Apple reportedly pre-ordered 80% of TSCM’s initial output of 5nm chips and 100% of its 3nms. Is that a problem?
From Bloomberg’s “Carmakers Face $61 Billion Sales Hit From Pandemic Chip Shortage” posted last month:
When buyers came back, the auto industry didn’t have enough semiconductors. Chip foundries were busy supplying gadget makers.
From “Amid Shortfalls, Biden Signs Executive Order to Bolster Critical Supply Chains” in Thursday’s New York Times:
“Right now, semiconductor manufacturing is a dangerous weak spot in our economy and in our national security,” Mr. Schumer said. “Our auto industry is facing significant chip shortages. This is a technology the United States created; we ought to be leading the world in it. The same goes for building-out of 5G, the next generation telecommunications network. There is bipartisan interest on both these issues.”
Republicans emerged from the White House meeting optimistic that such efforts could soon move forward. Representative Michael McCaul, Republican of Texas, said he was pleased to see that the White House made the issue a top priority and that the president was receptive. “His words were, ‘Look, I’m all in,’” he said.
My take: Hmm. I could be all wet here. But if there’s a way for Tim Cook to get ahead of this bipartisan buzz, he should take it.
But if Apple does this, they’d still expect to consume 100% of the output, right? So that would have no real net impact on US chip supplies as a whole.
“TAIPEI (Taiwan News) — Taiwan Semiconductor Manufacturing Company (TSMC) purchased a large piece of land in north Phoenix for a planned semiconductor plant last Wednesday (Dec. 9) at a state land auction.
TSMC’s US$89 million bid was the only one received by Arizona at an auction for a 1,129-acre tract of undeveloped land off Interstate 17, according to AZ Central. The auction made the location of the future facility public, which TSMC had previously kept quiet since May, when the project was first announced.
TSMC has said it plans to spend US$12 billion on an advanced 5-nanometer fab, which is expected to begin construction in 2021, with chip production slated to start by 2024. The company has said the new factory will create 1,900 full-time jobs over a five-year period.
Phoenix is still working with TSMC to finalize the development deal. TSMC has also said that it hopes U.S. federal subsidies will help cover the extra costs of manufacturing chips in America. Once the plant is completed, it is expected to use 12-inch wafers and have a production goal of 20,000 wafers per month.”
https://www.taiwannews.com.tw/en/news/4078343#:~:text=TSMC%20announced%20in%20May%20it,5%2Dnanometer%20fab%20in%20Arizona&text=TAIPEI%20(Taiwan%20News)%20%E2%80%94%20Taiwan,at%20a%20state%20land%20auction.
In the meantime, because current chip demand has so far outstripped vendor supply, would probably see some chip price inflation and additional competition for whatever supply is available, assuming off the shelf designs and parts. Lining up fab space and time for custom made parts could be problematic. It would take some time (months to a couple of years) for current chip suppliers to expand any chip production line to meet current short term future demand, if there is such an expansion. Like the Automakers, I suspect the chipmakers are loathe to expand supply without concrete evidence the demand is going to be there in 12-36 months. This may be a short term boom for the chipmakers while a drag on automaker production and revenue until both balance out.
The issue isn’t future supply, it is current supply. Demand spurs the expansion of production capacity. If Apple pre-paid for future supply, then its supplier has the funds to expand capacity, which it is doing now.
The problem is US fabs have been unable to advance their production technology in step with manufacturers’ needs.
https://focustaiwan.tw/business/202102250012
Now we are in a position wherein a critical business segment can’t produce its product. It’s going to take massive tax incentives to get a new fab online in the US. But then I don’t think business should be taxed in the first place.
There’s an interesting discussion on the Texas power grid collapse. Apparently Texas power producers are paid for actual production, where most other grids pay for capacity (and then reimburse based on retail payments of consumption.) This provides them with spare capacity that the Texas system does not. The analog for chips (or any other critical material) is whether government should sent up incentives to pay for (excess/spare) capacity.
It’s worth noting the reason Texas is on its own statewide grid, rather than a regional grid, is explicitly to avoid any Federal oversight.
Apple providing humanitarian relief is a worthy cause. Apple bailing other industries out of their own short-sighted failures to plan according, is an entirely different issue.
Please know that Apple coming to the car industry’s rescue will do little to enhance the tech’s company’s image because the car industry is a cut-throat industry that has the same short-sighted appreciation of gratitude as it did in its failure to prepare for avoiding or mitigating its current chip shortage problem.
Meanwhile Apple, having much higher electronics parts counts while producing far more widgets compared to automakers, and having much higher cash flow, develops and invests in partner relationships and the capacity to for future supply. Long term thinking.
Apple invested in capacity upgrades for their parts. Let the Auto industry invest in dedicated production for their needs. Apple ponied up capital for current and future production, not Apple’s fault that Automakers did not, and not Apple’s responsibility to solve automaker or supplier capacity restraints. Of course, no one wants to overbuild production capacity either.
Neither politicians nor top auto execs can think long term. Neither can explain (if they understood) long term issues to their constituents, voters or investors. When in crisis, they blame someone else.
In contrast, Tim and Elon saw that incremental improvements in chips mfg and battery tech have a powerful ‘compounding effect’. See Albert Einstein’s quote on compounding.
How much should auto makers add to the cost of a new car for parts inventories sitting in a warehouse? If the lines shut down once every 5 years for parts shortage, there’ll be a point when the downtime costs exceeds the carrying cost for the parts. I don’t know what that number is, but I know it’s more than ‘0 days’. (I studied inventory theory in college. There’s a “denial cost” that is part of the calculation, along with “carrying cost”, and a big part of inventory theory is balancing those two costs. See https://en.wikipedia.org/wiki/Inventory_theory )
In California, industries have come and left, can’t say whether there is a net increase or decrease in energy demand as of yet, but certainly after last year’s energy issues, California is likely to be / hopefully better prepared to deal with continuous hot weather energy demands, and to address/update on power sources which are in partial or complete shutdown (for maintenance or refurbishment, or closing) and cannot be relied on for additional power reserve generation. Certainly, consumer energy demand is a big issue during hot weather so more reliable power generation capacity and reserve is needed and needs to be coordinated properly.
Texas’ unique experiment (Enron birthed) has resulted in disastrous failure. I read a report where Texas ordered Samsung’s fab to disconnect from the grid for a significant period. If true, that means Samsung risked a disastrous power interruption if the fab’s on-site backup power failed.
The chip industry has been forewarned and may avoid Texas until they regulate for reliable power delivery.
PS – It isn’t to be overlooked, but Enron sold a power philosophy where utilities delivered cheap power and weatherizing their facilities would just be an extra uncompensated cost. Net Net. Cheap power for decades, all to end in a huge dangerous disaster with accompanying huge government bailout. Profits – private. Bailout cost – public.