D2D China: RF Chips as a microcosm of China Semis
China's semis industry has grown to the point where the need for consolidation is becoming urgent.
In the near future, this newsletter will only be available to paid subscribers. We appreciate your support in helping deliver more content on all the interesting topics shaping the industry today. Please click the link below to subscribe and thank you for reading.
Interesting News
Automotive
Anyone interested in China EVs, or the auto market in general, should listen to Bill Bishop’s podcast with Tu Le of Sino Auto Insight. Our key learnings: 1) Foreign auto makers have seen their brand premium, built over decades, rapidly start to evaporate; 2) China auto OEMs seem to really understand software much better than their foreign competitors; 3) China’s auto industry is building up vast capacity and exports are set to grow even further.
Chinese auto makers are building capacity in Southeast Asia. We have written a lot about China EVs, and an important part of their export surge is the move into less covered markets like Indonesia and Thailand. If EVs are truly easier to manufacture than gasoline engines we expect to see a lot more of these plants opening around the world.
Even while many foreign companies are looking to move production out of China, for many others it remains an important market. Auto parts supplier Continental has to walk that fine line.
A tour inside Tesla’s Shanghai Gigafactory.
Software
There seems to be a quiet exodus of US software companies from China. Citrix and Radware are shutting down, joining Salesforce, Teradata and Tableau a few months back. We have not seen much reporting on this elsewhere, but it is bracing. US companies spent many painful years prying open the Chinese market, and now they seem to be joining the de-coupling train as well.
Huawei appears to be doing everything they can to completely remove every dependency on foreign suppliers. This includes writing much of their own software, their own ERP software, and even their own EDA chip design tools.
Semis and Deep Tech
Stewart Randall on China MEMs companies. This is one area where Chinese companies seem to be doing fairly well and do not seem threatened by sanctions. These companies are still small but show some signs of hope.
A list of China’s top 130 semis companies. Not a lot of household names here.
VeriSilicon is now the 6th largest global vendor of silicon IP. This is one possible backdoor for China’s companies to regain some toehold with TSMC.
Car maker Nio invests in Silicon Carbide start-up Qingchun.
We found a tantalizing link to a post titled “Huawei is Killing Huawei” - but like so many interesting things this post had been deleted from the Internet.
Competition in Semis
A big theme right now is how hard competitive conditions are in much of China Semis. While researching the analysis below, we came across many examples of this online. Here is a sampling of some of that.
There are many posts like this one “16 Chip Companies Close Every Day” which has a very cynical take, but not a lot of hard data. This post, “I Told Investors not to Put Their Money Here”, had a very interesting summary:
Many local governments are actively developing the integrated circuit industry and have also set up local industry investment funds, hoping to introduce more and better chip projects through investment. When the chip industry was hot, local industry funds had no advantage or attraction; When investment in the chip industry cools down, local industry funds have become the choice that chip design companies have to consider.
The translation leaves a bit to be desired, but the point is that local governments are falling over themselves to invest in chip local companies and they end up investing in some of the more desperate companies. Not an encouraging combination. Note the assumption that the central government is scaling back on investments in fabless companies.
This post on changes to the Big IC Fund actually covers the US Chip Act more. This poses an interesting contrast, wondering which is better - China’s investment-led approach to semis or the US’s government-led policies.
China RF Semis spiraling downward
We recently came across an interesting article on WeChat surveying China’s RF chip industry. So much of the article resonated with us that we wanted to discuss it here. In particular, we feel that what is happening in the RF segment is indicative of what is happening across much of the rest of China’s fabless companies. The quick summary is that there are too many companies and too much investor capital resulting in ruinous competition.
A couple quick notes. We cannot vouch for the source. We believe he is an investment analysis based in Beijing, but does not always list the sources for his data. That being said, we can corroborate much of what he says from other sources, and believe most of his conclusions. Also, the various translation apps struggle a bit with this piece, but they work well enough to give the gist of the post.
Radio Frequency or RF components are chips that straddle the divide between analog radio signals and the digital transceivers and modems made by companies like Qualcomm and Mediatek. RF systems are actually fairly complex, with some RF modules containing dozens of tiny parts. Loosely speaking there are three main product groups in RF - Power Amplifiers (PAs), filters and switches. These products are typically bundled together into modules (usually called front-end modules or FEMs) in a confusing mix with companies selling discrete components to erstwhile competitors for assembly into modules. The majority of these products are used in smartphones and other cellular devices, but there is a sizable adjacent market selling modules for Wi-Fi access points.
The global RF market today is dominated by Broadcom, Qorvo, Skyworks, Murata and Qualcomm. These are all big companies with global reach, but they all do a lot of business with Chinese handset makers like Xiaomi and BBK (Vivo & Oppo) who are top 5 customers for the global players.
By contrast, there are about 100 RF vendors in China today. Each segment and sub-segment has 10 to 20 companies supplying into the market. And critically more are still entering.
Not surprisingly, all this competition has led to ruinous pricing. The author of the piece points out double digit price declines in many RF segments in the past year alone. For example, pricing for 4G PAs are now $0.40, and Cat 1 (4G IoT PAs) are at $0.20. Wi-FI FEMs are even worse coming in at $0.17. We remember when prices for these components broke $1, which triggered a rush of consolidation in the RF industry.
Another big problem for these companies is that they are generally not competitive on the global scale. The author estimates that the cellular PA market in China is RMB 150 million - which means none of them are supplying the big domestic brands much. The Wi-FI FEM market is much larger at RMB 3 billion, but there are at least 30 (!) companies vying for that market.
As a result, these companies are left to struggle with gross margins in the 20% - 30% range, and many see much lower levels. Maxscend (江苏卓胜微电子), arguably the market leader, commands gross margins in the 50% range. Several years ago they embarked on a strategy to sell modules bundling multiple parts. By contrast, low-cost leader Lansus (深圳飞骧科技) has gross margins that sometime dip into single digits. And at least these two companies have strategies. From what we can tell the others operate largely opportunistically chasing design wins.
But that is just the start of these companies problems. Despite double digit price declines this year, venture investing in the category is reportedly up 20% to 30%. So more competitors are on the way.
To put this in some context, in the early 2000’s there were about 30 companies competing in the RF market. Each subsequent advance of the wireless standard (starting with 2G to 3G) sparked a wave of consolidation to the point where we are down to five vendors today in the US and Japan.
The author is downright bearish on the sector. He forecasts three more years of price declines and pain, using terms like “life or death” and “hard bones to swallow”. While he acknowledges that the space has seen some consolidation much more is needed.
At heart, the problem is that most of these companies had hoped to build up export capabilities, but for the most part none have managed to move beyond supplying domestic companies largely for domestic consumption.
The author actually notes a perverse dynamic right now. He claims that many investors look at the RF segment and see that there is no established market leader. And since no one dominates the market yet that means there must be room for someone else to come in and take over.
Over the past ten years, China’s venture industry has poured money, supplied by State-backed LPs, into thousands of fabless chip companies. In some senses, this is an immense success, creating an industry where none existed before. But viewed another way, there is too much capital and too many competitors. As we noted last month, a big part of China’s industrial policy is to ignite brutal competition which eventually winnows down the field to a few battle-hardened champions. This is how the handset industry played out, and seems to be the plan for electric vehicles. This view permeates through much of China’s semis industry as well. This approach has some logic when viewed from a national economic development perspective, but from an individual company or venture investor perspective the quest to back the ultimate industry champion leads to some very distorted incentives.
Eventually, China’s RF sector, like the broader semi industry will consolidate. The world cannot support thousands of Chinese chip companies. There is some ‘hope’ that the recent changes to China’s Big IC Fund will lead to a broader push for industry consolidation, but there is no sign of that actually happening right now. And judging from the RF segment there is no slowing in investments in new companies.