Still with added video!
The mobile industry’s default semiconductor manufacturer is building its next fab in the US, a move that could have major geopolitical implications.
Rumours of the US incentivising big chip-makers to expand their presence on US soil circulated at the start of this week and now we know their provenance. Intel is a US company anyway, so the big prise was always going to be TSMC, which has most of its operations in Taiwan. It looks like bribes offers of support from the US government were too good to refuse.
“TSMC today announced its intention to build and operate an advanced semiconductor fab in the United States with the mutual understanding and commitment to support from the U.S. federal government and the State of Arizona,” opened the TSMA announcement.
“This facility, which will be built in Arizona, will utilize TSMC’s 5-nanometer technology for semiconductor wafer fabrication, have a 20,000 semiconductor wafer per month capacity, create over 1,600 high-tech professional jobs directly, and thousands of indirect jobs in the semiconductor ecosystem. Construction is planned to start in 2021 with production targeted to begin in 2024. TSMC’s total spending on this project, including capital expenditure, will be approximately US$12 billion from 2021 to 2029.”
This development is likely to be highly antagonistic to the Chinese government. Not only is the US already trying to restrict Chinese access to the US component ecosystem, Official Chinese policy insists Taiwan is part of China. Hence every move closer to the US Taiwan makes will be perceived as a strategic threat by the CCP. It will be interesting to see how TSMC deals with the political fallout from this announcement and it wouldn’t be surprising to see it announce an offsetting move there too.
As global commercial Balkanisation ramps, countries are seeking to make themselves as self-sufficient as possible.
The coronavirus pandemic, and the consequent global shortage of things like personal protective equipment, has brought to a head how dangerous it is to be reliant on other countries for essential kit. The mounting hostility between the US and China, coupled with the fact that much of the world’s manufacturing takes place in China, has served to further stoke concern.
Now we have reports that the US government is in talks with two of the world’s biggest semiconductor manufacturers – Intel and TSMC – to build new fabs in the US. Intel has confirmed it’s in discussions with the Defense Department about improving domestic technology sources, while TSMC has confirmed it has been chatting to the Commerce Departments, but not what they discussed.
Intel has fabs in ten different locations, five of which are already in the US and only one of which is in China. As the name Taiwan Semiconductor Manufacturing Company implies, most of TSMC’s fabs are located in Taiwan, but it does have a couple in China and one in the US. In the case of Intel, the US government seems to want a fab that it can call upon to ensure supply of chips in the worst-case scenarios.
The TSMC angle is more intriguing. On a practical level it’s the world leader mobile chip manufacture, an area in which Intel has shown impressively consistent ineptitude. As smartphones have become the single most important smart device, a major interruption to their supply chain would be a significant blow to consumers, businesses and governments alike.
But the really juicy aspect concerns China’s relationship with Taiwan, which it insists is part of China. The Taiwanese people and government beg to differ and the country is a consequently key pawn in many of the geopolitical games China and the US like to play with each other. Persuading TSMC to significantly expand its presence in the US would be a major symbolic victory and seriously antagonise the Chinese Communist Party, which President Trump may consider to be reason enough alone.
The Balkanisation megatrend this would appear to be following raises at least a couple of major issues. Firstly a lot more redundancy looks set to be built into supply chains, as companies and countries wean themselves off just-in-time imports. Secondly the west seems set to adopt an ‘if you can’t beat em, join em’ approach with respect to Chinese subsidising of domestic companies to give them significant advantages over foreign ones.
While this will improve supply chain security, it will also raise prices as companies pass on the additional cost of having to make more stuff themselves and no longer being able to import wage deflation from China. It also seems to herald a permanent enlargement of the state through a greater involvement in the private economy. The cost of this ultimately has to be faced by taxpayers, so it looks like the cost of living is set to be significantly higher for the foreseeable future.
President Trump’s Executive Order and the decision to place Huawei on the US ‘Entity List’ is going to dominate the headlines over the next couple of days, but what will be the impact on US suppliers?
During the ZTE saga last year, where the firm was banned from using US components in its supply chain, several US firms faced considerable difficulty. With Huawei potentially facing the same fate, the next few days will certainly make for uncomfortable reading for some.
Although the main focus of the news has been on the Executive Order banning any Huawei components or products in US communications infrastructure, the entry onto the ‘Entity List’ should be considered as big. This is effectively the commerce version of a dirty bomb, and some might suggest it is being used to disrupt Huawei’s supply chain and dent its ability to dominate the telco vendor ecosystem.
But what is the impact of losing a major customer? What are the realities these US firms will face if the Secretary of Commerce turns down their application to work with Huawei?
Speaking to members of the financial community, it could be pretty severe.
Losing a customer which accounts for 2-3% of total revenues would be a concern but nothing major. For 5% of revenues, this is a headache, but something the spreadsheets could most likely tolerate. When you start getting to 10% the panic button needs to be hit.
A customer which accounts for 10% of total revenues is a major prize. Losing this revenue would result in a complete rethink in how the business operates, as this could effectively wipe out any profit for the year. If you are in the services industry, it isn’t as much of an issue, but when it comes to manufacturing and components, there are so many different implications.
For example, in the first instance you have to consider how this hits budgets, forecasts, resource allocation and manufacturing strategy.
Sales staff are probably the safest here, as the lost revenues will have to be replaced as soon as possible with new customers, but what about the marketing strategy? Do you want to replace the lost capacity with short-term customers (i.e. quicker) or long-term customers which may offer larger orders?
On the R&D side, does a company have dedicated resource working on projects for that customer? What will these staffers do now? Can those projects be re-orientated for another customer?
Finally, on the manufacturing side, there are all sorts of issues. How will the loss of revenue impact the resource recovery plan? How are the manufacturing facilities configured – do you have to close plants?
Another consideration is on your own supply chain and procurement strategies. When supplying products to said customer, you will have to source your own raw materials. Will the loss of this customer result in contracts with suppliers having to be re-negotiated? Will this mean quantity discounts are now impacted?
These are all the considerations when you are losing a customer worth 10-15% of total revenues. Anything above this and you would have to question whether the company can survive, or at least face a major restructure.
|Share price of US suppliers to Huawei|
All share prices at the time of writing (UK: 16:20) – in comparison to market close on 15 May 2019
Looking at Qorvo, executives at semiconductor supplier might certainly have something to worry about. Huawei is features in the ‘top three’ customers for the firm, while on the most recent earnings call, the team discussed the success of Huawei’s smartphone division and in particular the ‘P’ series as a contributor towards a successful quarter. Some have suggested 11% of Qorvo revenues are dependent on Huawei.
Skyworks Solutions, another semiconductor company, has been suffering in recent years. With large parts of the business reliant on smartphone shipments, the global slowdown has been tough. The team work with Huawei on both the mobile and infrastructure side, and while it does work with many tier one firms in both segments, the market is clearly worried about a competitive field and an inability to work with one of the largest telco vendors worldwide.
Both Qorvo and Skyworks supply radiofrequency chips to Huawei, which might have an effect on the Chinese vendors ability to manufacture devices. That said, the supply chain disruption will not be anywhere near as damaging to Huawei as it was to ZTE as it has HiSilicon which manufacturers many of its components.
Xilinx is another which seems to have worn the news quite negatively. The team work with Huawei’s enterprise business unit, helping with video streaming challenges. This might be the smallest business group at Huawei, though the 5G euphoria is set to offer considerable opportunities. Xilinx share price has been recovering after a 17% drop in April, though this has proved to be another set-back.
NeoPhotonics is a company which should be seriously concerned. As a customer, Huawei accounted for more than 46% of the total revenue across 2018. The executive team is relatively open with investors regarding this fact, and this might have been factored into any decision to invest, though this is a massive loss for the business to absorb.
Lumentum is another business which is somewhat reliant on Huawei. While we were not able to nail down specific numbers, the firm supplies fiber optic components to Network Equipment Manufacturers (NEM) and considering there aren’t many of them to supply to, losing Huawei will be a headache.
At Finisar, Huawei described as one of the company’s major customers, though it has seemingly been diversifying its customer base in recent years. In 2017 and 2016, Huawei accounted for 11% and 12% of the annual total respectively, though the percentage is not listed for 2018. This is because the percentage has dipped below 10%, though we were unable to ascertain what the figure now is.
We might have to wait a few weeks to understand the full extent of the impact, and how stringently the US will enforce Huawei’s entry onto the ‘Entity List’, but we suspect there will be some very stressful meetings taking place in numerous offices throughout the US.
Qualcomm and Apple agreed to settle all the ongoing litigations with the iPhone maker paying the chipset maker an undisclosed amount and signing a six-year licensing agreement.
On Monday, Qualcomm and Apple went to court over the allegation that Qualcomm has been abusing its monopoly position to over-charge for its chips. The stakes could have run up to tens of billions of dollars, with the OEMs Foxconn and Pegatron already demanding compensation of $9 billion dating back to 2013. The case at the Southern District Court of California in San Diego was meant to last for five weeks.
On Tuesday, the two companies released a brief statement to announce a settlement. “Qualcomm and Apple today announced an agreement to dismiss all litigation between the two companies worldwide. The settlement includes a payment from Apple to Qualcomm. The companies also have reached a six-year license agreement, effective as of April 1, 2019, including a two-year option to extend, and a multiyear chipset supply agreement.”
This is definitely good news for the two companies especially for Qualcomm, and good for the industry and consumers. Specifically, for Qualcomm it means its business model will remain intact and the company can put an end to a multi-year legal saga; for Apple, in addition to avoiding the punitive $31 billion penalty, this settlement will be able to quicken its steps to launch a 5G iPhone, making up the gap already expanding between itself and the leading pack.
A few hours later, Intel announced that it intends “to exit the 5G smartphone modem business and complete an assessment of the opportunities for 4G and 5G modems in PCs, internet of things devices and other data-centric devices. Intel will also continue to invest in its 5G network infrastructure business. The company will continue to meet current customer commitments for its existing 4G smartphone modem product line, but does not expect to launch 5G modem products in the smartphone space, including those originally planned for launches in 2020.”
It must have been a blow to Intel’s mobile ambition, especially after it announced only late last year that it would bring the launch of its first 5G modem forward by half a year to the second half of this year, an act to prove the doubters wrong. That originally planned 5G modem to be launched in 2020 referred to in the announcement, presumably a second generation, was supposed to power the first 5G iPhone, after Apple all but officially declared that it would enter into an exclusive relationship with Intel.
Putting the two things together it may be reasonable to infer that Apple agreed to settle after it had realised that it does not have other options than coming back to Qualcomm for the supply of 5G modems (assuming Intel had updated Apple about its imminent decision to withdraw from the market).
In addition to leaning in on Intel, Apple has also been reported to be strengthening its in-house modem development capability, ultimately aiming to rid itself of reliance on external suppliers. Based on the terse announcement released together with Qualcomm, it looks Apple does not believe the home-grown modems will be good enough to compete with Qualcomm in the next few years. Huawei is another supplier that has launched its own 5G modem, but it may be safe to estimate that the chance of Apple going for Huawei chips is slim.
In keeping with the normal practice of settlement cases like this, the companies did not disclose the amount Apple will pay. However, Qualcomm updated the SEC shortly after the settlement announcement was made, as the settlement would have material impact on the earnings. The company expected an EPS incremental of about $2 “as product shipments ramp” without giving a specific timespan. As a reference, in the quarter ending 30 December 2018, Qualcomm delivered an EPS of $0.87 on the back of a total revenue of $4.8 billion. Therefore, assuming Qualcomm’s operational efficiency remains largely constant, the payment Apple will make could run into the $10 billion range.
Payment aside, there must be some soul-searching going on inside Apple, including by Tim Cook, the CEO, who came from a supply chain management background: how could Apple have let itself be cornered so badly in the first place? It’s hard to view this as anything other than complete humiliation for Apple, especially when you consider how aggressively it pursued this case.
On top of the millions it will have paid to lawyers Apple’s negotiating position in arriving at this settlement, considering what was widely assumed about its 5G modem situation, must have been very weak. So it’s quite possible Apple has ended up paying considerably more for Qualcomm’s chips than it would have if it had never initiated this war. Having said that, Apple’s share price seems completely unaffected by the news, probably indicating offsetting relief that it’s back in the 5G game. Qualcomm’s share’s however, surged 23% on the news.
Samsung has reported its earnings for the final quarter of 2017, and while profits shrunk year-on-year it does seem to be pretty good news for the Koreans.
Over the course of the quarter, operating profit fell slightly to $14.25 billion, though total revenues did increase 24% compared to the same period to roughly $62 billion. Over the course of the 12 months, revenues increased 19% to just over $224 billion, while operating profit jumped a monstrous 83% to just over $50 billion.
Looking more specifically at the individual business units, the semiconductor was attracting all the praise once again. The largest contribution came from the Memory business that manufactures DRAM and NAND, with Samsung gobbling up market share. This growth is expected to continue over the next couple of months, as new data centre builds will continue the demand for DRAM, while demand for NAND is likely to remain stable.
Interestingly enough, Samsung could now be considered the world leader in the chip market. A title long-held by Intel has now seemingly been handed over, as the $69.1 billion generated by the Samsung semiconductor business exceeds the $62.8 billion brought in by Intel over the last twelve months. Admittedly it is not a direct comparison as the pair operate in different sub-sectors, but looking at a cash basis Samsung is number one.
In the mobile business, earnings declined due to increased marketing costs prior to Christmas and lower shipments. Top-range devices maintained sales, though Samsung stopped selling as wide a range of low-range devices which has been blamed for the slowdown in total shipments. This dip might only be temporary however, as the launch of Galaxy S9 over the next coming months will provide a notable boost for the business.
This focus on the top end will only continue over the next twelve months as well. The business unit might not match total revenue numbers but the theory here seems to be focusing on profitability. In the low- and medium-range devices market the aim will be to simply hold-strong and maintain the current position, while the moves will be made further up the food chain.
Elsewhere on the mobile side of things, the Networks Business anticipates some bites when it comes to 5G commercialization in major markets including Korea, the US, and Japan. We’ve already seen a glimpse of the 5G potential in the US over the last couple of months, Samsung did win a useful contract with Verizon, but this is only the tip of the iceberg. 5G will start paying off before too long, but it can’t come soon enough as Samsung notes ‘the completion of LTE investments’.
Samsung might be known for a flashy phone with a big screen in most places worldwide, but the chip business is proving to be the saviour here. It might not be the most exciting aspect of the technology industry, but considering the anticipated need for components relating to software, connected devices, and services based on AI/IoT platforms, it might just pay-off to be boring.
Ahead of CES in Las Vegas, Samsung has launched its latest blitz on the AI processor market, unveiling the Exynos 9 Series 9810.
Built with the upcoming AI revolution in mind, Samsung has said the chip features a 2.9 GHz custom CPU, a 6CA LTE modem and deep-learning processing capabilities. Build on Samsung’s second-generation 10-nanometer FinFET process, the team will hope to continue the very healthy progress this division has been making over recent months.
“The Exynos 9 Series 9810 is our most innovative mobile processor yet, with our third-generation custom CPU, ultrafast gigabit LTE modem and deep learning-enhanced image processing,” said Ben Hur, VP of System LSI Marketing at Samsung.
“The Exynos 9810 will be a key catalyst for innovation in smart platforms such as smartphones, personal computing and automotive for the upcoming AI era.”
Looking specifically at the LTE modem, Samsung claims it will be able to support 6x carrier aggregation for download speeds of up to 1.2 Gbps, and upload speeds of 200 Mbps. Should this turn out to be accurate, this would comfortably exceed todays maximums, now we just need devices which can support them.
Now the technical bit is over, the new chip will most likely feature in the next Galaxy S flagship device, powering features such as facial recognition and object detection, both of which are fast gaining mass market acceptance.
Energy efficiency will likely be another prominent feature for the phone, a selling point which has historically been popular with the Samsung marketing team. Considering the PR headache Apple is facing with its batteries, Samsung might be able to benefit from the misery.
Embedded GPU maker Imagination Technologies has accepted an acquisition bid of £550 million from Canyon Bridge, which receives its funding from China.
Imagination has been looking for a buyer all summer, having concluded this was the best option for its shareholders after its biggest customer – Apple – announced it was going to do its own GPU thing earlier in the year. Its board has accepted an offer of 182p per share – a 42% premium on the share price immediately before the announcement that values the deal at around £550 million.
Canyon Bridge Capital Partners seems to have been created in order to buy US company Lattice Semiconductor late last year, based in the US but with initial funding from Chinese investors. This provenance seems to have been the reason the Committee on Foreign Investment in the US subsequently blocked the move.
With the Lattice acquisition having been valued at $1.3 billion, the move for Imagination has the feeling of a consolation prize. “Imagination has a world-class management team and highly talented employees,” said Ray Bingham, Partner at Canyon Bridge. “With our backing and investment Imagination can continue to invest in developing its technology, attract and hire the best engineers, and acquire and service customers globally.
“This transaction is in line with Canyon Bridge’s strategy of providing equity and strategic capital to enable technology companies to reach their full growth potential by opening new markets through our collaborative investment approach. We are investing in UK talent and expertise in order to accelerate the expansion of Imagination, particularly into Asia, where its technology platform will lead the continued globalization of British-developed innovation.”
“The proposed acquisition is a very good outcome for Imagination’s shareholders which the Imagination directors are intending to recommend unanimously,” said Imagination CEO Andrew Heath. “Imagination has made excellent progress both operationally and financially over the last 18 months until Apple’s unsubstantiated assertions and the subsequent dispute forced us to change course.
“The acquisition will ensure that Imagination – with its strong growth prospects – remains an independent IP licensing business, based in the UK, but operating around the world. Imagination employs a large number of hugely talented individuals who have developed our market leading technology. They and the business as a whole will benefit from Canyon Bridge’s investment in Imagination as it moves to the next stage of its development.”
One condition of the acquisition will be the separate sale of the MIPS embedded CPU unit, which Imagination acquired back in 2012. While the UK doesn’t seem to be either as protectionist (e.g. Softbank buying ARM) or as twitchy about Chinese involvement in its tech sector as the US it will still be interesting to see if its government decides to have a look at the deal.