The global computer chip battle & CHIPS Act

Investment Analysis Clubs Macro Economic Trends and Risks
1

https://www.wsj.com/tech/chips-act-funding-semiconductor-investments-us-22cc1ea8?mod=hp_lead_pos9

The U.S. Gave Chip Makers Billions. Now Comes the Hard Part.

Chips Act’s early stages have been challenged by rivals overseas and the sheer expense of making chips

By Asa Fitch, The Wall Street Journal, June 2, 2024

Two years into a nearly $53 billion government effort to shore up the U.S. chip industry, the program’s impact is becoming clearer: Big companies making advanced chips are getting a boost, but there are limits to what the money can do.

The Chips Act, passed in 2022 to jump-start domestic semiconductor production, is supposed to supercharge chip making in the U.S. But even in its early stages, it is being challenged by fast-growing chip industries in competing countries, political complexity regarding the allotments at home and the sheer expense of manufacturing chips…

The program is forecast to triple the number of chips made in the U.S. by 2032…

Chips Act grants have focused on cutting-edge chip factories that require tens of billions of dollars of capital outlays, and in that arena, the Boston Consulting report projected the U.S. share would grow from zero to 28%…As a result, new chip-making plants—known as fabs—are sprouting up in Arizona, Texas, New York, Oregon and Ohio…TSMC and Samsung are expected to keep their most advanced chip production back home in Taiwan and South Korea…A single advanced chip fab can cost more than $20 billion, and the planned U.S. facilities won’t be operating until later this decade. …

The Chips Act included a 25% tax credit for chip manufacturing equipment…The tax credit expires in 2026, and industry lobbyists are already preparing to push for an extension. [end quote]

image
image1081×706 11 KB

The CHIPS Act is a long-term effort to enhance U.S. national security by providing support for domestic advanced chip manufacturing.

Despite the infusion of cash, INTC stock is down 40% YTD and hovering near a 5 year low.
https://www.marketwatch.com/investing/stock/intc

On the other hand, Micron Technology (MU) is up 48% YTD and rocketing to an all-time high.
https://www.marketwatch.com/investing/stock/mu?mod=mw_quote_switch

Wendy

4 Likes
2

They will definitely need to extend it because these facilities take many years to build and to get running. They will have to extend it to 2028 or even 2030.

This is because people (investors) don’t trust Intel to not screw it up. They’ve made multiple bad decisions over the last decade or two, and there is no indication of any large management changes that might make better decisions in the future. But there’s still hope.

The only issue I see going forwards is that advanced chips (let’s say 3nm or smaller geometry) will likely become more expensive. That’s because when it was only TSMC making them, they could amortize the MASSIVE costs across all the chips sold. Now with 3 manufacturers, assuming each manufacturer gets about a third of the volume, even if volume goes up a bit, they have to amortize the costs across a lesser number of chips. Government kicking in some of the initial capital helps, but there are still very large investments required to keep the fabs going and growing.

7 Likes
3

This is reminding me of something I heard on LinkedIn over 8 years ago, that the costs of bleeding edge fabs (back then) was so high that only 3 companies in the world could afford them: Intel, TSMC and Samsung. And that soon only one would be able to, Samsung, due to tight relationship with their gov’t. i.e. they are subsidized. And now we are seeing Intel get subsidized as well. And I really do think it was going to be the only way for a domestic supplier to fight back, given the costs involved these days. They are enormous.

4 Likes
4

It seems obvious – to everyone except maybe politicians – that it is impossible to “catch up” to some state of the art in a specific area of science or technology by simply spending tens of billions on that topic in a concentrated window of time. There is some metaphysical / mathematical upper bound on how much humankind in general and how much any specific group of mankind can advance its understanding and abiilty to use new science or technology for some productive or lucrative purpose. In the case of integrated circuit manufacturing, advancing the state of the art requires bleeding edge expertise in ALL of the following areas:

  • material science
  • chemistry
  • modeling / design software
  • control systems
  • physics

The number of people on the planet capable of contributing at that edge in these areas for chip manufacturing probably only numbers about 100-200 and they require a support pool of people not quite as sharp that probably numbers 1000-2000. No one of that caliber becomes available overnight because a government wants to catch up and spend $53 billion dollars in the attempt. It’s at least a 4-8 year pipeline to wait for people with these backgrounds to emerge from an undergrad / post-grad degree program as raw material to learn the ropes actually working in the industry.

That general topic aside, it is worth noting that Intel has not had a very good 10-15 years or so competing with other fabricators in density, functionality and quality. However, I saw a video on YouTube recently about an advance Intel is making in chip fabrication with HUGE impacts on power consumption and heat generation in large chips.

The technology is called Backside Power Delivery (or Backside Power Delivery Network). The upshot of the video is that most chips today use a design which delivers power to all transistors on the chip from one side of the chip downwards. Much like the human circulatory system, the traces carrying current down into the chip start out large towards the top and get narrower as they go from layer 1 to 2 to 3 to N (some chips may have nearly 100 layers). Even though one might think of these as “wires” that carry the same voltage across their length, because they DO pose a resistance, there is a drop in voltage from the top surface of the chip to the bottom-most layer across these leads. And since a voltage across a resistance generates HEAT, when there are THOUSANDS of these vertical leads supplying current to lower layers, the HEAT lost in just conveying power to the lower layers is substantial. (A modern Intel Core i9 chip at 4 GHz can draw 253 watts of power).

Intel has devised a manufacturing technology that will allow power to be supplied to the chip from the top and bottom of the chip. Reducing the width of these leads at the edge as they narrow towards the center reduces this Ohm’s Law heat generation across these tiny voltage drops. Since less horizontal room is taken up by the wider power leads, this can also help increas transistor density per layer and per chip.

The level of engineering expertise required to make a modern chip makes the technology used only 20 years ago for Pentium era chips look like steam-punk technology.

WTH

16 Likes
5

Thanks all on this XLNT thead, from OP on down. I have forwarded copies to three college kids I am mentoring who all want to have a future in high tech, potentially chip fab, in Mexico. I pointed out all the thread as relevant, and especially WTH’s off the cuff about the bleeding edge 100 - 200 crux folk and the smart motivated 1000 - 2000 people enabling the crux folk.

They are aiming that high.

d fb

2 Likes
6

Readers of this thread might be interested in the Acquired podcast’s interview with NVIDA CEO Jensen Huang. Huang is one in billion.

2 Likes
7

The market could be discounting the possibility of something happening in a few months that would cause a reversal of government policy: repeal of the “chips” thing, and using the money to hand the “JCs” another tax cut instead.

Have you noticed that several states have filed suit challenging the “big gummit defacto EV mandate”, and an OEM which has invested heavily in EVs, taking the government’s side, lest it’s huge investment be rendered irrelevant?

Steve

3 Likes
8

Yep.
It took China 20 years to develop their EV edge.
2001; that year, EV technology was introduced as a priority science research project in China’s Five-Year Plan.

Then, in 2007, the industry got a significant boost when Wan Gang, an auto engineer who had worked for Audi in Germany for a decade, became China’s minister of science and technology. Wan had been a big fan of EVs and tested Tesla’s first EV model, the Roadster, in 2008, the year it was released. People now credit Wan with making the national decision to go all-in on electric vehicles. Since then, EV development has been consistently prioritized in China’s national economic planning.

Starting in 2009, the country began handing out financial subsidies to EV companies for producing buses, taxis, or cars for individual consumers.

From 2009 to 2022, the government poured over 200 billion RMB ($29 billion) into relevant subsidies and tax breaks.

And China continues pumping money into battery & Chip technology.
https://www.msn.com/en-us/money/news/china-s-pumping-big-money-not-just-into-chip-sector-but-also-ev-batteries-key-tesla-nio-suppliers-among-those-eligible-to-get-830m-r-d-funding/ar-BB1ngqXl
China is reportedly set to allocate more than 6 billion yuan ($830 million) to a state-led initiative aimed at advancing solid-state battery technology, in addition to its recent investments aimed at bolstering its domestic semiconductor industry.

2 Likes
9

AT&T spent 20 billion to upgrade from 3g to LTE. This was done in a 4 or 5 year period, not 13 years.

  • The overall project cost of the F-35 program has increased from an initial $200 billion to over $400 billion, with an estimated lifetime cost of $1.7 trillion.

The costs of the Iraq War are often contested, as academics and critics have unearthed many hidden costs not represented in official estimates. The most recent major report on these costs come from Brown University in the form of the Costs of War, which totaled just over $1.1 trillion. The United States Department of Defense’s direct spending on Iraq totaled at least $757.8 billion,

So for the Chinese investment into EV’s has been less than 1/5 what the U.S. spent looking for weapons of mass destruction.

I think I found those weapons, they sit in congress.

Cheers
Qazulight

7 Likes
10

That is including the jet fuel and the retirement package of every service member that looks at the plane.

We have no air defense without the F35. Everything else in the Pacific theater would be shot down by China within the opening few hours.

2 Likes
11

Some of that money (several hundred million $$) is to be used to expand a local chip fab in the metro Twin Cities–which makes a lot of chips for the auto, appliance, and similar markets. It will take 2-3 yrs because the machines needed to make the chips take a LONG time to get built and installed. Capacity will be at least doubled, possibly even more.

2 Likes