most part. Japan, Korea, China
especially — they have all engaged in this “crime” against free trade. Yet the
U.S. still dominates the chip industry, and our position has not deteriorated
in the last two decades in any way that is due to these mercantilist policies
and/or truly significant (i.e., truly threatening).
Nevertheless, many policy analysts (and evidently the Senate staffers
who drafted CHIPS) believe that these foreign subsidies – aside from being
offensive to our sense of fair play – are the crux of the problem facing the
U.S. semiconductor industry and the source of our national vulnerability. The
(obvious) remedy, as they see it (and CHIPS is the manifestation), is to ramp
up our own government subsidies to match these wily foreigners. Does this make
sense?
The answer is No.
First of all, the chip industry is fantastically profitable. This is not
the auto industry circa 2007. They don’t need the money.
American semiconductor firms are not coasting. Even as they spew cash in
enormous quantities, they are investing heavily in R&D – three times as
much, as a percentage of sales, as the Chinese – and 18 times as much in
absolute dollars.
Semiconductor R&D in Absolute
Dollars, USA vs China
·
“Chinese companies accounted for only $8.3
billion, or 37%, of China’s IC output, the rest being produced by the local
silicon wafer fabs (semiconductor factories) of TSMC,
Samsung, SK Hynix, Intel INTC +0.3% and other foreign
companies. That was just 6% of China’s IC market and only 2% of the global market.The total value of semiconductor production by
Chinese companies in 2020 was less than that of AMD, which ranked 15th in
the industry with estimated sales of $9.5 billion.”
If this is correct, and the SIA figure on R&D
as a percent of sales – 6.8% for Chinese firms – is applied, the absolute value
of Chinese private sector R&D would be much lower.]
The next ten years will see an enormous
acceleration of the rate of private sector investment. It will be driven not by
government incentives, but by the profit motive – the desire to capture a share
of a rapidly expanding market.
The leading firms are already investing
aggressively to build capacity. Some of the numbers bandied about recently –
·
Intel’s proposed
investment in the EU in next 10 years: $95 billion
·
Micron’s spending
plans in the next decade: $150 billion
·
Samsung’s proposed
investment next three years: $205 billion
·
SK Hynix (Korea,
formerly Hyundai Electronics – with whom I used to do business) will spend
$97bn on expanding its existing foundry facilities - in addition to a
previously pledged $106bn for four new plants.
·
TSMC’s investment
next three years: $100 billion
Stack them all up – against CHIPS, or against
China’s purported plans, and the picture is clear.
Admittedly, these numbers are not perfectly
calibrated, apples-to-apples comparables. They are
implicitly qualified or conditionalized in various
ways, according to different definitions. Some may be partly “aspirational.”
All are subject to “market conditions.” They may include R&D as well as
construction. They may include investments in chip design, as some of the
companies mentioned here are so-called Integrated Device Manufacturers (IDMs),
which combine multiple business segments. However, it is reasonable to view
these figures as investments in service of adding or improving chip production
(capacity and capability).
And it is not all “in the future.” The value of currently
announced near-term plans by just four companies to add domestic
capacity in the United States is 40% larger than the CHIPS program. These
numbers are less conditional.
Tentative conclusions emerge:
1.
The Private Sector
is capable of investing far more than the government, and is planning to do so.
2.
CHIPS is not going
to be a major factor here. If there were no CHIPS, it would not significantly
alter the outcome.
3.
China is not
investing nearly enough to catch up, even with massive government support, and
their mediocre performance reflects this.
In fact, China is so far behind at this point that
a catch-up plan may simply be infeasible, for the foreseeable future (the next
decade or so). According to one industry analyst:
·
“We estimated a
few years ago for China to become a fourth-equal semiconductor manufacturing
region (equal to US, Korea and Taiwan), would cost approximately $500 bn over ten years. The price tag is now higher.”
If China would have to spend half a trillion
dollars or more just to win 4th place (out of 4) in this global
competition, that alone suggests that chip supremacy or even parity are not
feasible goals — and hence not a threat to the U.S. in any broad sense. And
nobody is standing still to wait for Beijing to catch up. Korea alone will
spend $450 Bn over the same time period. China will
find it very hard to gain any ground.
Which brings us back to the question at the top: Is
the US CHIPS program really necessary?