90% of Huawei U.S. purchases can be rapidly replaced.

D.C. has a fantasy that the world stops at Washington’s command. The U.S. is ahead in a handful of high-tech fields, but fewer and fewer. My research shows that about 90% of the purchases from the U.S. can be replaced within months from other parts of the world. Update 5/22. British company ARM blocked Huawei from using Arm designs for future 5G & AI chips. This could be a major problem. Update 6/17 Huawei and China are well on the way to replacing Android, ARM, and RF chips. The only thing that’s biting is the Google blockade, costing tens of billions in phone sales.

For example, Huawei buys memory chips from Micron in the U.S. It could buy similar chips from SK Hynix in Korea, Toshiba in Japan, and imminently Yangtze Memory in China. Many of the most important parts already come from Asia, including screens for mobile phones. 

The hardest to replace will be RF for phones, advanced optics, FPGA’s, and chip design software. Here are some details and possible solutions. 

There are non-U.S. sources for Flex, Seagate, Micron, Western Digital, Corning, and several others. Intel and Xilinx FPGA’s; Skyworks, Qorvo, & Broadcom RF cell phone parts; advanced optics and lasers from several; and chip design software from Cadence & Synopsys have no easy replacements. Here are details and possible solutions.

Radio Frequency components 

Critical for cell phones. These are things like power amplifiers for the incoming signal.  Huawei spends something like $2B/year.

Primary sources Four American companies are world leaders: Skyworks, Qorvo, Broadcom RF division and Qualcomm RF division.

Alternative Murata in Japan and several up and coming Chinese companies. With Huawei and Chinese government support (Made in China, 2025), I believe that that nearly all will find reasonable replacements in 6-24 months. If they can’t develop state of the art RF components, the lesser quality would mean compromises like the phone being 1/8 of an inch larger. (The state of the art chips do more in less area and draw less power.)
My conclusion A problem for now, but solvable. Probably a large stockpile. Qorvo mentioned recent large shipments.

High-end optics/lasers

Used in Huawei’s important core switches, backhaul, etc.

Primary sources Lumentum, NeoPhotonics, Finisar, II VI (two six)
Alternative Huawei is making many similar units, some of which I’ve seen myself. Huawei’s components are not quite of equal quality. So are other Chinese firms. There is very active university research.
My conclusion A problem for certain high-end products, such as 400-gigabit data centre switches. My guess is they could be there in 12-36 months. 

FPGA’s (Field programmable Gate Arrays.) 

FPGA’s are semi-custom chips used in many products. They are used wherever you don’t need 10,000 or 100,000 units, including early prototypes.

Primary sources Three American companies dominate the market: Xilinx. Intel/Altera, and Lattice
Alternative Using multiple chips rather than an FPGA. Designing custom chips (costly)
My conclusion Tough but not impossible. Might require some serious redesign, and make products slightly large. It’s not my expertise, but I’d guess that might add 1/8th or even 1/4th inch to a high-end phone if they don’t develop their own technology.

Chip Design Software

Huawei is matching Qualcomm and Apple with the very best wireless chips. They require more than 1,000 engineers and 15 billion transistors. (Not a typo.) It also makes state of the art AI and network processor chips. Sales were about US$8 billion in 2018 – almost all were used internally at Huawei.

That level of complexity requires extraordinarily complicated software.

Primary sources American companies Cadence and Synopsys totally dominate. They almost certainly will cut Huawei off from updates. 
Alternative Only these two offer a complete set of chip design tools. There are a number of companies that can do parts of the job.
My conclusion Huawei is producing state of the art chips today with the Cadence and Synopsys software. It can continue doing so for a few years. In time, not getting updates will hurt. (?2-3 years.)

Ren has told his people they will need to struggle. He talks of a “long march” to victory. He likes to reference the Battle of Shangganling (Triangle Hill), which effectively ended the Korean War in 1952.

The Americans had massive advantages in equipment and aircraft. The Chinese decided they would do whatever was necessary to hold out. At a high cost in lives, the Americans were thrown back. They reached a stalemate at the 38th parallel which has continued until today.

No lives will be lost in this war, fortunately.

Leave a Reply

Your email address will not be published. Required fields are marked *

Scroll to top