In early January I published a writeup on my investment into TSMC (TSM) and Nvidia (NVDA) betting on the future of ARM based semiconductor tech. The world is now experiencing a semiconductor shortage, Intel (INTC) has announced a new strategy and in fact so much has happened in this space that it has been almost hard to keep up. So without further ado; here are my expanded thoughts on the semiconductor industry in 2021.
The semiconductor shortage
Building on the events of 2020 and only a few weeks into the new year we started seeing reports of a shortage of chips – particularly within the auto industry. Initially it seemed limited to this field and was given a relatively short time frame. Early on we saw automakers like Ford (F) and General Motors (GM) lower expectations for their earnings and announcing downtime in production facilities. Within long, warnings of the problem potentially expanding to more areas appeared and by late February The President of the United States, Joe Biden invited lawmakers and industry specialists to the White House to announce an executive order addressing the problem. Seeking $37 billion to fund manufacturing in America and underlying the seriousness of the issue. Trouble is that building a semiconductor manufacturing facility and ramping up production takes years and will not solve the issue at hand.
It is a correlation of several factors – many related to the pandemic. Here is a chart I made of the most significant events and their impact on the space. I think it is interesting to see how the many smaller issues stack up to reach a boiling point – The balance between production and demand is hard and sometimes unpredictable.
|Geopolitical tension between China and America, Huawei placed on entity list.||Huawei starts stockpiling chips in anticipation for eventual supply cutoff.||TSMC already working at full capacity to meet demand. China strives for own local semiconductor manufacturing.|
|The Pandemic hits mainland China, soon after spreads globally.||Global production plants and raw material extraction facilities shut down.||Lack of raw materials. Global semiconductor production slows down in general.|
|Automakers brace for lower sales.||Automakers book fewer future production slots for critical components like micro-chips.||Risk increases as ‘Just-in-time’ inventory management style complicates future production ramp up.|
|Increased demand for home office equipment as people work from home.||Consumer technology brands gobble up all available production slots within semiconductors.||Automakers left with too few available production slots due to lack of foresight.|
|Next generation video graphics cards and game consoles go on sale during a time of peak gaming demand.||Demand for semiconductors further increase.||Prices skyrocket on gaming hardware in general as demand overtakes supply.|
|Increased cloud infrastructure and network strain as more people take advantage of available tools from home.||Cloud data center providers needs to rapidly expands to meet demand.||Network, processing and graphics equipment further explode in demand and price. Production backlog builds up.|
|Bitcoin reaches new ATH and new processors are released from major brands in Q4.||Crypto mining reaches new levels and consumers are upgrading their gear.||Even more demand for high end CPUs and GPUs is created.|
|Society slowly reopens as vaccines roll out and governments grant stimuli.||Consumers starts spending again. Production ramps up.||No production slots available for auto makers, others. Vast production backlog. |
|Taiwanese drought, power outages, fire in Japanese semiconductor plant, winter storm in Texas.||Localized production grinds to a halt temporarily, no continuation of new facilities.||World reaches Peak Semiconductor shortage. Hopefully.|
The shortage have been seen as an investment opportunity by many – And I get why. Some see it as a short term opportunity gain, due to increased demand leading to higher revenues. For me it is still about the longer term – When I bought into TSMC and Nvidia I had no idea that the world would be hit by a shortage – However what I had caught on to was that semiconductors would become increasingly more important in the near future and that the ARM based architecture specifically will dominate the next few decades.
What bodes well for me however is that these events have lead to an increased focus by regulators and larger financial commitments into future production by TSMC. In call with analysts TSMC that they expect the severity of the shortage to decrease from next quarter, but that the issue will likely persist for the rest of year. Since then, also Intel and Nvidia have warned that we might see this continue all the way into 2023. TSMC now plans to spend around $30 billion on expanding production this year – something that will cut into their gross margin this year but is likely to benefit me as a long term investor – Just what I like to see. I have doubled my initial investment in TSMC from last year and will continue to hold for many years to come.
A shift in strategy from Intel
Only a week following my original entry, Intel announced that they would replace their CEO. In place of Bob Swan they put Pat Gelsinger, an actual engineer, long time board member and who served as CTO of the company back in the 1980s. Pat Gelsinger has already managed to bring with him a nice change of pace with a slew of new announcements and a shift in the overall strategy of the company.
Most importantly they announced two new manufacturing facilities in Arizona – Taking advantage of some of the incentives Joe Biden has put in place to support local chip manufacturing. A $20 billion investment that will help Intel compete with the likes of TSMC and Samsung (005930) in more ways than one. Uniquely to these two plants is that they will not only act as part of Intel’s usual IDM (Integrated Device Manufacturing) plan – Meaning that they both design and manufacture their own chips – But also as a semiconductor foundry business. Traditionally most of Intel’s said competition has been fabless chip designers like Advanced Micro Devices (AMD), Nvidia, Qualcomm (QCOM) and even the likes of Apple (AAPL) that design their own chips but have them manufactured elsewhere. But as Intel started to experience slowdowns and challenges in their manufacturing plants, people have realized that they also competing with foundry businesses. And Intel has now finally picked up the glove and accepted this challenge: These two new plants will also function as foundries to turn some of these fabless competitors into potential customers.
This renewed strategy of opening up Intel’s manufacturing facilities and competing directly where they have not before, I think will ultimately turn out to be the right decision in the long term. I think they can now finally focus on the most important emerging technologies – Not only for their own chip design departments sake – but also for third parties and whatever potential earnings they might see from this. I am still far from bullish on Intel as a company overall – These decisions should have happened years ago and I fear it is too little, too late. But I am really glad to see that the new management have acknowledged these required changes and avoid an otherwise certain demise. All the news surrounding the company has lead to a lot of new interest from investors and while I understand that the potential upside of buying stock in Intel rather than something like TSMC is much greater, due to the low current valuation and P/E rating of the company – I also recognize that they are now only starting this journey in one of the most notoriously difficult and cost intensive industries in the world.
What I do find really interesting as an investor however is the quote.
“For the 2023 roadmap we will also leverage our relationship with TSMC to deliver leadership CPU products for our client and data center customers. This is the power of our new IDM 2.0 model combined with a modular approach to design and Intel’s industry leading packaging technologies.”Pat Gelsinger, Intel CEO.
This means we are about to see even Intel – go to their fiercest competitor to try and take advantage of their vast manufacturing lead. And that should tell you a little bit about just how far they are ahead.
Nvidia’s perfect trifecta to dominate in datacenters
In mid April, Nvidia held their GTC 2021 conference and with it came a few very important announcements. Nvidia has designed a new ARM-powered datacenter CPU called ‘Grace’ to compliment their GPU dominance in this area. On top of this they also announced a new so-called DPU, called ‘Bluefield’ which is a dedicated chip to help with datacenter processing tasks specifically. With these announcements, Nvidia steps up to challenge both Intel and AMD more directly in this space while at the same time building the capability to deliver a much more tightly controlled and optimized software package as well as boosting to a higher connection bandwidth and performance across components. They now offer a complete package for their datacenter customers to rely on and to become an end-to-end supplier of these much in demand high-end datacenter tasks.
Nvidia at the conference also announced Orin and Atlan, two generational SoCs for cars in an effort to vastly improve chip capabilities within each generation. More importantly Nvidia has built a full vehicle kit called ‘Hyperion’ which includes these SoCs along with suite of sensors and cameras to power the entire car – everything from its self driving capabilities to its infotainment system. The platform supports over-the-air updates and can be customized by the car OEMs to fit exactly their needs and look. Nvidia is now on the way to become much like what Qualcomm and Google is to Android smartphone manufacturers today. On one hand providing the hardware to any OEM to make use of and at the same time offer a software basis to build upon and customize.
These announcements makes me even more confident in my investment in Nvidia and I truly do see this become this space shape up to become much like what we see in the smartphone field today. We have Tesla (TSLA) with their in-house integrated hardware and software designs, mirroring Apple and now Nvidia has potentially opened the door for other car manufacturers to tap into an underlying shared platform, just like we see the various Android OEMs do today using Qualcomm chips. I am really excited for the future of cars and Nvidia just made it even more exciting.
Speaking of Qualcomm and Android OEMs…
On a final note on what has been going on in the semiconductor space, within this very short, yet tremendously exciting times I want to talk about a rumor that is not only very likely to be true, but also likely to eventually impact Qualcomm significantly in the future.
Earlier this month Android Authority, among others reported that Xiaomi and OPPO, two Chinese Android device OEMs are teaming up to develop their own 5G enabled smartphone SoCs. Devices using these new chips are expected to arrive late this year or early 2022. Xiaomi has attempted this before back in 2017 with the Surge S1 chip – a midrange SoC that failed to gather much attention. However, at a product event earlier this year, Xiaomi announced a new in-house custom Image Signal Processing chip, a much less ambitious project, but one that confirms that their dreams are revitalized.
Xiaomi and OPPO are among the largest smartphone vendors in the world and are looking to not only grow significantly domestically but also globally. now that Huawei is more or less out of the picture. You may never have heard of OPPO as this brand does not fare well outside of its own borders, but I am willing to bet that you may have heard of OnePlus, Vivo or Realme – All brands connected to the same parent company ‘BKK electronics’. And if these rumors are true and as serious about the plan as I expect them to be Qualcomm have good reason to be worried.
While there is no reason to panic just yet, it is worth thinking about just how reliant Qualcomm has been on Chinese smartphone OEMs the past three or four years. Every Android brand, with the exception of Samsung, has lost more or less all their market share to Chinese OEMs in this time frame. LG has shut down their mobile division for good and both Sony and HTC are closing in on 0% market share. Nokia is attempting to claw their way back in with low budget offerings along with Motorola – The latter now owned by Lenovo, another Chinese OEM. That leaves Huawei, which already uses their own custom SoC ‘Kirin’ platform, Xiaomi and then those brands I mentioned that are owned by BKK.
If Qualcomm stands to lose these two large customers, they will eventually be left only with providing the chips inside Samsung devices. And that is only half true, as most Samsung devices outside of the US use their own ‘Exynos’ platform. Yet as I have iterated so many times before – Chip manufacturing and design is difficult and there is no reason to believe that even if this all turns out to be true, it will still take years before we see Qualcomm phased out of Xiaomi and OPPO devices. But if you are looking to invest long term in Qualcomm right now, I certainly think it is worth keeping in mind.
These plans are in part what have driven me to invest a small amount of money into Xiaomi stock – A position I plan on growing slowly over the years and which I will make an update and overview on at a later date.
Disclaimer: I am not a financial advisor, the opinions expressed in this article are entirely my own – always invest at your own risk.
If you enjoyed this read and are looking for more content like it, consider signing up for my email list to recieve new posts from jesbaek.com directly in your inbox. It is completely free and will not be used for anything else.