A Brief History of Semiconductors: How The US Cut Costs and Lost the Leading Edge

[Daniel Nenni]

This post is the second in a series that uses the history and economics of the American semiconductor industry to ask big picture questions about the future of fiscal policy and industrial policy. As the pandemic ends, the US will have a historic opportunity to revamp its public and economic infrastructure. However, to ensure that industrial policy is effective, many older strategies need to be updated to ensure that they are consistent with the suite of macroeconomic policy settings that support tight labor markets. Today’s post argues that the history of semiconductor manufacturing offers clear lessons for using industrial policy not just in resolving the present shortage, but in building a robust innovative ecosystem to secure the technological frontier for the long term.

While this is a history of the semiconductor industry, the policy takeaways it highlights hold for a wide range of industries. First, fiscal mechanisms play a crucial role in providing liquidity and mitigating financial uncertainty for highly uncertain sectors operating at the economy’s technological frontier. At the same time, industrial policy which inculcates robust supply chains through the reduplication of investment and employment plays a central role in gaining and holding the technological frontier. Science policy — the coordination of R&D undertaken by universities, private companies and public-private partnerships — is not enough. Finally, policy ambition is critical. Though bipartisanship is important, the scale of industrial policy must be such that it is able to achieve its goals.

A Brief History of Semiconductors: How The US Cut Costs and Lost the Leading Edge

By Alex Williams and Hassan Khan

employamerica.medium.com

 

[hist78]

This post is the second in a series that uses the history and economics of the American semiconductor industry to ask big picture questions about the future of fiscal policy and industrial policy. As the pandemic ends, the US will have a historic opportunity to revamp its public and economic infrastructure. However, to ensure that industrial policy is effective, many older strategies need to be updated to ensure that they are consistent with the suite of macroeconomic policy settings that support tight labor markets. Today’s post argues that the history of semiconductor manufacturing offers clear lessons for using industrial policy not just in resolving the present shortage, but in building a robust innovative ecosystem to secure the technological frontier for the long term.

While this is a history of the semiconductor industry, the policy takeaways it highlights hold for a wide range of industries. First, fiscal mechanisms play a crucial role in providing liquidity and mitigating financial uncertainty for highly uncertain sectors operating at the economy’s technological frontier. At the same time, industrial policy which inculcates robust supply chains through the reduplication of investment and employment plays a central role in gaining and holding the technological frontier. Science policy — the coordination of R&D undertaken by universities, private companies and public-private partnerships — is not enough. Finally, policy ambition is critical. Though bipartisanship is important, the scale of industrial policy must be such that it is able to achieve its goals.

A Brief History of Semiconductors: How The US Cut Costs and Lost the Leading Edge

By Alex Williams and Hassan Khan

employamerica.medium.com

The authors made several questionable assertions and omissions based on their poor understanding of the history and semiconductor industry. It's no match for many contributors on this SemiWiki.

Above all, the authors spent time to examine the impact on semiconductor industry from US and Japanese industrial and science policies from 1960s all the way to today. But after so much work, they totally forgot to write "Taiwan" once or anything about Taiwan's semiconductor industry and her policy (only mentioned "Taiwan" in a diagram). Korea is luckier. They were casually spelled out once.

Another shortfall of this paper is they don't understand why pure play foundry business model came to exist. Cost is one important thing, but in reality no IDM back then sincerely wanted to grow their foundry business, regardless of the price tags. When Intel, IBM, and Texas Instruments were making tons of money, why would they need to entertain those much smaller and weaker fabless companies? And those new fabless companies were potentially taking business away from Intel, IBM, or Texas Instruments! This is not something a national science policy or industrial policy can help to resolve.

BTW, the authors probably do not know Intel was offered an opportunity to become a major founding investor of TSMC around 1986 or 1987. But it was rejected by Intel.

 

[hist78]

Another policy approach stated in the paper is that DoD can use its buying power to influence semiconductor industry and DoD only buys products developed domestically. It's a typical misunderstanding shared by many so-called experts but not necessarily by DoD and other national security agencies like DOE, NSA, NRO, NGA, etc.

First, in reality the DoD order quantity is just too small to ask much from fabs.

Examples:

~ The famous Javalin antitank missile started around 1996 and less than 50,000 units have been manufactured since. Each of them needs about 200 various kinds of chips.

~ The super expensive F-35 stealth fighter. First flying in 2006, F-35 annually maximum production rate is set at 156 units per year until the program completed.

~ About 70 mighty US Navy Arleigh Burke class destroyers have been built since 1988 and 20 more are planned or under construction.

By all measurements, these are not huge volume considering it's spreading over multiple decades. DoD can only hope to hitchhike to large commercial production capabilities to make it feasible.

Second, contradict to a common belief, DoD is always willing to spend money on weapon systems/components originally designed/made by foreign suppliers. DoD does ask them to be made or majority made in US during the project life. For example:

~ M777 howitzer. This model is probably mentioned much more often in the news than any Ford or Toyota car models recently. It's designed and built by UK's BAE system. A lot works are done in BAE US in Mississippi nowadays.

~ Naval Strike Missile (NSM). This US Navy and Marine Corp next generation guided anti ship missile is designed and manufactured by a Norwegian company, Kongsberg Defence & Aerospace. It will be gradually transforming to US manufacturing facility with US partners.

~ The new US Navy Constellation class frigate. This new class of frigate will be built in Wisconsin by Fincantieri Marinette Marine utilizing their Italian parent company's existing base design.

Pentagon always buys solutions from foreign suppliers as long as they can transition to be made on the US soil. Due to the cost and time constraints, DoD can't afford to wait too long for a domestic company with good intentions to come out a possible solution for every need. It's more obvious when a foreign supplier from a friendly country can ship the product today or in a reasonable timeframe.

For DoD, this is really a serious business and a live or death business. I'm not surprised DoD, through the Chips Act, is willing to subsidize the new TSMC's Arizona fab.

 

[smeyer0028]

I think the US problem in semiconductors is inability to enforce
on the book anti-trust (better called pro-competition) laws. An example
in my EDA area is that the DEC Alpha chip was much better for EDA
tool speed optimization and would have taken over the EDA tool and
I think scientific computing market. If DEC had been given the time
to establish the Alpha chip there would now be good DEC fabs. Old HP
had good and innovative fabs. If plans to break up IBM in the 1970s
had been followed through the would now be good IBM hardware with fabs. Pattern is innovation in the oil and gas industry that started with
breaking up Standard Oil. This argument is presented in Senator Klobuchar's book on Anti-trust.
"Antitrust: Taking on Monopoly Power from the Gilded Age to the Digital Age"

 

[mozartct]

The evolution of semi manufacturing over 60 years is way too complicated to warrant a "brief history" treatment. To take one example, Japan was not dumping DRAM so much as they were out-yielding their US competitors (See Mostek et al.). Japan, Korea, China, Taiwan were reduced to rubble after WWII, Chinese Civil War and Korean war. What did we expect? That they would go back to a bartering economy and never try to rise up the economic ladder? They rose by copying no doubt, but overtime, the local twist (see Toyota) started to be more important and in fact critical. GM gave up on besting Toyota years ago. TSMC and Samsung are world leaders and they earned it (with some government help no doubt but less that what is generally discussed). The fact is that leading edge manufacturing takes place in VERY large factories that require highly-disciplined well-managed work forces, something we are not good at.

There would be much more to discuss but hey, it's Saturday. Beware the magic wand!

 

[hist78]

n The evolution of semi manufacturing over 60 years is way too complicated to warrant a "brief history" treatment. To take one example, Japan was not dumping DRAM so much as they were out-yielding their US competitors (See Mostek et al.). Japan, Korea, China, Taiwan were reduced to rubble after WWII, Chinese Civil War and Korean war. What did we expect? That they would go back to a bartering economy and never try to rise up the economic ladder? They rose by copying no doubt, but overtime, the local twist (see Toyota) started to be more important and in fact critical. GM gave up on besting Toyota years ago. TSMC and Samsung are world leaders and they earned it (with some government help no doubt but less that what is generally discussed). The fact is that leading edge manufacturing takes place in VERY large factories that require highly-disciplined well-managed work forces, something we are not good at.

There would be much more to discuss but hey, it's Saturday. Beware the magic wand!

I remember an author said: "any person tries to explain Quantum Computing in two sentences is a liar".

 

[smeyer0028]

Here is a "what if". What if Bell Labs had continued its 1950s
tradition (See "Crystal Fire" by Riordan and Hoddeson). INTC
founders had been less greedy stayed and let Shockley determine
semiconductor manufacturing and development strategy. I am
thinking of SLAC and CPI International collaboration that was
a spin off of Varian to build Klystron tubes. Semiconductor
manufacturing would then be something US is good at.

 

[benb]

Gonna edit this to sound less crazy:
Point 1: US semiconductor industrial policy, as the authors state, has historically been MILITARY industrial policy, emphasis on MILITARY.
Point 2: The strategic landscape now includes the possibility of a China-Russia scorched-earth alliance in which US-allied but non-nuclear states fall victim to aggression. Korea and Taiwan have targets on their backs. This is not new, but the scorched-earth alliance is new. Mariupol is new.
Point 3: The US can be strategically weakened via the scorched-earth strategy, economically, and militarily. We have depleted our Javelin supply for example, giving it to Ukraine. This has implications outside Ukraine.
Point 4: Semiconductors fit into the new landscape in that up until 5-10 years ago, there was a balance of production, with the US leading (arguably, let's say), and Asia following. Asia made up 60% of production but it was (again, arguably) trailing edge less strategically important production. Today, we find ourselves with the US entirely dependent on Taiwan and Korea for the leading edge, dual use (MILITARY and Industrial uses), in countries within missile range of both Russia and China, which is a strategic vulnerability.
Point 5: MILITARY Industrial policy will most likely be cloaked. The CHIPs act seems like 50% of the policy, the carrot. The stick is, I think, Taiwan and Korea to remain in the US alliance, need to ramp down fab construction in the conflict zone, at least until a more strategic balance is achieved.
Point 6: What is a strategic balance? Who knows, but 100% leading edge in Asia is bad. It begs for exploitation.
Last point: The scorched earth alliance may plan to destroy weaker states and replace their industries. For example, Russia prefers Ukraine not exploit oil and gas reserves to compete with Russian firms, and so is destroying that capability. Analogously, Chinese semiconductor foundries and memory fabs would benefit if Samsung and TSMC fabs suddenly couldn't produce chips anymore. Again, very arguably, approximately, etc.

 

[mozartct]

@benb Agree 100% that whatever passes for industrial policy is military policy, period.

This issue of manufacturing decline has been covered in many posts - no need to re-hash here. I just wanted to add 3 points:

1- Since the early 80s, finance (and all the values that come with it) has taken more and more space. Wall Street wants high profit margin, high growth and low capitalization. Semiconductor is neither of these things.
2- Again since the 80s but rapidly accelerating in last 20 years, healthcare costs (to employers and employees alike) and healthcare profits (for Wall Street) have skyrocketed so much that they take up 17% of the entire US economy. This favors jobs with high profit margins.
3- And even with all that, the US still manages to capture most of the salary and profit dollars of the broader ecosystem. See Apple, Google,Microsoft, AMAT, LAM, fabless firms... Wall Street loves it.

Money is money, it does not care about you and me, it does not give a damn where chips are made by whom. Ultimately, it does not care about anything other than its self-preservation.

 

[Paul2]

Klobuchar's

Ironic. What I hear now is that "Klobuchar's group" been quietly lobbying for Russia, and China

 

[Paul2]

Point 5: MILITARY Industrial policy will most likely be cloaked. The CHIPs act seems like 50% of the policy, the carrot. The stick is, I think, Taiwan and Korea to remain in the US alliance, need to ramp down fab construction in the conflict zone, at least until a more strategic balance is achieved.

That's very hard to do when everybody remembers what US did to Japanese DRAM industry. They were exploited, and then thrown out, and forgotten. Motorola made its money, and managed to keep afloat for half a decade until leaving semi, but in the end it croaked.

US precluding the rise of Japan, made rise of China an inevitability.

 

[mozartct]

@Paul2 No idea what the innuendo is about. If you have something to say, say it.

Change is the great constant. What made a low-yielding 3" wafer sold to DOD a hit (new functionality during Viet Nam war) has very little relevance to a gate-all-around chip made on 300 mm wafers in a fully automated factory (that is by definition extremely large).

The story we tell ourselves is just that, a story. Korea and Taiwan bested us in the 300 mm foundry world. We have good 200 mm foundries but arguably not a single native 300 mm one (sorry GF). How we got there is for the history books. Increasing production in the USA means more leading edge foundries period. Nothing else is economical. And who is doing it? Primarily Samsung and TSMC (with some DOD prodding for sure).

I mention finally the one exception to the rule: Texas Instruments, perhaps our unheralded champion.

 

[Paul2]

@Paul2 No idea what the innuendo is about. If you have something to say, say it

Sorry, I was quoting @smeyer0028